Cyclisation of citronellal over heterogeneous inorganic fluorides - Highly chemo- and diastereoselective catalysts for (±)-isopulegol

Article abstract of DOI:10.1039/b817572a

Based on a fluorolytic sol-gel synthesis, nanoscopic metal fluorides and partly hydroxylated metal fluorides were synthesized; varying the F: OH ratio inside these solids yielded catalysts with different combinations and variable strength Lewis and Br?nsted acid sites, which demonstrated unexpected catalytic properties for the diastereoselective synthesis of (±)-isopulegol. The Royal Society of Chemistry.

Full text of DOI:10.1039/b817572a

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
Cyclisation of citronellal over heterogeneous inorganic fluorides—highly
chemo- and diastereoselective catalysts for (Æ)-isopulegol
Simona M. Coman, Pratap Patil, Stefan Wuttke and Erhard Kemnitz*
ab
a
a
a
Received (in Cambridge, UK) 6th October 2008, Accepted 30th October 2008
First published as an Advance Article on the web 1st December 2008
DOI: 10.1039/b817572a
Based on a fluorolytic sol–gel synthesis, nanoscopic metal
fluorides and partly hydroxylated metal fluorides were synthe-
sized; varying the F : OH ratio inside these solids yielded
catalysts with different combinations and variable strength
Lewis and Brønsted acid sites, which demonstrated unexpected
catalytic properties for the diastereoselective synthesis of (Æ)-
isopulegol.
formed. Both types of acid site (Brønsted/Lewis) can be tuned
over a wide range by applying different molar ratios of HF to
8
O during the modified sol–gel synthesis. We have already
H
2
successfully applied this type of catalyst to the synthesis of
8
,10
(all-rac)-[a]-tocopherol,
be applied as new active and highly diastereoselective catalysts
and will show here that they can also
for the cyclisation of citronellal to (Æ)-isopulegol (Scheme 1).
8
–11
The catalysts were synthesised as reported elsewhere.
The world’s largest demand for menthol exists among the mint
products widely used in pharmaceuticals, agrochemicals,
HS-AlF
surface metal fluorides using anhydrous HF, while AlF
was synthesised using aqueous HF solutions with a concentra-
3
was synthesised through the sol–gel route for high
1
1
3
-H
1
cosmetics and flavouring applications. Among the four pairs
1
0
of optical menthol isomers, only (À)-menthol exhibits the
characteristic peppermint odour, and exerts a unique cooling
sensation on the skin and mucous membranes. Nevertheless,
racemic (Æ)-menthol also has industrial applications, although
its refreshing effect is lower than that of (À)-menthol alone.
On an industrial level, one of the major routes to menthol is
the Takasago process, which involves, as one of the key steps,
the isomerization of (+)-citronellal to (À)-isopulegol with an
tion of 50 wt% HF. The last procedure was also applied to the
preparation of MgF -40, MgF -57, MgF -71, MgF -87 and
MgF -100 samples using an aqueous HF solution, with con-
2
2
2
2
2
8
centrations of 40, 57, 71, 87 and 100 wt% HF, respectively.
Activity tests in batch mode were carried out, as described in the
following procedure: racemic citronellal (5.6 mmol, 860 mg) was
dissolved in 5 ml of solvent (toluene, cyclohexane, n-heptane,
i-propanol). The catalyst (50 mg) was added to this mixture, and
it was then heated to 40, 60 or 80 1C with stirring for 1–6 h. After
the reaction, the catalyst was filtered off and the reaction mixture
2
,3
aqueous ZnBr catalyst. As only (À)-isopulegol affords the
2
correct (À)-menthol configuration during its selective hydro-
genation, its synthesis from (+)-citronellal requires a high
diastereoselectivity (ds).
1
13
analysed by GC, and H and C NMR spectroscopy.
All samples were X-ray amorphous and exhibited high
surface areas connected with their mesopores. Nevertheless,
the acidic properties, e.g. the nature, the strengths and the
densities of their acidic sites, were quite different, as a function
Although heterogeneous catalysts are more attractive,
almost all of those reported are less selective and form
all four isopulegol isomers, as well as having a ds for
8
,10
(
Æ)-isopulegol of 52–75%. Therefore, much attention has been
of the preparation method (Table 1).
The cyclisation of citronellal to isopulegol is catalysed by
paid to the search for highly active and diastereoselective
4
heterogeneous catalysts for the cyclization of citronellal.
7
Thus, a recent report by Corma and Renz showed that
–6
7
12
Lewis and/or Brønsted acid sites, but according to Chuah
13
et al., the desired heterogeneous catalysts for citronellal cyclisa-
Sn-beta zeolite catalyses the cyclisation of citronellal in aceto-
5
tion should have strong Lewis and weak Brønsted acidity.
nitrile with a ds for (Æ)-isopulegol of 85%. Chuah et al. also
An initial screening of the catalysts showed that the HS-AlF
3
showed that over a Zr-beta zeolite catalyst, the cyclisation of
citronellal takes place with a high ds for (Æ)-isopulegol (93%).
Recently, we have found and developed novel nanoscopic,
partly hydroxylated inorganic fluorides with bi-acidic (Lewis/
sample was the most active and selective catalyst among the
tested samples (Table 2). Therefore, in the presence of this
sample, the conversion was almost complete after 1 h, and the
desired product—isopulegol—was obtained with a selectivity of
90.4%. Among the four isopulegol isomers, only (Æ)-isopulegol
8
Brønsted) properties. The catalysts’ synthesis follows the clas-
1
13
sical sol–gel route for high surface metal fluorides (characterised
9
by pure Lewis acidity) and a modified sol–gel route, in which
and (Æ)-neo-isopulegol were identified by H and C NMR
spectroscopy,w in a ratio of 85.5 : 14.5 (Table 2, entry 1).
Recently, using IR spectroscopic investigations of CO adsorp-
selective solvolysis of the original M–OR bond by an aqueous
HF solution results preferably in M–F, and to a minor extent,
M–OH bond formation. Thus, Lewis acid sites (partly coordi-
nated metal) and Brønsted sites (acidic M–OH groups) are
3
tion at surfaces, we showed that HS-AlF displays a large
number of very strong Lewis acid sites with just a few weak
a
Humboldt-Universita¨t zu Berlin, Institut fu¨r Chemie, Brook-Taylor-
Straße 2, D-12489 Berlin, Germany.
E-mail: erhard.kemnitz@chemie.hu-berlin.de; Fax: +49 3020937277
University of Bucharest, Faculty of Chemistry, Department of
b
Chemical Technology and Catalysis, Bdul Regina Elisabeta 4-12,
030016 Bucharest, Romania
Scheme 1 The cyclisation of citronellal to isopulegols.
This journal is ꢀc The Royal Society of Chemistry 2009
4
60 | Chem. Commun., 2009, 460–462

View Article Online
Table 1 Textural and chemical characterisation of the catalysts
BET surface Pore Quantity of Number of
but of a similar strength (Fig. 1). Furthermore, CO-IR
investigations of these samples showed that both sites are
8
2
Entry Catalyst area/m g
À1
À1a
17
À2b
m
centres of medium strength.
˚
size/A AS/mmol g
AS/Â10
MgF -100 and MgF -87 samples, also characterised by the
2
2
1
2
3
4
5
6
7
HS-AlF
AlF -H
3
323
187
107
65
23
25
24
23
20
0.978
0.624
0.168
0.262
0.163
0.332
0.537
18.2
20.0
5.4
6.6
3.6
predominant presence of Lewis acid sites, with only a few
Brønsted acid sites on the surface, are significantly less active
3
MgF
MgF
MgF
MgF
MgF
2
-40 180
2
-57 235
2
-71 276
2
-87 424
2
-100 372
and selective for isopulegols than HS-AlF
3
(Table 2, entries 1,
6
and 7). Their low catalytic performances are due to two
4.8
8.4
important parameters: the density and the strength of their
acid sites, both of them being at the opposite end of the scale
a
AS = acidic sites. Determined from NH
Calculated from NH -TPD measurements and N
3
-TPD measurements.
sorption isotherms.
compared to HS-AlF
3
. On the other hand, the ds for (Æ)-
b
3
2
isopulegol is almost 100% (Table 2, entry 7). The high ds
observed for this sample may be due to steric hindrance
imposed by the catalyst’s pore diameter, the most favored
geometrical diastereomers of isopulegols being formed in
excess (Table 1, entries 1 and 7). A similar effect of pore size
Table 2 Comparison of the catalysts in terms of conversion (X), yield
(
a
Y), selectivity (S) and diastereoselectivity (ds)
1
5
Entry
Catalyst
HS-AlF
AlF -H
MgF
MgF
MgF
MgF
MgF
X (%)
Y (%)
S (%)
ds (%)
upon the ds was also observed by Murzin et al.
b
The fine tuning of Lewis/Brønsted acid site ratios seems to
be essential for this synthesis (Table 2, entries 3–7). Therefore,
the existence of a predominant density of Brønsted acid sites
1
3
99.1
94.7
24.3
46.6
95.0
5.7
89.6
80.0
18.2
35.6
82.6
0.9
90.4
84.4
75.1
76.4
87.0
16.0
36.7
85.5
83.3
75.3
78.7
84.7
499.0
499.0
2
3
4
5
6
7
3
2
2
2
2
2
-40
-57
-71
-87
-100
(
MgF
the absence of Brønsted acid sites (or only a very low density
of these) makes MgF -100 and MgF -87 almost non-active.
2
-40) leads to a relatively low yield of isopulegols, while
3.0
1.1
2
2
a
On the other hand, irrespective of the catalysts’ nature, the ds
for (Æ)-isopulegol is higher than 75%, the value characteristic
of a thermodynamic equilibrium of an isopulegol mixture.
Reaction conditions: 860 mg citronellal, 5 ml toluene, 6 h, 80 1C;
citronellal/catalyst molar ratio = 180 : 1. 1 h of reaction.
b
14
16
Lewis acid sites at the surface. Interestingly enough, during the
course of these measurements, it turned out that handling this
solid, very strong Lewis acid in air (several seconds for sample
preparation) created weak Brønsted acid sites at the surface due
to the interaction of the very strong Lewis acid sites with water
molecules from airborne moisture. However, these weak
Brønsted acid sites could be reversibly converted into Lewis acid
sites by an appropriate thermal treatment at around 300 1C.
Contrary to the statement of Ravasio et al., who claimed a
correlation between the stereoselectivity for (À)-isopulegol
and the strength of the Lewis acid sites, in our case, there
seemed to be no relationship between these two parameters.
The activity, overall selectivity and isomer distribution were
also affected by the solvent polarity, reaction temperature and
amount of citronellal used in the reaction. Table 3 compiles
the solvents that were used in this study, taking HS-AlF as
3
Since the HS-AlF
3
samples used in our catalytic reactions were
the example. After 1 h, the conversion of citronellal was lower
stored and handled in air, they consequently provided—beside
very strong Lewis sites—some weak Brønsted sites.
than 34% in i-propanol, but higher than 99% in toluene
3
On the other hand, an IR spectrum of AlF -H immediately
displays two bands: an intensive band assigned to medium
À1
Brønsted acid sites (2172 cm ) and a band assigned to strong
À1
Lewis acid sites (2207 cm ), which only develops as a small
1
0
feature. Therefore, the Lewis : Brønsted acid site ratio is
quite small. As long as both samples display a similar number
of acid sites (Table 1, entries 1 and 2) per square metre, the
catalytic performance of these two samples can be explained in
terms of the different nature of their acidity (Lewis/Brønsted)
and their acidity strength. Therefore, in accordance with
1
3
Chuah et al.,
the desired heterogeneous catalysts for
citronellal cyclisation should have strong Lewis and weak
Brønsted acidity, which is perfectly fulfilled in case of
HS-AlF . Nevertheless, the activity and selectivity of AlF -H
3
3
is only slightly lower than that of HS-AlF (Table 2, entry 2).
3
Therefore, we cannot neglect a cyclization mechanism based
on almost exclusively medium Brønsted acid sites.
In the case of the MgF
obtained in the presence of MgF
Pyridine-IR measurements show that all of the catalysts in
the MgF series are characterised by the simultaneous pre-
2
sample series, the best results were
2
-71 (Table 2, entry 5).
2
Fig. 1 Infrared spectra (Py-IR) of the sol–gel nanoscopic MgF2
sence of Lewis and Brønsted acid sites, in different amounts
samples.
This journal is ꢀc The Royal Society of Chemistry 2009
Chem. Commun., 2009, 460–462 | 461

View Article Online
Table 3 The influence of the solvent’s nature upon HS-AlF
performance
3
catalytic
reactions. Their catalytic performance was influenced by both
the ratio and strength of their Brønsted/Lewis acid sites, and the
reaction conditions. Therefore, the best activities and selecti-
vities for isopulegols were obtained in the presence of catalysts
with very strong Lewis and weak Brønsted acidities. This
a
Dielectric
constant (e) X (%) Y (%) S (%) ds (%)
Entry Solvent
1
2
3
4
Toluene 2.40
Cyclohexane 2.02
99.1
86.0
87.3
33.7
89.6
81.2
81.0
26.5
90.4
94.4
92.9
78.5
85.5
84.0
89.3
28.3
optimum ratio could be adjusted by tuning the HF : H O ratio
2
n-Heptane
i-Propanol
1.92
20.30
during the synthesis. The ds for (Æ)-isopulegol was superior
(
91.7%) to most conventional heterogeneous catalysts used for
a
Reaction conditions: 860 mg citronellal, 5 ml of solvent, 1 h, 80 1C;
citronellal/catalyst molar ratio = 180 : 1.
this reaction and seems also to be influenced by the textural
properties of the catalyst.
Dr S. M. Coman is a fellow of the Alexander von Humboldt
Foundation.
(
Table 3, entries 1 and 4). The selectivity followed the same
trend as the conversion; the values obtained in non-polar
media being higher than in polar media. A better definition
of polarity should take into account the strength of adsorption
of a substrate onto an adsorbent material (Snyder’s eluotropic
Notes and references
1
10
). Isopulegol: 4.87 (m, C HH, cis to
w H NMR (400 MHz, CDCl
3
1
0
3
, 1H), 3.43 (dt, C H(OH),
CH
3
, 1H), 4.83 (s, C HH, trans to CH
3
2
J = 10.4 and 4.3 Hz, 1H), 2.03 (m, C HH, 1H), 1.86 (m, C H, 1H),
4
1
7
series). Strong solvent adsorption site interactions generate
strong competition between the substrate and solvent for these
9
.69 (d, C H
5
6
, J = 0.5 Hz, 3H), 1.65 (m, C HH, C HH, 2H), 1.49
1
3
1
m, C H, 2H), 1.32 (m, C HH, 1H), 0.97 (m, C HH, C HH, 2H) and
5
2
6
(
0.92 (d, C H , J = 6.5 Hz, 3H). Neo-isopulegol: 3.67 (dt, C H(OH),
J = 10.4 and 4.2 Hz, 1H).
7
3
sites. Therefore, as the HS-AlF
3
catalyst is a very strong Lewis
3
acid, the predominant adsorption of i-propanol renders the
active site inaccessible to substrates, and consequently results
in a low concentration of them on the catalytic surface.
1 G. S. Clark, Menthol, Perfumer Flavorist, 1998, 25, 33.
2 Y. Nakatani and K. Kawashima, Synthesis, 1978, 147.
3
4
S. Otsuka and K. Tani, Synthesis, 1991, 665.
C. Milone, A. Parri, A. Pistone, G. Neri and S. Galvagno, Appl.
Catal., A, 2002, 233, 151.
Consequently,
a weak interaction between the solvent
(
i.e., non-polar solvents) and the catalyst surface allows a
higher concentration of the substrate to be adsorbed onto
the catalyst surface, and implicitly leads to a faster reaction.
The decrease in selectivity for isopulegols in i-propanol
5 Z. Yongzhong, N. Yuntong, S. Jaenicke and G. K. Chuah,
J. Catal., 2005, 229, 404.
6
7
8
K. Arata and C. Matsuura, Chem. Lett., 1989, 1788.
A. Corma and M. Renz, Chem. Commun., 2004, 550.
S. Wuttke, S. M. Coman, G. Scholz, H. Kirmse, A. Vimont,
M. Daturi, S. L. M. Schroeder and E. Kemnitz, Chem.–Eur. J.,
DOI: 10.1002/chem.200801702.
(
Table 3, entry 4) can be explained by the citronellal reduction
to citronellol in the Meerwein–Ponndorf–Verley (MPV) reac-
1
tion and the subsequent dehydration of the products.
8
9
¨
S. Rudiger, U. Groß and E. Kemnitz, J. Fluorine Chem., 2007, 128,
The results obtained in different solvents were also con-
firmed by tests with different concentrations of substrate
solution. Therefore, when the substrate concentration
increased from 9 to 16.5 and 23 wt%, the TON increased
from ca. 56 to 112 and 165, respectively.
3
10 S. M. Coman, S. Wuttke, A. Vimont, M. Daturi and E. Kemnitz,
53.
Adv. Synth. Catal., 350, 2517.
¨
1 E. Kemnitz, U. Groß, S. Rudiger and C. S. Shekar, Angew. Chem.,
Int. Ed., 2003, 42, 4251.
2 D.-L. Shieh, C.-C. Tsai and A.-N. Ko, React. Kinet. Catal. Lett.,
2003, 79, 381.
3 G. K. Chuah, S. H. Liu, S. Jaenicke and L. J. Harrison, J. Catal.,
1
1
1
1
One attempt to improve ds values by lowering the reaction
1
9
temperature, as has been done successfully in other cases, did
not result in any significant improvement. Decreasing the
temperature from 80 to 40 1C slight decreased the selectivity
2
4 T. Krahl, A. Vimont, G. Eltanany, M. Daturi and E. Kemnitz,
001, 200, 352.
J. Phys. Chem. C, 2007, 111, 18317.
15 P. Maki-Arvela, N. Kumar, V. Nieminen, R. Sjoholm, T. Salmi
and D. Y. Murzin, J. Catal., 2004, 225, 155.
obtained (from 91.8 to 88.3%) in the case of HS-AlF
3
, but the
conversion decreased significantly (from 99.3 to 41.6%). The
best results in terms of selectivity for isopulegols and ds for
1
6 N. Ravasio, M. Antenori, F. Babudri and M. Gargano, Stud. Surf.
Sci. Catal., 1997, 108, 625.
17 Y.-M. Chung and H.-K. Rhee, J. Mol. Catal. A: Chem., 2001, 175,
(
Æ)-isopulegol were obtained at 60 1C, while the conversion
remained more or less at the same level (X = 97.4%, S =
249.
8 Z. Yongzhong, N. Yuntong, S. Jaenicke and G.-K. Chuah,
J. Catal., 2005, 229, 404.
1
9
2.3%, ds = 91.7%).
In summary, nanosized inorganic fluoride catalysts have been
1
9 V. K. Aggarwal, G. P. Vennall, P. N. Davy and C. Newman,
Tetrahedron, 1998, 39, 1997.
employed for the first time in intramolecular carbonyl-ene
4
62 | Chem. Commun., 2009, 460–462
This journal is ꢀc The Royal Society of Chemistry 2009