A study of V-pillared layered double hydroxides as catalysts for the epoxidation of terpenic unsaturated alcohols

Article abstract of DOI:10.1006/jcat.2000.3104

Mg,Al-Layered double hydroxides (LDHs) are pillared with vanadium under mildly acidic conditions. Via XRD, Raman, and IR characterization, it is confirmed that the material is a phase-pure LDH with decavanadate (V₁₀O₂₈^6-) anions between the brucite sheets. This V₁₀-LDH has the highest activity for the epoxidation of geraniol when used with tert-butyl hydroperoxide as the oxidant in anhydrous toluene. In a series of tests, the possible release of catalytically active dissolved V from the catalyst is investigated. Under all conditions tested, V₁₀-LDH behaves as a heterogeneous catalyst, in contrast with some other tested V-containing materials (e.g., VAPO-5, or V in a siliceous matrix). Regioselectivity in the epoxidation of geraniol evidences the coordination of the alcohol group of geraniol on the active V center. Finally, it is demonstrated that the V₁₀-LDH catalyst can be applied to the epoxidation of a series of allylic and homoallylic alcohols, mostly of terpene origin.

Full text of DOI:10.1006/jcat.2000.3104

Journal of Catalysis 198, 223–231 (2001)
doi:10.1006/jcat.2000.3104, available online at http://www.idealibrary.com on
A Study of V-Pillared Layered Double Hydroxides as Catalysts
for the Epoxidation of Terpenic Unsaturated Alcohols
�
�
�
�
,1
Aida L. Villa, Dirk E. De Vos, Francis Verpoort,† Bert F. Sels, and Pierre A. Jacobs
�
Centre for Surface Chemistry and Catalysis, Katholieke Universiteit Leuven, Kardinaal Mercierlaan 92, 3001 Leuven, Belgium; and †Department
of Inorganic and Physical Chemistry, University of Gent, Krijgslaan 291 (S-3), 9000 Gent, Belgium
Received July 10, 2000; revised October 13, 2000; accepted October 13, 2000; published online February 13, 2001
of dissolved metal, typically VO(acac)2. Initial attempts to
Mg,Al-Layered double hydroxides (LDHs) are pillared with immobilize such V catalysts have focused on V-substituted
vanadium under mildly acidic conditions. Via XRD, Raman, and aluminophosphates, V silicates, or V-pillared montmoril-
IR characterization, it is confirmed that the material is a phase-
lonite (7–10); however, the mixed success of these materials
proves that there is a need for the exploration of new, solid
catalytic materials.
Several groups have recently reported on the use of
metal-exchanged layered double hydroxides (LDHs) as
catalysts for liquid phase oxidations (11–13). While LDHs,
due to their basic nature, display catalytic activity in the
epoxidation of, e.g., unsaturated ketones (14), their use as
pure LDH with decavanadate (V10O6 ) anions between the brucite
�
28
sheets. This V10-LDH has the highest activity for the epoxidation
of geraniol when used with tert-butyl hydroperoxide as the oxidant
in anhydrous toluene. In a series of tests, the possible release of
catalytically active dissolved V from the catalyst is investigated.
Under all conditions tested, V10-LDH behaves as a heterogeneous
catalyst, in contrast with some other tested V-containing materials
(e.g., VAPO-5, or V in a siliceous matrix). Regioselectivity in the
epoxidation of geraniol evidences the coordination of the alcohol catalysts for many other reactions requires exchange with
group of geraniol on the active V center. Finally, it is demonstrated a catalytically active anion (15). For instance, tungstate-
that the V10-LDH catalyst can be applied to the epoxidation of a exchanged LDHs have been used as catalysts for the
series of allylic and homoallylic alcohols, mostly of terpene origin.
�
c 2001 Academic Press
Key Words: vanadium; layered double hydroxides; geraniol; ter-
penes; epoxidation.
biomimetic oxidation of bromide anions with H2O2 as the
oxidant (16); in contact with the same H2O2, molybdate-
exchanged LDHsare solid catalystsfor generation ofdioxy-
gen in the singlet excited state (17). The success of such
catalysts is in part due to the infinite oxidative stability of
the inorganic support, and to the large anion exchange ca-
pacity, which efficiently retains the anions, particularly in
solvents of low dielectricity.
INTRODUCTION
Oxidative transformation of fine chemicals is an impor-
tant target in catalysis, particularly if one desires to obtain
products that are free of dissolved metals. Efficient hetero-
genization routes exist for, e.g., Ti and W, for instance via
isomorphous substitution of Ti into molecular sieves, or via
exchange ofW-based anions(1, 2). However, progressin the
design of heterogeneous V catalysts has been less marked.
Homogeneous V catalysts are capable of effecting olefin
epoxidation, and hydroxylation of alkanes and aromatics,
but they are particularly suitable for the epoxidation of al-
lylic and homoallylic alcohols (3–6). The latter reactions are
stereospecific and mostly chemoselective, make efficiently
use of the oxidant, and often proceed under ambient condi-
tions. However, the typical procedure for V-catalyzed epox-
idation of allylic alcohols uses up to a few mole percent
Incorporation of V anions into LDHs has been exten-
sively described, and several procedures have been es-
tablished to pillar LDHs with the polyoxoanion V10O6
�
2
8
(
18–21). However, these materials have received little at-
tention in liquid phase catalysis. There is a report on the aer-
obic photo-oxidation of isopropanol to acetone catalyzed
by V10O6 -pillared LDH (22).
�
2
8
This paper presents a detailed study of the use of a
V10O6 -pillared LDH (V10-LDH) as a catalyst for the se-
�
2
8
lective epoxidation of allylic and homoallylic alcohols. First,
standard characterization techniques are applied to check
whether the pillaring is successful, and to evaluate the state
of the V in the pillars. Next, optimum reaction conditions
are designed for the epoxidation of the typical substrate
geraniol. Under the optimized conditions, the heterogene-
ity of the observed catalysis is critically evaluated with sev-
eral tests. For the sake of comparison, similar evaluations
of other solid V materials are included. These demonstrate
1
To whom correspondence should be addressed. Fax: 32 16 32 1998.
E-mail: pierre.jacobs@agr.kuleuven.ac.be.
223
0021-9517/01 $35.00
Copyright �c 2001 by Academic Press
All rights of reproduction in any form reserved.

2
24
VILLA ET AL.
that V10-LDH is the only material of the set that can be V2O5 was added to the stirring gel, causing a color change
considered as a heterogeneous catalyst. Finally, the applica- to yellow. After 20 min of homogenizing, 15 ml of tripropy-
bility of V10-LDH to the epoxidation of a series of terpenic lamine was added. The gel was further homogenized for
and other (homo)allylic alcohols is demonstrated.
90 min and finally crystallized for 6 h at 423 K in Teflon-
lined autoclaves. The aluminophosphate was isolated and
calcined as described by Montes et al. (24). The dry material
contains 1.4 wt% V.
EXPERIMENTAL SECTION
VO-PVP. One gram of polyvinylpyridine (Reillex
Preparation of Catalysts
4
5
25 polymer) was added to a solution of 0.4 g of VOSO4 �
LDH. The precursor Mg,Al-LDH was prepared via co-
H2O in 100 ml of ethanol. After overnight shaking, the
precipitation of Mg2 and Al nitrate salts at pH 10. All
preparations of LDHs were performed with deionized wa-
ter that was decarbonated by boiling for 15 min. The typical
procedure for synthesis of Mg,Al-LDH has been described
elsewhere (23).
+
3 +
solid was filtered, extensively washed with ethanol, and
dried under vacuum at room temperature (25). The room-
temperature, vacuum-dried material contains 7.6 wt% V.
V-SiO2. To a mixture of 22.3 ml of tetraethyl orthosili-
cate (0.1 mol) and 0.39 g of Na3VO4 (2.12 mmol) in 270 ml
of water was added 8 ml of 37% HCl. After 75 h of stir-
ring at room temperature, the solid was filtered, extensively
washed with distilled water, and dried at room temperature.
Finally, the material was calcined in air by heating to 900 K
at 1 K per minute, and maintaining the final temperature
for 4 h. A material is obtained with 1.7 wt% V.
V10-LDH. For the synthesis of a decavanadate-pillared
LDH (V10-LDH), the suspension of freshly prepared
Mg,Al-LDH is used (about 1 g of solid per 100 ml of wa-
ter), without a drying step (20). The pH of this suspension
is brought from 9 to 4.5 with 2 M HNO3. Next, 1.0 g of
NaVO3 (8.2 mmol) is dispersed in 33 ml of deionized and
decarbonated water, and the pH is adjusted to 4.5 with 2 N
HNO3, resulting in a color change to the characteristic or-
ange of decavanadate. The orange decavanadate solution
is then added over 15 min to the LDH suspension, while
maintaining the pH at 4.5. After 4 h of stirring at room
temperature and centrifugation, a yellow solid is obtained,
which is washed five times to remove excess decavanadate.
The dry powder is eventually obtained by lyophilization.
This direct exposure of the LDH to a decavanadate so-
lution leads to the purest decavanadate-pillared LDH, as
indicated by physicochemical data (see below). The V con-
tent of this material on a dry weight basis is 21% .
V10-Amberlite. To 75 ml of water was added 0.456 g of
NaVO3, and the pH was adjusted to 4.5 by addition of a
Dowex 50W � 4 resin. The formation of decavanadate was
evidenced by the yellow-orange color of the solution (26).
The Dowexwasremoved byfiltration, and to 50mlofthe so-
lution wasadded 1gofAmberlite IRA-400. After overnight
shaking, the orange polymer beads were isolated by filtra-
tion, and extensively washed with distilled water.
Extraction of Decavanadate Species into a Toluene Layer
Alternatively, an indirect procedure was used to pre-
A solution of sodium decavanadate was prepared as in
pare decavanadate-pillared LDH, via intermediate pil- the previous paragraph, resulting in a 5 mM solution of
6
�
6�
laring with terephthalate anions (V10-LDH(TA)). In this V O . To 1 ml of this solution (5 �mol of V O ) were
10
28
10 28
case, the reported procedure of Drezdzon was followed added 4 ml of toluene and 0.068 g of tetrapentyl ammonium
(
18).
bromide (180 �mol). This resulted in complete transfer of
the yellow decavanadate into the toluene layer. The solu-
tion of decavanadate in toluene (1.25 mM) was used as such
in catalytic experiments.
V-LDH. Vanadate-exchanged LDH (V-LDH) was pre-
pared by dispersing 0.2 g of NaVO3 in 15 ml of decarbon-
ated water. One gram of NO -exchanged Mg,Al-LDH was
�
3
added, and the suspension was stirred for 24 h at room tem-
perature under N2. In contrast with the preparation of V10- Instrumentation
LDH, the pH was not adjusted in this case. The solid was
extensively washed, and air-dried. In an analogous proce-
dure, a terephthalate-pillared LDH was exposed to NaVO3
X-ray diffractograms were recorded with a Siemens D-
000 diffractometer. IR spectra were recorded with KBr
pellets, using a Nicolet 730 spectrometer. For Raman spec-
troscopy, a Bruker Equinox 55-FRA106/S apparatus was
5
(
V-LDH(TA)).
VAPO-5. The preparation of this V-substituted alu- employed, with excitation in the near infrared, and an ir-
minophosphate was based on a literature procedure (24). radiation power between 50 and 100 mW. Visible spectra
An 8.5 g amount of 85 wt% H3PO4 was diluted with 6 g of were recorded with a Perkin Elmer spectrophotometer. El-
H2O. This solution was added to a slurry of 5.5 g of pseu- emental analyses were performed with inductively coupled
doboehmite in 16 g of H2O, and the white viscous gel was plasma (Bodemkundige Dienst van Belgi e¨ ) after calcina-
stirred for 90 min at room temperature. A 0.35 g amount of tion of the weighed sample and dissolution in acid.

EPOXIDATION OF ALLYLIC ALCOHOLS WITH V-LDH
225
Catalytic Reactions and Product Analysis
Reactions were started by addition of the oxidant (H2O2,
aqueous tBuOOH, or 5 M tBuOOH in decane, Aldrich) to
a magnetically stirred suspension of the substrate and the
catalyst in toluene. Products were analyzed on an HP chro-
matograph with a 60 m CP-Sil 5 column. Authentic epox-
ides were prepared with m-chloroperbenzoic acid as the
stoichiometric oxidant, or using a homogeneous VO(acac)2
catalyst (1–5 mol % ) with tBuOOH in toluene.
Materials
All products were purchased in the highest grade avail-
able and used as such: 35% H2O2, cinnamyl alcohol,
H3PO4 (85% ), and tert-butyl hydroperoxide (70% so-
lution in water) (all from Akros); (-)-carveol, (R)-(-)-
carvone, citronellol, crotyl alcohol, 2-cyclohexen-1-ol,
geraniol, isophorol, (1R)-(-)-myrtenol, nerol, (1R)-(-)-
nopol, (S)-(-)-perillyl alcohol, (S)-cis-verbenol, (1S)-(-)-
verbenone, Mg(NO3)2 � 6H2O, tert-butyl hydroperoxide
(
(
5.0–6.0 M solution in decane), and Amberlite IRA-900
all from Aldrich); (-)-isopulegol, (-)-trans-pinocarveol,
FIG. 2. Raman spectra of (a) decavanadate-pillared LDH, prepared
by direct exposure of an LDH to a decavanadate solution (V10-LDH),
b) decavanadate-pillared LDH, prepared by exposure of a terephthalate-
pillared LDH to decavanadate (V10-LDH(TA)), (c) terephthalate-pillared
LDH, and (d) vanadate-LDH, prepared by exposure of an LDH to NaVO3
without pH adjustment.
tetraethyl orthosilicate, tripropylamine, vanadyl acetylace-
(
tonate, Al(NO3)3 � 9H2O, NaVO3, and Dowex 50W � 4
(
Fluka); and VOSO4 � 5H2O, V2O5, and HClconc (Merck).
RESULTS AND DISCUSSION
positions of the second and the third reflections. The value
Characterization of the V10-LDH Catalyst
c/3 = 11.8 A˚ is in excellent agreement with values reported
in the literature (18, 19, 22).
XRD. X-ray diffraction is the easiest method to moni-
tor the spacing of the brucite layers in layered double hy-
droxide type materials. For the V10-LDH material prepared
by direct exposure to decavanadate anions, reflections are
observed at 10.2, 5.89 (006), and 3.92 A˚ (009) (Fig. 1a). As
Alternatively, samples were prepared by intercalation of
terephthalate anions, and subsequent intercalation with de-
cavanadate anions. For the terephthalate-intercalated sam-
ple, there are reflections at 14.07, 7.19, and 4.71 A˚ (Fig. 1b).
The resulting c/3 value of 14.2 A˚ corresponds to a per-
the first reflection is somewhat broadened, it is most ap-
pendicular orientation of the terephthalate anions between
the brucite layers (18, 19). Subsequent exposure of the
terephthalate-intercalated LDH to a decavanadate solution
results in complete phase transformation. The new mate-
rial has reflections at 10.6, 5.86, and 3.94 A˚ , which is in close
propriate to calculate an average basal spacing c/3 from the
agreement with the structure obtained by direct exposure
of the LDH to decavanadate (vide supra).
For all samples, the detection of the (110) plane at 1.52 A˚
proves that the layered double hydroxide structure of the
material is maintained. Moreover, the characteristic reflec-
tions of carbonate-containing LDHs (e.g., 7.6 A˚ ) were not
observed, indicating that major CO2 contamination was
succesfully avoided during synthesis.
Vibrational Spectroscopy
Raman. In order to characterize the nature of the in-
terlayer anions, and especially the state of the V, Raman
spectra were recorded (Fig. 2). In the Raman spectra of
FIG. 1. X-ray patterns of (a) V10-LDH and (b) terephthalate-pillared
LDH.
�
1
most LDH samples, a peak at 557 cm can be assigned to

2
26
VILLA ET AL.
the lattice vibrations of the brucite octahedral layers (23).
Raman data are particularly useful to distinguish between
TABLE 1
Optimization of the Reaction Conditions in the Epoxidation
such species as VO3 , V2O7 , V4O , and V10O28 (27–29).
�
4�
4�
6�
a
4
12
of Geraniol with V -LDH as the Catalyst
1
0
In the Raman spectrum of V10-LDH, prepared by direct
exchange of an LDH with an acidified V solution, deca-
vanadate is the dominant species, as evidenced by peaks at
Time, Conversion, Selectivity,
b
Entry Solvent
Oxidant
h
%
%
� 1 � 1
9
93 and 322 cm , and a shoulder at 965 cm (Fig. 2a). A
1
2
3
4
5
Dioxane H2O2, 50%
24
4
4
4
4
9
39
55
96
98
97
31
highly similar spectrum is observed for aqueous vanadate at
a pH between 3 and 6 (27). Additional decavanadate peaks
Dioxane tBuOOH, 70%
CH3CN tBuOOH, 70%
Toluene tBuOOH, 70%
Toluene tBuOOH, decane
Toluene tBuOOH, decane
4
28
59
95
3
�
1
are observed at 600 and 465 cm . Only the minor peak in
�
1
the spectrum at 827 cm seems somewhat shifted from the
reported value for decavanadate (840 cm ).
c
6
34
�
1
a
The terephthalate-intercalated LDH has characteristic
vibrations of the aromatic carboxylate anion at 3075, 1610,
Conditions: 1 mmol of geraniol, 1.1 mmol of oxidant, 20 mg of catalyst,
ml of solvent, 293 K. Conversion of geraniol in % .
2
b
�
1
Selectivity = 100 � (moles of epoxide formed)/(moles of geraniol
1
413, 1130, 855, and 634 cm (Fig. 2c). After intercala-
converted). Except for entry 6, the ratio (2, 3)/(6, 7) epoxide is at least 50.
The main side product is geranial.
tion with decavanadate (V10-LDH(TA)), a small amount
of residual terephthalate was observed in some cases, as
c
No catalyst; (2, 3)/(6, 7) epoxide = 1.
�
1
evidenced by the subsistence of, for instance, the 1610 cm
peak (Fig. 2b). The most intense peaksofV10-LDH(TA) can
be assigned to intercalated decavanadate (996, 963, 833, and
lyst (Table 1). A reaction with H2O2 resulted in poor con-
version and low epoxide selectivity. Better results were ob-
tained with tBuOOH, even if the reaction outcome depends
markedly on the solvent used. With 70% tBuOOH in water
as the oxidant, dioxane was not an effective solvent, while
better results were obtained in acetonitrile and particularly
in toluene (entries 2–4). However, the best yield and selec-
tivity are obtained when 5 M tBuOOH in decane is used as
the oxidant, in toluene as the solvent (entry 5). The chemos-
electivity is excellent, with minimal formation of geranial.
Moreover, the regioselectivity for the formation of the 2,3-
epoxide is high (>98% ). Hence, it seems that the V10-LDH
is particularly effective if anhydrous conditions are applied.
For comparison, an LDH was tested in which the deca-
vanadate was introduced after first pillaring with terephtha-
late (V10-LDH(TA)). Eventual conversions and selectivi-
ties are practically indistinguishable from those obtained
with V10-LDH, prepared by direct pillaring (Table 2, en-
tries 1 and 2). Both catalysts fully preserve their character-
istic yellow color during and at the end of the reaction; the
filtrate is completely colorless. In contrast with the V10 cat-
alysts, low conversions and selectivities are obtained if the
vanadate is introduced on an LDH support without acidi-
fication (Table 2, entries 3 and 4).
�
1
3
23 cm ). Obviously, incomplete exchange of terephtha-
late by decavanadate is not a problem when the V10-LDH is
prepared by direct exposure of the LDH to decavanadate.
Hence, it seems that the direct procedure leads to purer
samples than the indirect method via terephthalate inter-
calation.
For comparison, Fig. 2 also includes the Raman spectrum
of the vanadate-exchanged V-LDH material, which was ob-
tained without acidification (Fig. 2d). Apart from a strong
�
1
and a weak nitrate vibration at 1061 and 710 cm , respec-
�
1
tively, the major band is observed at 935 cm , together with
weak bands at 474 and 352 cm . This spectrum matches
�
1
that of aqueous vanadate solutions at a pH between 8 and
n�
1
1, and corresponds to cyclic species (VO3)n , with n at
least 3, in which VO2 units are linked by –O– bridges (27).
IR. IR spectra are in general less useful for the study of
these LDH samples, since the LDH structure itself strongly
�
1
absorbs below 1000 cm . For V10-LDH, the only band that
�
1
can be assigned to V species is located at 960 cm , which
is the frequency of dissolved decavanadate at pH 4.5 (27).
In the IR spectrum of V10-LDH(TA), some residual tereph-
thalate may be observed at 1405 and 1567 cm , in addition
to the decavanadate V == O stretching vibration at 960 cm
�
1
�
1
.
Decavanadate, anion exchanged on an Amberlite, was
used under the same conditions as the V10-LDH. The gen-
eral course of the reaction is similar to that with V10-LDH:
the reaction starts without any induction period, and the
2,3-epoxide is the main product. However, much more
time is needed to reach even modest conversion levels
Hence, there is full agreement between the Raman and IR
data.
Summarizing, the physicochemical evidence proves that
V10O6 is the prevailing anion when the LDHs are exposed
�
2
8
to vanadate solutions at pH 4–5, whether the LDH has been
first pillared with terephthalate or not.
(
Table 2, entry 5). It was noted that the yellow color of
the decavanadate-loaded beads was preserved during the
reaction of the V10-Amberlite catalyst.
Finally, a solution of decavanadate anions, solubilized in
Optimization of the Reaction Conditions with V10-LDH
and Comparison with Other Decavanadate Catalysts
Several sets of reaction conditions were tested, with toluene by ion pairing with tetrapentyl ammonium cations,
geraniol as a model substrate and V10-LDH as the cata- was used as a catalyst, with a geraniol:V ratio of 100, under

EPOXIDATION OF ALLYLIC ALCOHOLS WITH V-LDH
227
TABLE 2
decavanadate, immobilized by ion exchange on an organic
or inorganic solid.
Epoxidation of Geraniol with Solid V Catalystsa
Comparison of a Series of Solid V Catalysts
Epoxide
Conversion, selectivity,
Other products
(selectivity, % )
Table 2 gives a comparison of various solid organic and
inorganic V catalysts for the epoxidation of geraniol. With
Entry
Catalyst
V10-LDH
V10-LDH(TA)
V-LDH
V-LDH(TA)
V10-Amberlite
%
%
2+
1
2
3
4
5
6
80
86
4
97
98
77
0
Geranial (1)
Geranial (1)
VOSO4 � H2O as a V source, VO was introduced by coor-
dinative bonding on polyvinylpyridine (VO-PVP) and by
ion exchange on the acid cation exchanger Amberlyst 15.
With the latter material, residual acidity lowers the epoxide
selectivity(entry6);an excellent epoxide yield wasobtained
with VO-PVP (entry 7).
Geranial (20)
Geranial (100)
Geranial (1)
Geranial (2),
products of acid
rearrangement (72)
Geranial (2)
8
b
15
64
98
26
VO-Amberlyst
c
1
5
Other solids that are effective in the epoxidation of
geraniol are the crystalline V-substituted aluminophos-
phate (VAPO-5, entry 8), and a silica-containing dispersed
7
8
9
VO-PVP
95
97
85
97
97
80
d
VAPO-5
Geranial (1)
Geranial (3)
d
V-SiO2
V
V prepared via a sol–gelprocess(V-SiO2, entry9). A larger
a
General reaction conditions: 1 mmol of geraniol, 1.1 mmol of weight of catalyst was used in the latter two cases, since the
tBuOOH in decane, 15 mg of catalyst, 2 ml of toluene, 293 K, 4 h (unless
V content of these solids is lower (1.4 wt% V for VAPO-
stated otherwise).
b
5, and 1.7 wt% V for V-SiO2, vs 7.6 wt% V for VO-PVP,
With 30 mg of catalyst.
Prepared by exchange of commercial Amberlyst-15 (in the H form)
c
+
and 21 wt% V for V -LDH). Only a meticulous evaluation
10
with excess VOSO4.
of the heterogeneous nature of the observed catalysis can
reveal the true value of these materials as heterogeneous
catalysts.
d
With 60 mg of catalyst.
conditions that are otherwise identical to those of the het-
erogeneously catalyzed reactions. In this case, the visible
spectrum of the solution was monitored in parallel to the
progress of the reaction. The yellow-colored decavanadate
has intense absorption between 400 and 500 nm. Remark-
ably, this decavanadate solution initially displays hardly any
activity. When the visible spectrum indicates that at least
Stability of the V Under the Reaction Conditions
Different methods were applied to determine whether
the catalytic activity was due to homogeneous or really
supported species. The catalysts under scrutiny were V10-
LDH, VAPO-5, VO-PVP, and V-SiO2. The optimized con-
ditions for the V10-LDH catalyst were applied, i.e., anhy-
drous tBuOOH (0.55 M) as the oxidant in toluene, with
geraniol (0.5 M) as the substrate.
5
4
0% of the decavanadate anions have disappeared (after
h), the reaction suddenly starts (Fig. 3). Thus, surpris-
ingly, the catalytic properties of quaternary ammonium-
solubilized decavanadate are quite different from those of
In the first type of experiment, the role of dissolved vana-
dium was studied by splitting the reaction suspension at a
conversion level between 10 and 20% . Next, the progress
of the reaction was monitored in the remaining suspen-
sion and in a filtrate, obtained by low-speed centrifugation
and filtration over a standard HPLC filter. An example of
such an experiment is shown in Fig. 4, for the reaction of
geraniol with tBuOOH over V10-LDH. It is obvious that
the progress of the reaction in the clear filtrate is negligible.
Moreover, the reaction starts immediately, without any no-
ticeable induction time, and the rate is practically constant
up to 20% conversion. This indicates that the catalytic ac-
tivity is not due to slow dissolution of the V species, since
this would entail a gradually increasing reaction rate. A
practically identical result is obtained when crotyl alcohol
is used as the substrate. Thus, the type of allylic alcohol does
not affect the stability of the V in V10-LDH. For the other
vanadium catalysts, slight to substantial increases in geran-
iol conversion were observed after solid–liquid separation,
and other heterogeneity tests were applied.
FIG. 3. Epoxidation of geraniol, catalyzed by decavanadate, solubi-
+
lized by (C5H11)4N in toluene. Evolution of the absorbance at 400 nm
(
A400; ᭹) and the 2,3-epoxygeraniol yield (Y, % ; ᭡) as a function of time.
Conditions: 1 mmol of geraniol, 1.1 mmol of tBuOOH, 5 M in toluene,
�mol of decavanadate, 2 ml of toluene, room temperature.
In the second test, the solid catalysts were pre-exposed
to the oxidant in the absence of an allylic alcohol. Thus,
1

2
28
VILLA ET AL.
lar reaction, and was compared to a fresh material. This
test reveals particularly clear results. With V10-LDH, pre-
exposure to oxidant does not affect the activity of the solid,
while the solution contains only a hardly perceptible activ-
ity, comparable to the background reaction without cata-
lyst. However, pre-exposure to the oxidant causes an at
least 50% reduction of the activity of the VAPO-5, VO-
PVP, and V-SiO2. Moreover, in the latter three cases, the
filtrate has an appreciable activity.
In the third approach, a regular reaction was conducted.
At the end of the reaction, the catalyst and the clear solu-
tion were separated by centrifugation. The various catalysts
were dried (6 h at 383 K for V10-LDH and V-SiO2; 6 h at
3
38K for VAPO-5;6h under vacuum at 296K for VO-PVP);
they were then used in the next reaction (“catalyst reuse”
FIG. 4. Heterogeneity test in the epoxidation of geraniol with V10- in Table 3). With V -LDH, the original activity of the first
1
0
LDH (2 mmol of geraniol, 2.2 mmol of tBuOOH, 8 mg of catalyst, 4 ml
of toluene, 293 K). After 120 min, catalyst was removed from half of
the reaction suspension. Reaction progress in the filtrate (᭺) was then
compared to that in the suspension (�).
run was fully recovered in a subsequent run. With VAPO-5,
a mild drying treatment is apparently not sufficient. Only
after calcination, the catalytic activity of a reused catalyst
approaches that of the fresh material. Finally, major activity
decreases were observed for V-SiO2 and for VO-PVP, even
after 2 h of stirring at room temperature in a solution of if several regeneration treatments were investigated.
tBuOOH in toluene, the catalyst was removed by cen-
In the fourth and final approach, a reaction was con-
trifugation. Next, allylic alcohol was added to the super- ducted with 1 mmol of geraniol and 1.1 mmol of tBuOOH.
natant, and the conversion was monitored (“pre-exposure” The reaction filtrate was collected at the end of a reac-
in Table 3). On the other hand, the recovered catalysts were tion, at over 95% conversion to geraniol epoxide. Next, new
dried, and subjected to a short treatment under vacuum aliquots of substrate (1 mmol) and oxidant (1.1 mmol) were
(
333–383 K) in order to remove adsorbed organics. The added to the clear filtrate (“addition of extra substrate”
oxidant-exposed catalyst was eventually used in a regu- in Table 3). This addition reduces the overall substrate
TABLE 3
Heterogeneity of Solid V Catalysts in Epoxidation of Geraniol with tBuOOH/Decane as the Oxidanta
V10-LDH
VAPO-5
VO-PVP
V-SiO2
Regular reaction
81 (3.5 h, 97 S)
27 (3.5 h, 90 S)
93 (3.5 h, 96 S)
91 (9 h, 80 S)
After pre-exposure of catalyst to oxidant and solvent, without substrate
Filtrate + substrate
6 (24 h, 70 S)
89 (3.5 h, 96 S)
10 (24 h, 71 S)
10 (3.5 h, 90 S)
6 (20 h, 86 S)
11 (3.5 h, 87 S)
26 (20 h, 31 S)
38 (9 h, 28 S)
b
Recovered catalyst + substrate,
oxidant and solvent
Catalyst reuse
First run
Second run
81 (3.5 h, 97 S)
80 (3.5 h, 97 S)
00 (17 h, 97 S)
27 (3.5 h, 90 S)
12 (3.5 h, 85 S)
25 (3.5 h, 97 S)
100 (12 h, 96 S)
38 (17 h, 93 S)
91 (9 h, 80 S)
27 (9, 54 S)
c
1
Addition of extra substrate to the filtrate of a completed reaction
d
First reaction
81 (3.5 h, 97 S)
6 (12 h, 98 S)
48 (7 h, 94 S)
1 (24 h, 93 S)
96 (3 h, 97 S)
99 (4 h, 97 S)
—
—
—
—
9
+Extra substrate (0.55 mol per ml)
51 (7 h, 92 S)
96 (24 h, 95 S)
54 (7 h, 92 S)
98 (24 h, 88 S)
5
Note: Conversions are given in % at a given reaction time (h), as well as the selectivity in % (S).
a
Reaction conditions: 0.5 mmol of geraniol and 0.55 mmol of tBuOOH in decane per ml of solvent (toluene). Catalyst concen-
trations were 6.8 mg/ml (V10-LDH, VAPO-5), 3.6 mg/ml (VO-PVP), and 13.6 mg/ml (V-SiO2).
b
Catalysts were dried before reuse: 6 h at 383 K for V10-LDH and V-SiO2; 6 h at 338 K for VAPO-5; 6 h under vacuum at 296 K
for VO-PVP.
c
After calcination of the spent catalyst, 4 h at 823 K.
With 25 mg of VAPO-5 per ml.
d

EPOXIDATION OF ALLYLIC ALCOHOLS WITH V-LDH
229
conversion to about 50% . If the reaction is really heteroge- the two geminal methyl groups. For other molecules with
neous, the conversion level should stay stationary around the 6,6-dimethylbicyclo[3.1.1]heptene structure (myrtenol,
5
0% . The latter situation is indeed observed with V10-LDH, trans-pinocarveol, nopol), the epoxidation occurs exclu-
even if the filtrate is stirred for 24 h at room temperature sively at the sterically less hindered side of the molecule.
Table 3). However, VAPO-5 and VO-PVP clearly fail in Some nonterpenic allylic alcohols were also subjected to
(
this test: over the 24 h after the addition of extra substrate the epoxidation. High conversions with excellent selectiv-
and oxidant, the substrate conversion gradually increases ity were obtained for crotyl alcohol and cinnamyl alcohol
to over 95% .
(entries 10 and 11). While the primary alcohol functions are
Summarizing, only in the case of the V10-LDH catalyst not oxidized in the presence of a double bond, secondary
can it be proven that the filtrates contain no catalytically ac- alcohols are more susceptible to ketonization, particularly
tive V compounds, while the recovered solid has an activity the cyclic structures such as carveol or 2-cyclohexen-1-ol.
that matches that of a fresh catalyst. Note that all sepa- Thus, appreciable amounts of the �,�-unsaturated ketones
rations of catalyst and liquor were performed at the same are formed (entries 12 and 13).
temperature as the reactions themselves (293 K); hence,
readsorption of dissolved catalytic species upon cooling of
reaction mixtures, which is a frequently encountered prob-
GENERAL DISCUSSION
lem, is not an issue in the present experiments.
While the heterogeneousnature ofsolid V catalystsin liq-
uid phase oxidations has been frequently questioned in the
past (7–9), the extensive tests in the present work demon-
Applicability to Terpene Olefinic Alcohols
Having established that V10-LDH is a truly heteroge- strate that V10-LDH can be considered as a really hetero-
neous catalyst for the epoxidation of geraniol, the potential geneous epoxidation catalyst. Nevertheless, it is important
of this catalyst for the epoxidation of a series of other un- to emphasize that this claim only pertains to the specified
saturated alcohols, mostly of terpene origin, was evaluated reaction conditions. For instance, we have experienced that
(
Table 4). The standard optimum conditions, as defined in a too severe heat treatment of V10-LDH (e.g., overnight at
�
the case of geraniol, viz., tBuOOH in decane, were applied. 200 C) results in some leaching when the material is after-
High conversions and epoxide selectivities were obtained ward used in a catalytic test, probably because of a minor
for a series of terpenic allylic alcohols, such as geraniol, structural degradation.
nerol, perillyl alcohol, myrtenol, and trans-pinocarveol (en-
V10-LDH has the highest activity with tBuOOH in an-
tries 1 to 5). For those molecules that contain two dou- hydrous toluene. It is likely that the anhydrous conditions
ble bonds (geraniol, nerol, perillyl alcohol), the epoxida- of the reaction prevent extensive hydrolysis of the deca-
tion is highly regioselective and prefers the double bond vanadate pillars and ensuing V leaching. Apparently, the
in proximity of the alcohol group. This proves that also conditions for maximal activity coincide with those for an
for the heterogeneous LDH catalyst, the mechanism of optimum heterogeneity. Another important element is the
the V-catalyzed epoxidation probably involves coordina- mild reaction temperature. The allylic alcohols are easily
tion of the allylic alcohol group to the V atom before oxy- epoxidized at room temperature, whereas previous stud-
gen transfer takes place. The reactions with geraniol and ies of oxidation of, e.g., olefins or aromatics with solid V
nerol illustrate that the epoxidations are stereospecific: the catalysts often were conducted at temperatures between 60
�
reaction with geraniol only yields the 2,3-epoxide of geran- and 80 C (7–9). At these temperatures, and under aqueous
iol, and not the isomeric epoxide of nerol. This means that conditions (e.g., with H2O2), leaching is more probable than
there is no rotation around the double bond during the under the mild conditions of our tests.
epoxidation.
It is striking that the only material which is really het-
Even homoallylic alcohols (C == C–C–COH) are success- erogeneous contains polynuclear V. In leaching materials
fully epoxidized with high selectivity, albeit that the yield such as VAPO-5 or V-SiO2, the V can be assumed to be
may be somewhat lower. This is illustrated by the reactions monomerically dispersed. In a decavanadate pillar, each V
with isopulegol and nopol (entries 6 and 7). If the distance atom is linked via at least five V–O–V bonds to the rest
between the alcohol group and the double bond is too large, of the pillar. Apparently, this fivefold bonding sufficiently
coordination of the alcohol does not accelerate the epoxi- withstands the conditions of the epoxidation reactions, so
dation, and a rather low conversion and epoxide selectivity that all V is effectively withheld within the structure. A
are obtained, as is observed with �-citronellol (entry 8).
somewhat similar situation is generally accepted to occur
While many allylic alcohols are smoothly epoxidized, a with the Ti catalyst TS-1: of the four possible Ti–O–Si links,
notable exception is verbenol (entry 9). A plausible reason at least two stay intact during catalytic oxidation reactions
is that coordination of the alcohol to the V is hampered by with H2O2 (1). At contrast, in a material such as V-SiO2, the
the bulky carbon skeleton; note that the alcohol group in binding of the V to the support may be rather loose, and
verbenol is on the same side of the six-membered ring as this may facilitate the release of active V species.

2
30
VILLA ET AL.
TABLE 4
Epoxidation of Terpenic and Nonterpenic (Homo)Allylic Alcohols with V10-LDHa
Conversion,
%
Main epoxide
product
Epoxide selectivity,
%
Selectivity, % ,
aldehyde or ketone
Entry
1
Substrate
Geraniol
t, h
4
95
98
97b
97b
1 (geranial)
2
Nerol
24
2 (neral)
3
4
Perillyl alcohol
Myrtenol
14
14
100
97
99c
91d
1
1 (myrtenal)
5
6
trans-Pinocarveol
10
14
91
81
87d
94e
1
4
Isopulegol
7
8
Nopol
48
24
61
25
90d
55e
2
1
�-Citronellol
9
Verbenol
24
14
10
65 (verbenone)
1
0
1
Crotyl alcohol
14
24
65
85
100f
100
—
—
1
Cinnamyl alcohol
1
2
3
2-Cyclohexen-1-ol
Carveol
24
24
28
64
75
22 (enone)
1
88g
12 (carvone)
a
b
c
Procedure: 1 mmol of olefin, 1.1 mmol, 5–6 M, of tBuOOH in decane, 12 mg of catalyst, 2 ml of toluene, 293 K.
2, 3) : (6, 7) epoxide > 50. Two enantiomeric epoxides (50 : 50).
Only the double bond in the ring is epoxidized. The diastereomeric (1, 2) epoxides are formed in a 70 : 30 ratio, see Refs. (31, 32).
More than 95% selectivity for the isomer shown.
(
d
e
f
Two diastereomeric epoxides (� 50 : 50).
Trans : cis = 28 : 1. Commercial crotyl alcohol contains 4–5% cis isomer.
g
Commercial carveol is a mixture of the trans and cis isomers (Ref. 33). The trans-carveol reacts much faster than cis-carveol.

EPOXIDATION OF ALLYLIC ALCOHOLS WITH V-LDH
231
The intercalation of decavanadate in the LDH materials
provides a unique opportunity to study the catalytic proper-
ties of these compounds under anhydrous conditions, with
tBuOOH as the oxidant. When homogeneous decavana-
date is dissolved as a quaternary ammonium salt in toluene,
it displays a very low activity, which may tentatively be as-
cribed to a too strong wrapping of the decavanadate by
the organic cations. Consequently, the peroxide and the
substrate have no access to the decavanadate; the reaction
only starts after an induction period, when the UV–visible
spectrum evidences disappearance of the decavanadate. A
similar situation has been observed with P-W Keggin het-
eropolyanions, which are degraded by a peroxide before
they display an appreciable epoxidation activity (30). This
contrasts with the behavior of decavanadate, exchanged on
an LDH or in an organic resin. Particularly in the LDH, the
decavanadate seems well accessible, and the epoxidation
starts immediately, without a noticeable induction period.
The results of the epoxidation of a series of allylic alco-
hols (Table 4) strongly evidence that the alcohol group in
3. Mimoun, H., Saussine, L., Daire, E., Postel, M., Fischer, J., and Weiss,
R., J. Am. Chem. Soc. 105, 3101 (1983).
4
5
. Mimoun, H., Mignard, M., Brechot, P., and Saussine, L., J. Am. Chem.
Soc. 108, 3711 (1986).
. Sheldon, R. A., and Kochi, J. K., “Metal-Catalyzed Oxidations of Or-
ganic Compounds.” Academic Press, San Diego, 1981.
6. Sharpless, K. B., and Verhoeven, T. R., Aldrichimica Acta 12, 63 (1979),
and references therein.
7
8
. Whittington, B. I., and Anderson, J. R., J. Phys. Chem. 97, 1032
1993).
. Haanepen, M. J., Elemans-Mehring, A. M., and van Hooff, J. H. C.,
(
Appl. Catal. A 152, 203 (1997).
9. Reddy, J. S., Liu, P., and Sayari, A., Appl. Catal. A 148, 7 (1996).
0. Choudary, B. M., Valli, V. L. K., and Durga Prasad, A., J. Chem. Soc.,
Chem. Commun. 721 (1990).
1
1
1
1. Tatsumi, T., Yamamoto, K., Tajima, H., and Tominaga, H., Chem. Lett.
815 (1992).
2. Sels, B. F., De Vos, D. E., and Jacobs, P. A., Tetrahedron Lett. 47, 8557
(1996).
1
1
1
3. Gardner, E. A., and Pinnavaia, T. J., Appl. Catal. A 167, 65 (1998).
4. Fraile, J. M., Garcia, J. I., and Mayoral, J. A., Catal. Today 57, 3 (2000).
5. Trifiro, F., and Vaccari, A., in “Comprehensive Supramolecular Chem-
istry” (G. Alberti and T. Bein, Eds.), Vol. VII, p. 251. Pergamon,
Elmsford, NY, 1996.
the substrate molecule coordinates on the peroxo-V cen- 16. Sels, B., De Vos, D., Buntinx, M., Pierard, F., Kirsch-De Mesmaeker,
A., and Jacobs, P. A., Nature 400, 855 (1999).
ter. This implies that the active V species participates in
1
7. van Laar, F., De Vos, D., Vanoppen, D., Sels, B., Jacobs, P. A., Del
Guerzo, A., Pierard, F., and Kirsch-De Mesmaeker, A., Chem. Com-
mun. 267 (1998).
the reaction with two coordination sites, one for the per-
oxide activation, and another for the alcohol coordination.
Since a V atom is initially bound to the pillar with at least
1
8. Drezdzon, M. A., Inorg. Chem. 27, 4628 (1988).
five V–O–V bonds, one or two of these bonds must be bro- 19. Ulibarri, M., Labajos, F., Rives, V., Trujillano, R., Kagunya, W., and
Jones, W., Inorg. Chem. 33, 2592 (1994).
0. Lopez-Salinas, E., and Ono, Y., Bull. Chem. Soc. Jpn. 65, 2465
ken to enable reaction. Nevertheless, this leaves a sufficient
2
number of V–O–V bonds intact to keep all V within the
polyanions, as is firmly substantiated by the heterogeneity
tests.
Summarizing, we have proven that under well-defined
conditions, decavanadate-pillared LDH is a useful hetero-
geneous catalyst for selective epoxidation of terpenic allylic
alcohols.
(
1992).
2
2
1. Twu, J., and Dutta, P. K., J. Catal. 124, 503 (1990).
2. Kwon, T., Tsigdinos, G. A., and Pinnavaia, T. J., J. Am. Chem. Soc. 110,
3653 (1988).
3. Sels, B., De Vos, D., Pierard, F., Kirsch-De Mesmaeker, A., and Jacobs,
P. A., J. Phys. Chem. B 103, 11114 (1999).
2
2
4. Montes, C., Davis, M. E., Murray, B., and Narayana, M., J. Phys. Chem.
94, 6431 (1990).
2
2
2
2
2
3
5. Linden, G. L., and Farona, M. F., J. Catal. 48, 284 (1977).
6. Corigliano, F., and Di Pasquale, S., Inorg. Chim. Acta 12, 99 (1975).
7. Griffith, W. P., and Lesniak, P. J. B., J. Chem. Soc. (A) 1066 (1969).
8. Twu, J., and Dutta, P. K., J. Catal. 124, 503 (1990).
9. Twu, J., and Dutta, P. K., J. Phys. Chem. 93, 7863 (1989).
0. Aubry, C., Chottard, G., Platzer, N., Br e´ geault, J. M., Thouvenot, R.,
Chauveau, F., Huet, C., and Ledon, H., Inorg. Chem. 30, 4409 (1991).
1. Mordini, A., Pecchi, S., Capozzi, G., Capperucci, A., Degl’Innocenti,
A., Reginato, G., and Ricci, A., J. Org. Chem. 59, 4784 (1994).
2. Brito Baptistella, L., and Mendes Aleixo, A., Liebigs Ann. Chem. 785
(1994).
ACKNOWLEDGMENTS
We are grateful to Colciencias (A.L.V.), IWT (B.F.S.) and FWO
D.E.D.V.) for fellowships. This work is sponsored by the Belgian Fed-
eral Government in an IUAP program, Supramolecular Catalysis.
(
3
3
REFERENCES
1
2
. Bellussi, G., and Rigutto, M. S., Stud. Surf. Sci. Catal. 85, 177 (1994).
. Villa de P., A. L., Sels, B., De Vos, D., and Jacobs, P. A., J. Org. Chem.
33. Arends, I. W. C. E., Pellizon Birelli, M., and Sheldon, R. A., Stud. Surf.
Sci. Catal. 110, 1031 (1997).
6
4, 7267 (1999).