Towards one-pot synthesis of menthols from citral: Modifying Supported Ionic Liquid Catalysts (SILCAs) with Lewis and Bronsted acids

Article abstract of DOI:10.1016/j.jcat.2009.02.005

The use of ionic liquids in catalysis is attracting ever more attention in chemical engineering. In line with this research we have studied Supported Ionic Liquid Catalysts (SILCAs) which consist of immobilized catalytic species, e.g. transition metal par

Full text of DOI:10.1016/j.jcat.2009.02.005

Journal of Catalysis 263 (2009) 209–219
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Towards one-pot synthesis of menthols from citral: Modifying Supported Ionic
Liquid Catalysts (SILCAs) with Lewis and Brønsted acids
Pasi Virtanen a,∗, Hannu Karhu b, Geza Toth c, Krisztian Kordas c, Jyri-Pekka Mikkola a,d
a
Laboratory of Industrial Chemistry and Reaction Engineering, Process Chemistry Centre, Åbo Akademi University, Piispankatu 8, FIN-20500, Turku/Åbo, Finland
Department of Applied Physics, University of Turku, FIN-20014, Turku, Finland
Microelectronics and Materials Physics Laboratories, University of Oulu, FIN-90570, Oulu, Finland
Technical Chemistry, Department of Chemistry, Umeå University, SE-90187, Umeå, Sweden
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
The use of ionic liquids in catalysis is attracting ever more attention in chemical engineering. In line with
this research we have studied Supported Ionic Liquid Catalysts (SILCAs) which consist of immobilized
catalytic species, e.g. transition metal particles residing in an ionic liquid layer immobilized on a solid
support. We found out that dissolution of Lewis or Brønsted acids into the ionic liquid layer of SILCA has
major effects to the activity as well as the product distribution of the catalyst in citral transformation
reactions. In fact, one-pot synthesis of menthols from citral was accomplished. The effects of different
amounts of added Lewis and/or Brønsted acids to SILCA (one of them or both) were studied in this work.
© 2009 Elsevier Inc. All rights reserved.
Received 18 November 2008
Revised 9 February 2009
Accepted 9 February 2009
Available online 10 March 2009
Keywords:
Catalysis
Catalyst modification
Ionic liquids
Multiphase reactions
One-pot synthesis
Supported Ionic Liquid Catalyst (SILCA)
1. Introduction
these features are not common to every single ionic liquid, as has
recently been stated by Deetlefs and Seddon [8]. Even if ionic liq-
Research in the field of ionic liquids is continuously attract-
ing more interest in scientific world. Many new research areas
and improved techniques were introduced at the 2nd International
Congress on Ionic Liquids in July 2007 in Yokohama, Japan, in-
cluding applications in: pharmaceutics, biotechnology, solar-cells,
cellulose processing and catalysis [1]. Indeed, catalysis is one of
the areas were ionic liquids have already shown their capability in
enhancing the reaction rates of many reactions [2–4]. Ionic liquids
are completely ionic compounds, meaning that they are formed of
ionic species only. Typically ionic liquids are formulated as a com-
bination of a large organic cation with the possibility of charge
delocalization and, consequently, a relatively large inorganic or or-
ganic anion. They have many exceptional properties compared to
conventional ionic compounds. The main difference is their sig-
nificantly lower melting point. For example, sodium chloride has a
uids have shown their ability in various chemical applications [1],
widespread use of ionic liquids in industrial applications is still
hampered by the cost aspects, limited knowledge of their physical
and chemical properties as well as concerns about their biodegrad-
ability. Thus, utilizing special characteristics of ionic liquids by
immobilizing small amounts of them on solid surfaces is an al-
ternative worth consideration [9]. In fact, catalysis is one of the
areas where this concept has already been widely utilized [10–15].
Immobilization or supporting of ionic liquids can be achieved
by several different techniques, such as simple impregnation, graft-
ing, polymerization, sol–gel method, encapsulation or pore trap-
ping [9–12,16,17]. A straightforward preparation method involves
impregnation of the support material with an ionic liquid, diluted
with a molecular solvent (e.g., acetone). The dilution followed by
the evaporation of the co-solvent results in a uniform and thin
ionic liquid layer on the pore structure of the support material.
When SILCA catalysts prepared by this manner are applied in a
liquid-phase process, a bulk solvent that is not miscible with the
ionic liquid should be selected. In parallel to the ionic liquid im-
mobilization, transition metal ions or complexes and other mod-
ifier compounds can be dissolved into the ionic liquid layer. The
organometallic species can further be reduced to metallic nano-
particles. A catalyst which contains palladium nanoparticles, to-
gether with an ionic modifier (i.e. Lewis acid) in an ionic liquid
◦
melting point of 801 C, whereas many ionic liquids have a melting
point below room temperature. Other special characteristic fea-
tures that can be attributed to most of them are: often negligible
−8
vapor pressure (∼10
bar), they are liquid over wide temperature
range, unique solvation properties, wide electrochemical window
and good ion conductivity [5–7]. Still, one should remember that
Corresponding author. Fax: +358 2 2154479.
E-mail address: pasi.virtanen@abo.fi (P. Virtanen).
*
0021-9517/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.02.005

210
P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
2.2. Testing of activity and selectivity of the catalysts
The catalysts were applied in the transformation of citral (Lan-
caster, 95%). Experiments were performed in a semi-batch reactor
(Parr Instrument company, total volume 600 ml, liquid volume
250 ml), equipped with heating jacket and a tailor-made stir-
rer/catalyst holder. Suitable pieces of the catalyst carrier ACC were
attached to the stirrer/catalyst holder. The temperature, pressure
and stirring rate were controlled by a Parr 4843 control unit (Wat-
low control series 982). The stirring rate was adjusted accordingly
(1400 rpm) so that external mass-transfer limitations were elim-
inated. That was confirmed in our previous studies [21]. All ex-
Fig. 1. A supported ionic liquid catalyst used in citral transformation.
◦
periments were performed at a constant temperature (80–120 C)
and at a constant partial pressure of hydrogen (AGA, 99.9999%, 5–
10 bar). Citral (approx. 3 g) was dissolved in 250 ml of n-hexane
(Merck, pro analysis) and the solution was bubbled with hydrogen,
in order to ensure that the solution was oxygen-free. Before each
and every experiment the reactor was pre-heated to the desired
temperature and hydrogen pressure was adjusted to the selected
level. The citral solution was introduced to the reactor and stirring
was engaged.
layer which, in turn, is immobilized on an active carbon cloth
(ACC), is presented in Fig. 1.
Selective hydrogenation of α,β-unsaturated aldehydes, ketones
and esters, in general, is a versatile method to obtain many inter-
esting products that find use in the perfumery industry, hardening
of fats, preparation of pharmaceuticals and synthesis of organic
chemical intermediates. Citral itself and its hydrogenation products
are widely used in the perfumery and fine chemical industry. Se-
lective hydrogenations of α,β-unsaturated aldehydes and ketones
are challenging, because these species can contain three differ-
ent double bonds: isolated and conjugated carbon–carbon double
bonds as well as a carbonyl group. Consequently, during hydro-
genation of citral, many competing and consecutive reactions can
take place, including formation of ring compounds such as pulegols
and menthols. Most of the menthol used worldwide is obtained by
freezing peppermint and cornmint oils, but it is also produced syn-
thetically [18]. Menthols from citral is an attractive synthetic route
because citral comes from renewable feedstock and is mainly ob-
tained by distillation of essential oils. According to previous stud-
ies, the best catalyst so far was a Ni supported on acidic zeolite
H-MCM-41 [19,20]. For this reason a plausible assumption was
that an acidic function incorporated into a Supported Ionic Liq-
uid Catalyst (SILCA), previously used on citral hydrogenation [21],
represented a worthwhile alternative. A lot of research in selec-
tive hydrogenation of α,β-unsaturated aldehydes has been carried
out with conventional heterogeneous and homogeneous catalysts,
in conventional as well as supercritical solvents and ionic liquids
[22–27]. The potential of supported ionic liquid catalysts has al-
ready been illustrated in e.g. hydrogenation of alkenes, displaying
competitive performance in comparison to biphasic systems and
conventional solvents [14,28,29].
The progress of the reaction was monitored by withdrawing
small amounts of liquid samples from the reactor analyzed by gas
chromatography (Hewlett Packard 6890 GC with FI detector). Since
the sampling volumes were small, compared to the overall vol-
ume of the reaction mixture, the volume changes in time were
neglected. In addition, a gas chromatograph coupled to a mass
spectrometer (Agilent 6890N GC with Agilent 5973 MS detector)
was applied for identification of peaks. In both cases, 500 μl of in-
ternal standard (0.02 M cyclohexanone in cyclohexane) was added
into a 500 μl of sample. The column in the system was an Agilent
DB-1 with a length of 30 m, inner diameter of 0.25 mm and, the
film thickness of 0.5 μm. The samples were analyzed with the fol-
lowing temperature program: temperature was at first held 10 min
◦
◦
◦
at 100 C, then raised 5 C/min to 160 C, and consequently held
◦
◦
10 min at 160 C, followed by a temperature ramp 13 C/min to
200 C, and was held 1 min at 200 C.
◦
◦
2.3. Catalyst characterization
The catalysts were investigated by means of X-ray photoelec-
tron spectroscopy (XPS) in order to determine the binding en-
ergies of all compounds and oxidation states of Pd on the ACC
support. A Perkin–Elmer PHI 5400 spectrometer was used for the
X-ray photoelectron analysis with MgKα radiation (1253.6 eV) and
pass energy of 35.75 eV. Pressure during the analysis was ca. 8 ×
2. Experimental
−9
10
mbar. Reduced samples were stored in a nitrogen atmo-
sphere after pre-treatment process. However, during transfer to
the XPS instrument the samples were in contact with ambient air
and they were out-gassed overnight. Depending on the vapor pres-
sure and the condition of the catalyst, the ionic liquid partially
evaporated during out-gassing due to ultra high vacuum condi-
tions. Hence, boron, fluorine and chlorine, which have relatively
low photoionization cross-sections, were in some conditions below
detection limit. The observations are largely based on residual ionic
liquid in the pores of the active carbon cloth. Binding energies
(BEs) were determined by line fitting procedure, where the line
shape of the fitted peaks was a convolution of a Doniach–Šunjic´
line shape with a Gaussian line shape. A linear background was
used for signal background subtraction. Accuracy of the binding
energies is 0.1 eV. In the quantitative analysis, sensitivity factors
for B 1s, Cl 2p, F 1s and Zn 2p3/2 were 0.159, 0.891, 1.000 and
3.726 [31]. The catalyst reduction was carried out ex-situ, under a
2.1. Catalyst preparation
Catalysts were prepared according to a similar straight-forward
preparation method introduced by us previously [30]. Palladium
acetylacetonate (Pd(acac)2) (approx. 50 mg) (Aldrich 99%), a Lewis
acid zinc chloride (ZnCl2) or ferrous chloride (FeCl2) and/or a
Brønsted acid tetrafluoroboric acid HBF4 or acetic acid, as well
as an ionic liquid N-butyl-4-methylpyridinium tetrafluoroborate
(NB4MPyBF4) (approx. 150 mg) (Merck, 98%), when applicable,
were dissolved in acetone (Merck, p.a.). Solution was poured over
ACC Kynol® (approx. 900 mg). ACC was dried a priori at 60 C over
◦
12 h and cut into suitable pieces. Acetone was evaporated in rotary
◦
evaporator at 50 C in vacuum of approx. 300 mbar and completely
dried in vacuum of approx. 4 mbar. Catalysts were pre-treated in a
high-pressure autoclave (Parr Inc.) at 120 C under hydrogen flow
of 10 bar. As a result, a catalyst containing palladium nanoparticles
and a Lewis and/or a Brønsted acid in an ionic liquid immobilized
on ACC was achieved.
◦
◦
hydrogen flow at 120 C for 60 min prior to analysis.
Leaching of ionic liquid was studied by High Performance Liq-
uid Chromatography (Hewlett Packard 1100 series HPLC). Reversed-

P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
211
Scheme 1. Reaction sequence for citral transformation.
phase HPLC with diode array detector (DAD) was utilized in
3. Results and discussion
the following conditions: sample injection volume 5 μl, eluent:
30% acetonitrile, 70% 0.02 M phosphate buffer pH 5, flow rate
1.0 ml/min, column: Hewlett Packard, Zorbax eclipse XDB-C8: in-
ternal diameter of 4.6 mm, length 150 mm, column tempera-
3.1. Activity and selectivity of the catalysts
Citral transformation reactions were performed with thir-
teen different SILCA-catalysts containing approximately the same
amount of palladium, ionic liquid N-butyl-4-methylpyridinium
tetrafluoroborate and support material active carbon cloth. Only
the amount of Lewis acid zinc chloride and Brønsted acid tetraflu-
oroboric acid were altered in catalysts which they were used.
During the hydrogenation of citral in the presence of acidic mod-
ifier, a number of competing and consecutive reactions can occur.
The reaction network for the main transformations is presented in
Scheme 1. Other minor products formed and identified during the
experiments were different alkanes and alkenes formed by dehy-
dration and cracking.
In our earlier studies we found out that the most active catalyst
of SILCAs containing palladium and different ionic liquids was the
one containing N-butyl-4-methylpyridinium tetrafluoroborate [21,
32]. Although, we tested no more than five different ionic liquids,
the catalyst containing [NB4MPy][BF4] was by far the most active
one and therefore was chosen for further development. We assume
that the reasons for the differences in activity could be explained
by differences in solubility of reactants and products between dif-
ferent ionic liquids and hexane. Thus, reactants are more soluble
in [NB4MPy][BF4] than other ionic liquids and main product di-
hydrocitronellal is less soluble in [NB4MPy][BF4] than other ionic
liquids.
◦
ture of 40 C, detection with DAD: qualitatively from UV spectrum
and quantitatively with wavelength of 220 nm. The ionic liquid
standards were prepared and analyzed with HPLC and calibration
curves were drawn according to these results. Liquid phase sam-
ples were analyzed and the amount of ionic liquid was determined
by using the linear part of the calibration curve.
Leaching of palladium from the catalyst was determined by
means of inductive coupled plasma emission spectrometer coupled
to a mass spectrometer (ICP–MS, ELAN 6000). The samples were
taken at the end of the experiment. Solvent was first evaporated
and the remaining solid was dissolved in Aqua Regia (1:3, con-
centrated nitric acid and hydrochloric acid). These samples were
diluted with water prior to analysis. The amount of leached pal-
ladium from the catalyst was calculated according to these re-
sults.
The surface area and the micropore volume of the catalysts
were determined by an automatic physisorption–chemisorption
apparatus (Carlo-Erba Instruments, sorptometer 1900). The Dubinin
method was used to calculate the surface area of the catalysts.
Dollim–Heal method was used to calculate the micropore volume
of the catalysts. The catalysts were analyzed as fresh and after be-
ing used in hydrogenation experiment.
Selected catalysts were also investigated by means field emis-
sion scanning electron microscopy (FESEM Jeol JSM-6300F) and
energy-filtered transmission electron microscopy (EFTEM LEO 912
OMEGA), equipped with an energy dispersive X-ray detector. For
the EFTEM analyses, the samples were crushed and dispersed ei-
ther in n-hexane or in water before dispensing them on the speci-
men holder copper grid (formvar/carbon).
3.1.1. The effects of different acids
Since the main products for the catalysts containing both pal-
ladium and an ionic liquid only were citronellal and 3,7-dimethyl-
octanal, we decided to study catalyst modification by added modi-
fier. At first we tested how the addition of different acids in SILCA
and in reference catalyst without ionic liquid affects the activity

212
P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
Fig. 2. Comparison of the initial reaction rates with catalysts containing different acid modifiers with and without ionic liquid, calculated from citral conversion at time
◦
15 min. The reaction conditions were: T = 100 C, p(H2) = 10 bar. The initial reaction rates were calculated as follows: (moles of citral converted to products)/(time × mass
of Pd).
Fig. 3. Comparison of selectivities of products from citral reaction over catalysts containing different acid modifiers with and without ionic liquid in hydrogen atmosphere
◦
after total conversion (approximately after 6 h, 1 h, 24 h, 24 h, 6 h, 6 h, 4 h, 4 h, respectively with the catalysts from left to right). The reaction conditions were: T = 100 C,
p(H2) = 10 bar. Other compounds are reaction products from dehydration and cracking of citral and its hydrogenation products.
of the catalyst and the catalyst selectivity. In Fig. 2 the activity of
various catalysts are compared by means of calculating the initial
reaction rates.
It can be seen that the catalyst containing palladium in ionic
liquid on ACC is the most active one and the catalysts containing
ionic liquid and Lewis acid are the worst when activity of the cat-
alyst is evaluated. However, when the catalyst selectivity is com-
pared the catalyst containing stronger Lewis acid, zinc chloride, in
ionic liquid shows exceptional selectivity profile (Fig. 3). With this
catalyst the selectivity of menthols was over ten times higher than

P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
213
Scheme 2. Suggested reaction mechanism for the citronellal transformation to isopulegols in the presence of a Lewis acid.
ring formation reaction as presented in Scheme 2. It is also evident
that ionic liquid must be present with acids in order to affect the
selectivity. We assume that the ionic liquid also enables the forma-
tion of ring compounds by stabilizing the ionic forms in Scheme 2.
3.1.2. The effect of a Lewis acid
With the promising results from the experiments with catalysts
containing different acids in catalysts, we decided to test how dif-
ferent amounts of a Lewis acid, ZnCl2 in SILCA, affects the activity
and selectivity profiles in citral hydrogenation. In Fig. 4 the activ-
ity of the various catalysts are compared by means of calculating
the initial reaction rates.
The catalysts in all experiments were otherwise similar, but
they contained different amounts of Lewis acid ZnCl2. When look-
ing at Fig. 2 it can be clearly seen that the activities of the SILCAs
containing ZnCl2 are much lower than the one containing Pd and
ionic liquid only. However, the specific features of these catalysts
become visible when selectivities of different products are com-
pared (Fig. 5).
Fig. 4. Comparison of the initial reaction rates with different catalysts, calculated
◦
from citral conversion at time 15 min. The reaction conditions were: T = 100 C,
p(H2) = 10 bar. The initial reaction rates were calculated as follows: (moles of citral
converted to products)/(time × mass of Pd).
3.1.3. The effect of a Brønsted acid
with the catalyst without any acid and over seven times higher
than with the catalyst containing acid without ionic liquid. Also,
a reasonable selectivity towards menthols was achieved with the
catalyst containing strong Brønsted acid in ionic liquid.
It seems evident that reactions over the catalyst containing
strong acids ZnCl2, or HBF4 in ionic liquid produced more ring
compounds, isopulegols and menthols. This is probably caused by
the electron withdrawing capability of strong acids, which helps
the oxygen to maintain the negative charge and thus enables the
In the next step we evaluated the effect of a Brønsted acid,
tetrafluoroboric acid (HBF4) in SILCA. HBF4 was chosen because
it has the same anion (BF4 ) as the ionic liquid used. The ef-
fect of HBF4 in total was not as significant as was the effect of
ZnCl2. Again, we tested the catalysts containing different amounts
of HBF4. The amount of HBF4 is not absolutely certain, since some
acid evaporates during catalyst preparation procedure. The same
vacuum treatment that was used in catalyst preparation was ap-
plied to the 1:1 mixture of acid in ionic liquid, in order to estimate
−
Fig. 5. Comparison of selectivities of products from citral reaction over Lewis acidic SILCAs in hydrogen atmosphere after total conversion (approximately after 24 h, with the
◦
catalyst without acid after 1 h). The reaction conditions were: T = 100 C, p(H2) = 10 bar. Other compounds are reaction products from dehydration and cracking of citral
and its hydrogenation products.

214
P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
were tested. Both catalysts contained the same amount of HBF4
(1:2 molar amounts compared to ionic liquid) and 1:1 as well as
1:4 molar amounts of ZnCl2 in relation to the ionic liquid amount.
The results were similar as with the catalyst containing a Brønsted
acid only (Fig. 8). However, the activity of the catalyst contain-
ing 1:2 molar amount of HBF4 and 1:4 molar amount of ZnCl2
was much higher than that of the catalysts containing the same
amount of only one of the acids. With the other catalyst contain-
ing the same amount of HBF4 and 1:1 molar amount of ZnCl2,
the activity was approximately the same as the catalyst containing
HBF4 only, but about twice as high than that of the catalyst con-
taining ZnCl2 alone. The selectivities of these two catalysts were in
the same level as the catalysts containing HBF4 only. Therefore, we
decided not to continue any further investigations in this line of
research.
Fig. 6. Comparison of the initial reaction rates with different catalysts, calculated
3.1.5. The effects of temperature and hydrogen pressure
◦
from citral conversion at time 15 min. The reaction conditions were: T = 100 C,
As the next step, we decided to study a catalyst that contained
1:2 molar ratio of ZnCl2 and an ionic liquid NB4MPyBF4 together
with Pd on ACC more thoroughly, since both the activity and selec-
tivity towards menthols were at a reasonable level with that cat-
alyst. The catalyst was investigated at two different temperatures
and two different partial pressures of hydrogen. Fig. 9 clearly illus-
trates that reasonable variations in temperature and pressure do
not have a large effect to the activity of the catalyst. However, the
selectivities towards different products were more significantly af-
fected as illustrated in Fig. 10. When the pressure of hydrogen was
lowered to 5 bar, more ring compounds were obtained. The results
were understandable, since lowering the pressure of H2 obviously
lowers the production rate of further hydrogenated products, di-
hydrocitronellal and 3,7-dimethyloctanol, thus giving possibility to
form ring compounds from citronellal. Also, a notable result is that
with lower pressure of H2 we did not produce almost any 3,7-di-
methyl-2-octenal which again means that we should get more cit-
ronellal and ring compounds in accordance with Scheme 1. We
previously obtained similar results with catalysts in the absence
of acidic compounds, meaning that the formation of citronellal
was favored by lower hydrogen pressure. Obviously, we were not
p(H2) = 10 bar. The initial reaction rates were calculated as follows: (moles of citral
converted to products)/(time × mass of Pd).
the amount of acid that evaporates. The experiment showed that
over 90% of the acid remained in the ionic liquid during vacuum
treatment. The amount of ionic liquid and Pd was approximately
the same in all of the catalysts. In Fig. 6 the activity of the cata-
lysts containing different amounts of HBF4 are compared.
On the basis of these results, it is clear that an addition of HBF4
to the SILCA lowers the activity of the catalysts, but the amount of
HBF4 does not seem to have any effect at all. When comparing the
selectivities of different products in Fig. 7, we can clearly see that
the selectivity towards menthols and dihydrocitronellal are some-
what higher with the catalysts containing larger amounts of HBF4.
However, the effect of an added Brønsted acid was significantly
less than the effect of an added Lewis acid.
3.1.4. Combined Brønsted and Lewis acids
In the following phase we decided to study the effect of addi-
tion of both a Brønsted and a Lewis acid. Two different catalysts
Fig. 7. Comparison of selectivities of products from citral reaction over Brønsted acidic SILCAs in hydrogen atmosphere after total conversion (approximately after 6 h, except
◦
with the catalyst without acid after 1 h). The reaction conditions were: T = 100 C, p(H2) = 10 bar. Other compounds are reaction products from dehydration and cracking
of citral and its hydrogenation products.

P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
215
Fig. 8. Comparison of: selectivities of products from citral reaction over SILCA containing both a Brønsted and a Lewis acids in hydrogen atmosphere after total conversion (a
and b) (approximately after 6 h) and the initial reaction rates over same catalysts, calculated from citral conversion at time 15 min (c and d). The reaction conditions were:
◦
T = 100 C, p(H2) = 10 bar. The initial reaction rates were calculated as follows: (moles of citral converted to products)/(time × mass of Pd). Other compounds are reaction
products from dehydration and cracking of citral and its hydrogenation products.
able to produce more ring compounds, since the acidic function
was missing in those catalysts [21,29,30]. In Fig. 11 the mole frac-
tions of citral and its reaction products are presented as a function
of time as well as selectivities of reaction products as a func-
tion of conversion from the experiment at hydrogen pressure of
◦
5 bar and at temperature of 100 C. This experiment was run over
48 h in order to estimate the whether the product distribution
changes with time. After total conversion was achieved, at 24 h
the product distribution changes only slightly. Some isopulegols
which are still present react to menthols and dihydrocitronellal
reacts slowly to 3,7-dimethyl-1-octanol and decomposition prod-
ucts.
3.1.6. Catalyst deactivation
Fig. 9. Comparison of the initial reaction rates over catalyst Pd in ZnCl2/NB4MPyBF4
The same catalyst system (1:2 molar amounts of ZnCl2 and
ionic liquid NB4MPyBF4 together with Pd on ACC), that was used
when temperature and pressure effects was demonstrated was
(1:2) on ACC, calculated from citral conversion at time 15 min. The reaction con-
ditions are displayed in the figure. The initial reaction rates were calculated as
follows: (moles of citral converted to products)/(time × mass of Pd).
Fig. 10. Comparison of selectivities of products from citral reaction over catalysts Pd in ZnCl2/NB4MPyBF4 (1:2) on ACC in hydrogen atmosphere after total conversion
(approximately after 24 h). The reaction conditions are displayed in the figure. Other compounds are reaction products from dehydration and cracking of citral and its
hydrogenation products.

216
P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
Fig. 11. Mole fractions of citral and its reaction products as a function of time (upper graph) and selectivities of reaction products as a function of conversion (lower graph)
during a batch run. The catalyst applied consisted of Pd in a mixture of 1:2 molar amounts of ZnCl2 and ionic liquid NB4MPyBF4 on ACC. The reaction conditions were:
◦
T = 100 C, p(H2) = 5 bar. Other compounds are different reaction products of dehydration and cracking.
chosen for catalyst deactivation testing as well. The catalyst was re-
used in four consecutive batches for transformation of citral. The
temperature and hydrogen partial pressure was kept in all batches
3.2. Catalyst characterization
The results from the XPS measurements for selected catalysts
are listed in Table 1. Palladium 3d5/2 binding energy (BE) in the
studied samples was ca. 335.5 eV, which is assigned to Pd in
Pd(0) state. It is chemically feasible to expect that zinc chloride
is present in ionized form in the ionic liquid. The presence of
Zn(0) should also be considered, since the catalyst samples were
reduced under flowing hydrogen prior to analysis and/or used in
reductive operating conditions, possibly resulting in the reduction
of zinc. This is observed in the Zn/Cl ratios that varied from 0.39
to 4.0. In the latter case, it is evident that most of the zinc is
not associated with chlorine. Metallic zinc has a binding energy
of 1021.8 eV, which was not observed in the measured ionic liq-
uid catalysts. Since the samples were in contact with air, zinc has
◦
at 100 C and 5 bar, respectively. Between the experiments the
catalyst was stored in the reactor in hydrogen atmosphere. Citral
conversion as a function of time in these four consecutive batches
is presented in Fig. 12. It can be seen that the catalyst is deac-
tivated during the first experiment. However, during consecutive
batches the deactivation is very minimal. The cause for deactiva-
tion is probably similar as has been reported previously in the case
of hydroformylation and citral hydrogenation with similar types of
catalysts [21,33].
The product distribution changed little in consecutive batches.
Selectivities of dihydrocitronellal and citronellal increased as selec-
tivities of all other products decreased.

P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
217
Fig. 12. Deactivation of a catalyst containing 1:2 molar amounts of ZnCl2 and ionic liquid NB4MPyBF4 together with Pd on ACC during four consecutive batches. The reaction
◦
conditions were: T = 100 C, p(H2) = 5 bar.
Table 1
Binding energies of atoms on selected SILCAs measured with XPS.
Catalyst
n(ZnCl2):
n(IL)
n(HBF4):
n(IL)
C 1s
(eV)
O 1s
(eV)
F 1s
(eV)
Cl 2p1/2
(eV)
Zn 2p3/2
(eV)
Pd 3d5/2
(eV)
B 1s
(eV)
Zn/Cl
(eV)
B/F
(eV)
Pd/ZnCl2/NB4MPyBF4/ACC (fresh)
1:9
0
284.5
532.1
685.4
197.8
1022.1 (53%)
1023.5 (47%)
1022.7
335.5
BDLa
1.1
BDL
1:1
0
284.6
532.5
685.4
198.0
335.4 (92%)
337.1 (8%)
BDL
1.5
BDL
Pd/ZnCl2/NB4MPyBF4/ACC (spent)
Pd/HBF4/NB4MPyBF4/ACC (fresh)
1:9
1:1
0
0
284.7
284.4
532.5
532.0
685.8
685.1
198.0
198.0
N/Ab
N/A
1022.7
1022.5
335.3
335.4
193.7
BDL
0.39
0.77
0.24
BDL
0
0
1:9
1:1
284.5
284.6
532.3
532.8
685.5
685.8
N/A
N/A
335.4
335.6
BDL
193.7
N/A
N/A
BDL
0.26
Pd/HBF4/NB4MPyBF4/ACC (spent)
0
1:1
1:2
284.6
284.6
532.6
532.1
685.8
685.2
N/A
N/A
335.7
335.4
193.7
N/A
0.31
Pd/ZnCl2/HBF4/NB4MPyBF4/ACC (fresh)
1:4
197.9
1022.6 (78%)
1021.0 (22%)
1022.1 (53%)
1023.5 (47%)
BDL
2.0
BDL
1:1
1:1
1:2
1:2
284.5
284.6
532.1
532.3
685.2
197.8
335.5
335.4
BDL
BDL
1.1
BDL
BDL
Pd/ZnCl2/HBF4/NB4MPyBF4/ACC (spent)
BDL
BDL
1022.3
BDL
a
BDL: below detection limit.
N/A: not applicable.
b
formed hydrates and oxides prior to XPS analysis. In literature,
binding energy values in the range of 1023.1 eV–1023.3 eV have
been reported for Zn 2p3/2 in ZnCl2 [34–36].
Table 2
Results from Pd leaching measurements.
Catalyst
n(ZnCl2):n(IL)
n(HBF4):n(IL)
c, Pd (ppm)a
In the studied catalyst samples, Zn 2p3/2 was observed in vac-
uum conditions at a wide binding energy range, ranging from
1021.0 eV to 1022.7 eV. This is explained by the formation of hy-
droxides and oxides. Chlorine 2p3/2 was observed at ca. 197.9 eV,
which is also lower than the value of Cl in ZnCl2, 199.1 eV [37].
Binding energies that were observed for boron 1s (193.7 eV)
are lower than the binding energies previously observed for
tetrafluoroboron containing compounds (ca. 195 eV) [31]. Instead,
the binding energies of F3 containing compounds correspond well
with the observed binding energy for B 1s. A feasible explana-
tion is that BF3 ligands have bonded with zinc chloride, explaining
the lowest observations of binding energy values (1021.0 eV) in
zinc.
1:9
1:4
1:2
1:1
0
0
0
0
0.89
1.3
1.5
Pd/ZnCl2/
NB4MPyBF4/
ACC
0.94
0
0
0
0
1:9
1:4
1:2
1:1
0.98
1.0
0.93
1.3
Pd/HBF4/
NB4MPyBF4/
ACC
Pd/ZnCl2/HBF4/
NB4MPyBF4/
ACC
1:4
1:1
1:2
1:2
1.3
1.3
a
(c, Pd) measured from the bulk phase after the experiment.
being a small amount. However, it was significantly higher than in
the case of the catalysts in our previous studies, that did not con-
tain any Lewis or Brønsted acids, varying from 25 to 48 ppb [21].
On the basis of nitrogen physisorption measurements, it was
possible to calculate the specific surface area and micropore vol-
ume of the catalysts. The catalysts were analyzed as fresh and
Results from the ionic liquid leaching studies with HPLC indi-
cated that, practically no leaching occurred during experiments.
The results from the palladium leaching measurements with ICP–
MS are listed in Table 2. The amount of leached palladium in the
bulk phase from each experiment was in the range of 1 ppm, this

218
P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
Table 3
Results from the nitrogen physisorption measurements.
Catalyst
n(ZnCl2):n(IL)
n(HBF4):n(IL)
As
mpV
(m2/g)a
(cm3/g)b
1:9
1:4
1:2
1:1
0
0
0
0
15.2
482
20.6
687
0.0054
0.17
0.0073
0.24
Pd/ZnCl2/
NB4MPyBF4/
ACC (fresh)
1:9
1:4
1:2
1:1
0
0
0
0
6.03
9.97
9.85
8.62
0.0021
0.0035
0.0035
0.0031
Pd/ZnCl2/
NB4MPyBF4/
ACC (spent)
0
0
0
0
1:9
1:4
1:2
1:1
182
0.065
0.19
0.17
Pd/HBF4/
NB4MPyBF4/
ACC (fresh)
530
485
1240
0.44
0
0
0
0
1:9
1:4
1:2
1:1
34.2
9.56
24.3
498
0.012
0.0034
0.0086
0.18
Pd/HBF4/
NB4MPyBF4/
ACC (spent)
Fig. 13. FESEM (a, b) and EFTEM (c, d) pictures of catalyst containing Pd in mixture
of 1:1 ZnCl2 and NB4MPyBF4 on ACC as fresh (a, c) and after being applied in citral
transformation.
Pd/ZnCl2/HBF4/
NB4MPyBF4/
ACC (fresh)
1:4
1:1
1:2
1:2
591
1490
0.21
0.53
Pd/ZnCl2/HBF4/
NB4MPyBF4/
ACC (spent)
1:4
1:1
1:2
1:2
168
172
0.060
0.061
a
As = specific surface area, calculated by Dubinin method.
mpV = micropore volume, calculated by Dollim/Heal method.
b
after being applied in citral transformation experiments. The re-
sults from these experiments are listed in Table 3.
Since the specific surface area of pure ACC is 1680 m2/g and
micropore volume is 0.6 cm3/g [21], one can estimate that ap-
proximately 20–70 vol% of the pores were filled with immobilized
species. The accumulation of reaction products is probably the rea-
son for the very low surface area and pore volume for spent cat-
alysts. This kind of accumulation has previously being reported in
hydroformylation and citral hydrogenation reactions with similar
catalysts [21,31].
The catalysts which contained highest concentrations of mod-
ificators were selected to be analyzed with FESEM and EFTEM.
The results proved that palladium nanoparticles were formed on
catalyst during pre-treatment process (Figs. 13–15). These figures
illustrate FESEM and EFTEM pictures of the catalysts containing
palladium in a mixture of 1:1 ZnCl2 and NB4MPyBF4 on ACC, pal-
ladium in a mixture of 1:1 HBF4 and NB4MPyBF4 on ACC and
palladium in a mixture of 1:1 ZnCl2 and NB4MPyBF4 as well as
1:3 HBF4 and NB4MPyBF4 on ACC. The catalysts were analyzed as
fresh and as spent ones.
Fig. 14. FESEM (a, b) and EFTEM (c, d) pictures of catalyst containing Pd in mixture
of 1:1 HBF4 and NB4MPyBF4 on ACC as fresh (a, c) and after being applied in citral
transformation.
FESEM images (a) and (b) reveal the fiber structures of the
catalyst support and also some palladium particles are visible. In
Fig. 13b one can observe that citral transformation products ac-
cumulate on the catalyst surface as were indicated by the results
from nitrogen physisorption measurements where spent catalysts
had lower specific surface areas than fresh catalysts. This is also a
plausible explanation for the catalyst deactivation process. EFTEM
pictures (c, d) show that the catalysts really contain palladium
nanoparticles in ionic liquid immobilized on ACC. Also, some ag-
glomeration of palladium particles can be observed when com-
paring fresh and spent catalysts. This might also contribute to the
deactivation of the catalyst.
4. Conclusions
During this study we successfully managed to demonstrate
that it is possible to modify Supported Ionic Liquid Catalysts by
co-immobilization of acidic modifiers during catalyst preparation.
These modified catalysts demonstrated exceptional selectivities on
Fig. 15. FESEM (a, b) and EFTEM (c, d) pictures of catalyst containing Pd in mixture
of 1:1 ZnCl2 and NB4MPyBF4 as well as 1:3 HBF4 and NB4MPyBF4 on ACC as fresh
(a, c) and after being exposed to the reaction conditions.

P. Virtanen et al. / Journal of Catalysis 263 (2009) 209–219
219
citral transformations when compared to unmodified ones. Selec-
References
tivity towards menthols was raised from ∼1%, gained over SILCA
without acidic modifier, up to ∼41%, gained over Lewis acid mod-
ified SILCA, thus being the main product in the reaction. The
characterization of the catalysts verified that metallic palladium
nanoparticles were formed on the catalysts during pre-treatment
process and leaching of both ionic liquid and palladium were min-
imal. As demonstrated by this study, the supported ionic liquid
catalysts are very interesting class of catalysts, since they can be
easily tuned by different modifiers in order to change the selec-
tivity profile of the reaction. Although, we only presented the acid
modified catalysts we anticipate that this concept can be applied
also in various other reaction. For example, it is plausible that the
catalyst could be modified with chiral modifiers for production of
asymmetric chemicals.
[1] 2nd International Congress on Ionic Liquids, Book of Abstracts.
[2] J. Dupont, G.S. Fonseca, A.P. Umpierre, P.F.P. Fichtner, S.R. Teixeira, J. Am. Chem.
Soc. 124 (2002) 4228.
[3] R. Sheldon, Chem. Commun. (2001) 2399.
[4] C.M. Gordon, Appl. Catal. 222 (2001) 101.
[5] P. Bonhôte, A.-P. Dias, N. Papageorgiou, K. Kalyanasundaram, M. Grätzel, Inorg.
Chem. 35 (1996) 1168.
[6] R.D. Rogers, K.R. Seddon, Science 302 (2003) 792.
[7] K.R. Seddon, Nat. Mater. 2 (2003) 363.
[8] M. Deetlefs, K.R. Seddon, Chim. Oggi (Chem. Today) 24 (2006) 16.
[9] M.H. Valkenberg, C. deCastro, W.F. Hölderich, Green Chem. (2002) 88.
[10] S. Breitenlechner, M. Fleck, T.E. Mueller, A. Suppan, J. Mol. Catal. A 214 (2004)
175.
[11] D.W. Kim, D.Y. Chi, Angew. Chem. Int. Ed. 43 (2004) 483.
[12] A. Riisager, R. Fehrmann, S. Flicker, R. van Hal, M. Haumann, P. Wasserscheid,
Angew. Chem. Int. Ed. 44 (2005) 815.
[13] H. Hagiwara, Y. Sugawara, K. Isobe, T. Hoshi, T. Suzuki, Org. Lett. 6 (2004) 2325.
[14] J. Huang, T. Jiang, H. Gao, B. Han, Z. Liu, W. Wu, Y. Chang, G. Zhao, Angew.
Chem. Int. Ed. 43 (2004) 1397.
Acknowledgments
[15] C.P. Mehnert, Chem. Eur. J. 11 (2005) 50.
The financial support from the Academy of Finland is gratefully
acknowledged (Decisions nos. 124640, 209391 and 211463). This
work is a part of the activities at the Åbo Akademi Process Chem-
istry Centre (PCC) within the Finnish Centre of Excellence Programs
(2000–2005 and 2006–2011) by the Academy of Finland.
[16] F. Shi, Q. Zhang, D. Li, Y. Deng, Chem. Eur. J. 11 (2005) 5279.
[17] R.T. Carlin, H.T. Cho, J. Fuller, Proc. Electrochem. Soc. 11 (1998) 180.
[18] G.S. Clark, Menthol Perfumer Flavorist 25 (1998) 33.
[19] A.F. Trasarti, A.J. Marchi, C.R. Apesteguía, J. Catal. 224 (2004) 484.
[20] P. Mäki-Arvela, N. Kumar, D. Kubicka, A. Nasir, T. Heikkilä, V.-P. Lehto, R. Sjö-
holm, T. Salmi, D.Y. Murzin, J. Mol. Catal. A Chem. 240 (2005) 72.
[21] P. Virtanen, H. Karhu, K. Kordas, J.-P. Mikkola, Chem. Eng. Sci. 62 (2007) 3660.
[22] P. Claus, Top. Catal. 5 (1998) 51.
Appendix A. Abbreviations
[23] G. Neri, L. Mercadante, A. Donato, A.M. Visco, S. Galvagno, Catal. Lett. 29 (1994)
379.
SILCA
ACC
Supported Ionic Liquid Catalyst
Active Carbon Support
[24] P. Gallezot, A. Giroir-Fendler, D. Richard, Catal. Lett. 5 (1990) 169.
[25] F. Joó, J. Kovács, A.C. Bényei, Á. Kathó, Catal. Today 42 (1998) 441.
[26] M. Burgener, R. Furrer, T. Mallat, A. Baiker, Appl. Catal. A 268 (2004) 1.
[27] K. Anderson, P. Goodrich, C. Hardacre, D.W. Rooney, Green Chem. (2003) 448.
[28] C.P. Mehnert, E.J. Mozeleski, R.A. Cook, Chem. Commun. (2002) 3010.
[29] M. Ruta, I. Yuranov, P.J. Dyson, G. Laurenczy, L. Kiwi-Minsker, J. Catal. 247
(2007) 269.
[30] J.-P. Mikkola, P. Virtanen, H. Karhu, T. Salmi, D.Y. Murzin, Green Chem. (2006)
197.
[31] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-Ray Photo-
electron Spectroscopy, Perkin Elmer Corp., Physical Electronics Division, Eden
Prairie, 1992.
[32] P. Virtanen, J.-P. Mikkola, T. Salmi, Ind. Eng. Chem. Res. 46 (2007) 9022.
[33] A. Riisager, R. Fehrmann, M. Haumann, B.S.K. Gorle, P. Wasserscheid, Ind. Eng.
Chem. Res. 44 (2005) 9853.
[34] L. Fiermans, R. Hoogewijs, J. Vennik, Surf. Sci. 47 (1975) 1.
[35] R.F. Mulligan, A.A. Iliadis, P. Kofinas, J. Appl. Polym. Sci. 89 (2003) 1058.
[36] W. Xie, Z. Yang, Catal. Lett. (2007) 19.
Pd(acac)2 palladium acetylacetonate
ZnCl2
HBF4
zinc chloride
tetrafluoroboric acid
NB4MPyBF4 N-butyl-4-methylpyridinium tetrafluoroborate
GC–MS gas chromatograph–mass spectrometer
XPS
X-ray photoelectron spectroscopy
HPLC
high performance liquid chromatograph
ICP–MS ion coupled plasma emission spectrometer–mass spec-
trometer
FESEM field emission scanning electron microscopy
EFTEM energy-filtered transmission electron microscopy
BE
binding energy
As
mpV
specific surface area
micropore volume
[37] R.D. Seals, R. Alexander, L.T. Taylor, J.G. Dillard, Inorg. Chem. 12 (1973) 2485.