Water soluble lanthanoid benzoate complexes for the kinetic separation of cis/trans-limonene oxide

Article abstract of DOI:10.1016/j.tetasy.2006.10.026

A new class of water soluble, environmentally friendly, lanthanoid 3,5-diacetamidobenzoate complexes (Ln = La, Gd, Yb) have been synthesized. The La and Gd complexes selectively catalyse hydrolysis of the cis-isomer of limonene oxide allowing for the separation of the trans-isomer (>98:2 dr) in up to 74% yield. Comparative studies with the corresponding chlorides and triflates reveal the lanthanoid benzoate complexes to be more active than the chlorides, but less active, though more selective, than the triflates.

Full text of DOI:10.1016/j.tetasy.2006.10.026

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Tetrahedron: Asymmetry 17 (2006) 2833–2838
Water soluble lanthanoid benzoate complexes for the
kinetic separation of cis/trans-limonene oxide
Philip C. Andrews,* Michael Blair, Benjamin H. Fraser, Peter C. Junk,
Massimiliano Massi and Kellie L. Tuck
School of Chemistry, Monash University, PO Box 23, Victoria 3800, Australia
Received 2 October 2006; accepted 19 October 2006
Abstract—A new class of water soluble, environmentally friendly, lanthanoid 3,5-diacetamidobenzoate complexes (Ln = La, Gd, Yb)
have been synthesized. The La and Gd complexes selectively catalyse hydrolysis of the cis-isomer of limonene oxide allowing for the
separation of the trans-isomer (>98:2 dr) in up to 74% yield. Comparative studies with the corresponding chlorides and triflates reveal
the lanthanoid benzoate complexes to be more active than the chlorides, but less active, though more selective, than the triflates.
ꢀ 2006 Elsevier Ltd. All rights reserved.
1. Introduction
Homogeneous catalysis in non-volatile organic solvents is
currently a field of active research due to interest in sustain-
able synthetic methods with low environmental impact.1 In
this respect, epoxide ring opening reactions in water have
O
great potential owing to the vast number of enantiomeri-
cally pure compounds that are accessible via this simple
transformation.2 As catalysts, lanthanoid triflates appear
to be an obvious choice since they have already been shown
to be effective in a number of other aqueous based organic
procedures.3 From an industrial perspective though,
triflates may not be ideal because of the highly corrosive
nature of triflic acid. Lanthanoid benzoates4 are a possible
alternative with the potential to provide similar reactivity
with improved selectivity. However, the relative insolubil-
ity of simple benzoates has inhibited comparative studies
in this area.
limonene oxide
O
O
2
1
cis
trans
limonene oxide
limonene oxide
Figure 1. Limonene oxide.
and industrial researchers and as such, a number of proce-
dures now currently exist, with varying efficiencies, for their
separation. These fall largely into two categories: (1) sepa-
ration and recovery of each diastereomer based upon dif-
fering physical properties, for example, chromatography
or distillation,8 and (2) kinetic separation where one diaste-
reomer reacts faster than the other, allowing for recovery
of the pure unreacted diastereomer, for example, photo-
assisted kinetic separation9 and biocatalysis,10 electrophilic
mercuration,11 amine addition12 and molybdenum(VI)
catalysis.13
The pure diastereomers 1 and 2 of limonene oxide (Fig. 1),
a naturally occurring epoxide, have been used in the total
syntheses of a number of natural products.5–7 Although
pure 1 and 2 are commercially available, they are expensive
and are more commonly purchased and used as a mixture
of the cis/trans (47:53) diastereomers. Efficient and com-
mercially viable methods for separating the diastereomers
have, therefore, been of some interest to both academic
While the degree of separation in these chemical processes
is generally >90%, they suffer from various drawbacks
including being non-catalytic, requiring high reaction tem-
peratures, the use of toxic reagents/catalysts, a need for
derivatisation, and air/moisture sensitivity.
*
Corresponding author. Tel.: +61 3 9905 5509; fax: +61 3 9905 4597;
e-mail: phil.andrews@sci.monash.edu.au
0957-4166/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2006.10.026

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
2834
P. C. Andrews et al. / Tetrahedron: Asymmetry 17 (2006) 2833–2838
Herein, we report the first application of water soluble lan-
thanoid (Ln = La, Gd, Yb) benzoate complexes, derived
from 3,5-diacetamidobenzoic acid, and their application
as catalysts in the ring opening and consequent kinetic
separation of limonene oxide.
Since limonene oxide is only sparingly soluble in water
(4.6 · 10À3 mol LÀ1 at 25 ꢁC), the reaction mixtures are
best described as dispersions rather than homogeneous
solutions. The suspensions were vigorously stirred at room
temperature and the reaction progress was monitored by
1H NMR spectroscopy. A reliable analytical procedure
involving extraction with ethyl acetate and in vacuo
concentration was initially hindered by the volatility and
significant loss of limonene oxide. After some experimenta-
tion, the most reliable and reproducible method involved
an initial extraction with CDCl3, to efficiently remove all
the unreacted epoxide, followed by a second extraction
with ethyl acetate to remove the remaining diol 7. The ethyl
acetate extraction was then evaporated to dryness leaving
the non-volatile solid diol. This was re-combined with the
CDCl3 extract to form the sample for NMR analysis. With
a good procedure for monitoring the reaction progress
established, it was discovered that the only product of the
reaction was the water soluble trans-diaxial diol 7 (Scheme
2), which was not unexpected since both cis- and trans-lim-
onene oxides have been reported to open through a diaxial
mechanism to yield the same diol.14 Further NMR analysis
revealed that the cis-diastereomer 2 reacts at a much faster
rate than the trans-diastereomer 1 and that a kinetic sepa-
ration of trans-limonene oxide 1 could be obtained if the
reaction was stopped at the appropriate time. Under nor-
mal, non-analytical, reaction conditions the remaining
unreacted trans-limonene oxide 1 could be isolated by pen-
tane extraction of the aqueous layer (pure by NMR and
>98:2 dr by GC analysis). The aqueous phase containing
the catalyst was used for three consecutive reactions and
extraction cycles without any significant reduction in the
yield of diol or loss in the recovered fraction of trans-
epoxide.
2. Results and discussion
To establish the effectiveness of lanthanoid benzoate cata-
lysts in an aqueous medium, the ligands first had to be
designed to allow solubility and stability. Convenient targets
were the complexes derived from 3,5-diacetamidobenzoic
acid 3. The ligand was efficiently prepared by the acylation
of 3,5-diaminobenzoic acid with acetic anhydride. Sub-
sequent treatment of 3 with lanthanoid bicarbonate,
prepared from the lanthanoid oxide, gave the appropriate
La, Gd and Yb complexes 4–6 in good yield (Scheme 1).
For examination as catalysts in the ring opening and
kinetic separation of cis/trans-limonene oxide, complexes
4–6 were dissolved in water (5 mol %, relative to total limo-
nene oxide) and the 53:47 mixture of limonene oxides 1 and
2 added (Scheme 2).
1. conc. HCl
2. sat. NaHCO3
Ln2O3
Ln(HCO3)3
Ln = La, Gd, Yb
O
HN
EtOH/water
4:1
O
O
N
H
Figure 2 shows the reaction profile of the commercially
available mixture of limonene oxides 1 and 2 with the
lanthanum catalyst 4. At t = 0, the species fractions of
limonene oxide are trans = 0.53 and cis = 0.47. As the
reaction proceeds there is a fast consumption of the cis-
epoxide and a slow consumption of the trans-epoxide.
After 13 h, the cis-epoxide is completely transformed to
the diol and 74% of the original fraction of trans-epoxide
is recovered.
OH
O
3
O
HN
O
N
Ln
.nH2O
3
H
O
The gadolinium complex 5 gave a similar reaction profile
(Fig. 3) where complete consumption of the cis-epoxide 2
was achieved in ca. 15 h. In contrast to the La catalysed
reaction, the fraction of trans-epoxide recovered, decreased
to 0.25 (47%).
4 Ln = La 78%
5 Ln = Gd 98%
6 Ln = Yb 83%
Scheme 1. Synthesis of 3,5-diacetamidobenzoate complexes 4–6.
catalysts 4-6
(5 mol%)
+
+
H2O, rt
OH
O
O
O
OH
trans-epoxide 1
cis-epoxide 2
trans-epoxide 1
trans-diaxial diol 7
Scheme 2. Separation of cis/trans-limonene oxide 1.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
P. C. Andrews et al. / Tetrahedron: Asymmetry 17 (2006) 2833–2838
2835
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
2
4
6
8
10
12
14
0
2
4
6
8
10
12
time (h)
time (h)
trans-diaxial diol
trans-epoxide
cis-epoxide
trans-diaxial diol
trans-epoxide
cis-epoxide
Figure 2. Reaction profile for the hydrolytic ring opening of limonene
oxide mixtures 1 and 2 in the presence of 5% lanthanum catalyst 4.
Figure 4. Reaction profile for the hydrolytic ring opening of limonene
oxide mixtures 1 and 2 in the presence of 5% La(OTf)3.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.6
0.5
0.4
0.3
0.2
0.1
0
0
2
4
6
8
0
5
10
15
time (h)
time (h)
trans-diaxial diol
trans-epoxide
cis-epoxide
trans-diaxial diol
trans-epoxide
cis-epoxide
Figure 5. Reaction profile for the hydrolytic ring opening of limonene
oxide mixtures 1 and 2 in the presence of 5% Gd(OTf)3.
Figure 3. Reaction profile for the hydrolytic ring opening of limonene
oxide mixtures 1 and 2 in the presence of 5% gadolinium catalyst 5.
The increased reactivity of gadolinium triflate (7 h for com-
pletion) over lanthanum triflate (11 h for completion) is
rationalised by the contraction in ionic radius and
increased Lewis acidity of gadolinium over lanthanum.
As was observed for the benzoates, the ytterbium triflate
proved to be much less reactive (Fig. 6) than the La and
Gd analogues. The reaction profile shows that after 9 h
only ca. 50% of the cis-epoxide 2 had been hydrolysed. This
observation is consistent with that obtained with ytterbium
complex 6 and most likely results from the epoxide sub-
Surprisingly, the Yb complex 6 gave no appreciable reac-
tion over a period of 2 days. Monitoring over a further
period of 2 days revealed a very slow, and non-selective
reaction which was not studied further.
It was important to compare the reactivity and selectivity
of complexes 4–6 with their corresponding lanthanoid
chlorides and triflates. The lanthanoid chlorides (La, Gd,
Yb) proved to be inefficient catalysts with very little reac-
tion occurring at room temperature over a period of
5 days. The triflates were prepared using an adaptation
of the procedure described by Kobayashi.15 The mixture
of limonene oxides 1 and 2 were added to the lanthanoid
triflates (5 mol %) in water and stirred vigorously. The
reaction progress was again monitored by 1H NMR
spectroscopy. The reaction catalysed by La(OTf)3 (Fig. 4)
was faster than that catalysed by lanthanum benzoate com-
plex 4 and complete consumption of the cis-epoxide 2 was
achieved in a reaction time of 11 h. However, there was a
decrease in the amount of trans-epoxide recovered (65%).
0.6
0.5
0.4
0.3
0.2
0.1
0
0
2
4
6
8
10
Gadolinium triflate also catalysed the hydrolysis of
limonene oxide faster than the corresponding gadolinium
benzoate complex 5 (Fig. 5). Complete consumption of
the cis-epoxide 2 was achieved in 7 h, but again, increased
reactivity of the trans-epoxide led to an inferior recovery of
this diastereomer (61%).
time (h)
trans-diaxial diol
trans-epoxide
cis-epoxide
Figure 6. Reaction profile for the hydrolytic ring opening of limonene
oxide mixtures 1 and 2 in the presence of 5% Yb(OTf)3.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
2836
P. C. Andrews et al. / Tetrahedron: Asymmetry 17 (2006) 2833–2838
Table 1. Summary of reaction times, yields and dr for the separation of
commercially available (+)-cis/trans-limonene oxide
points were recorded on a Kofler hot stage apparatus
and are uncorrected. Mass spectrometry (ESI) was per-
formed on a Micromass Platform QMS spectrometer. High
resolution mass spectra (HRMS) were recorded on a
Bruker BioApex 47e FTMS. Infrared spectra (IR) were
recorded on a Bruker Equinox 55 ATR spectrometer.
GC–MS were performed on a Varian 3700 gas chromato-
graph using a 30QC5/BPX5 1.0 lm column of internal
diameter 0.53 mm and length 30 m at a linear velocity of
40.6 cm/s. Elemental microanalyses were performed by
the University of Otago, Dunedin, New Zealand. Lantha-
noid oxides and (+)-limonene oxide, mixture of cis/
trans-diastereomers were obtained from Sigma–Aldrich
and used as supplied.
Catalyst
Time dr
(h)
(1:2)a
Fraction of trans-epoxide Yield of
1 recovered (%)
diol 7 (%)
4
13 >98:2
15 >98:2
39b (74)c
25b (47)c
—
—
—
61
75
<5
<5
<5
<5
73
68
23
5
6
>36 53:47d
>120 53:47d
>120 53:47d
>120 53:47d
11 >98:2
LaCl3
GdCl3
YbCl3
La(OTf)3
Gd(OTf)3
Yb(OTf)3
—
27b (65)c
32b (61)c
—
7
>98:2
>24 35:26
a Diastereomeric ratio of trans/cis-limonene oxide after given reaction time
(h).
b Yield as % of total limonene oxide starting material (max = 53%).
c Yield as % of only trans-limonene oxide starting material.
d No appreciable reaction with no change in limonene oxide fraction.
4.2. 3,5-Diacetamidobenzoic acid 3
To a stirred suspension of 3,5-diaminobenzoic acid (5.0 g,
32.9 mmol) in distilled water (150 mL) was added K2CO3
(13.3 g, 98.7 mmol). After stirring for 0.5 h, the solution
became homogeneous and acetic anhydride (7.80 mL,
82.6 mmol) was slowly added. The reaction was stirred at
rt o/n after which time all the water was removed under
reduced pressure and the residual solid re-crystallised from
boiling ethanol/water (5:1) to give a grey solid (7.06 g,
91%). 1H NMR (300 MHz): d 2.04 (s, 6H), 7.89 (d,
J = 1.8 Hz, 2H), 8.10 (t, J = 1.8 Hz, 1H), 10.08 (br s,
2H). 13C NMR (75 MHz): d 23.0, 113.5, 114.6, 131.4,
139.7, 167.1, 168.6. IR (ATR) 3313s, 2894m, 1684s,
1627s, 1572m, 1533s, 1479m, 1452m, 1353m, 1321m,
1286s, 1158m, 1115w, 1036s, 934w, 912m, 871m, 645m,
625m cmÀ1. MS calcd for C11H12N2O4Na+ = 259.1, found:
259.0. HRMS calcd for C11H12N2O4Na+ = 259.0695,
found: 259.0695.
strate not being able to successfully compete with water
and the sulfonate ligands for co-ordination to the smaller,
more Lewis acidic ytterbium centre.
The reaction times and yields obtained with each of the lan-
thanoid 3,5-diacetamidobenzoate complexes 4–6 and the
lanthanoid triflates and chlorides are summarised in Table
1. As can be seen, the La benzoate catalyst 4 is the most
efficient, followed by the La and Gd triflates, both of which
are marginally better than the Gd benzoate complex 5,
while the chlorides are ineffective.
3. Conclusion
This new procedure provides a high yielding and environ-
mentally benign procedure for the formation of trans-axial
diol 7 and consequent kinetic separation of trans-limonene
oxide. The La benzoate complex proved to be the most effi-
cient, outperforming all three metal triflate complexes.
While the degree of kinetic separation may not be as high
as those reported for other chemical systems this new reac-
tion process provides the benefits of being catalytic in
water, can be carried out at ambient temperature, mini-
mizes the use of volatile organic compounds with no need
for reagent derivatisation, and importantly does not
involve the use of toxic reagents/catalysts. Significantly,
lanthanide tris-2,5-diacetamidobenzoate complexes repre-
sent a new class of water soluble catalysts, for which this
study provides an initial assessment. Further investigations
are currently underway in our laboratories to develop and
test analogous complex systems in ring opening reactions,
and to compare the reactivity and selectivity of these cata-
lysts with lanthanoid triflates in a variety of other organic
transformations.
4.3. General procedure for the synthesis of benzoate
complexes
To solid Ln2O3 (0.5 mmol) was added concd HCl (12 M,
1 mL). Stirring was continued until all the solid had dis-
solved and then satd NaHCO3 (ꢀ40 mL) was added until
the pH reached 8.5. The precipitated Ln(HCO3)3 was col-
lected on a sintered funnel and washed with distilled water
(3 · 5 mL), acetone (3 · 5 mL) and ether (3 · 5 mL) and
then dried under vacuum for 2 h at rt. It was then added
to a solution of 3,5-diacetamidobenzoic acid (2 mmol) in
refluxing ethanol/water (4:1, 300 mL). The reaction was
heated at reflux for a further 3 h after which any unreacted
solid was removed by filtration and the ethanol/water
removed under reduced pressure. The residual solid was
suspended in THF and heated to reflux for 1 h before being
filtered and washed with boiling THF (3 · 50 mL). The
resulting grey powdery solid was dried under reduced
pressure at rt for 2 days.
4. Experimental
4.1. General
4.4. Lanthanum tris-3,5-diacetamidobenzoate 4
Following the general procedure as above with 3,5-diacet-
amidobenzoic acid 2 (1.00 g, 4.23 mmol) and La2O3
(173 mg, 0.53 mmol), lanthanum tris-3,5-diacetamido benz-
oate 4 was obtained as a greyish pink powder (0.7 g, 78%).
1H and 13C NMR were recorded at 300 and 75 MHz,
respectively, on a Bruker AM 300 spectrometer. Melting