Immobilizing heteropolyacids on zirconia-modified silica as catalysts for oleochemistry transesterification and esterification reactions

Article abstract of DOI:10.1016/j.jcat.2014.09.016

A new method of chemical immobilization of Keggin heteropolyacids (HPAs) was suggested. H₃PW₁₂O₄₀, H₄SiW₁₂O₄₀, and H₃PMo₁₂O₄₀ were immobilized on the silica which was previously grafted with zirconium butoxide. The immobilization method promoted strong interaction HPA-support and yielded 25 wt.% of well-dispersed HPAs, so increasing the density of acid sites. The catalysts were active in the reaction of transesterification of methyl stearate with n-butanol and esterification of oleic acid with trimethylolpropane. We demonstrate that, contrary to the immobilized H₃PMo₁₂O₄₀, the H₃PW₁₂O₄₀ and H₄SiW₁₂O₄₀-based catalysts are stable toward leaching in a non-polar oleic acid medium. A discussion on circumventing the leaching in non-polar versus polar media is proposed in terms of interaction strength HPA-support. The stronger interaction (i.e., better resistance for leaching) between the support and H₃PW₁₂O₄₀ (or H₄SiW₁₂O₄₀) is referred to the lower difference of electronegativity between Zr and W and the lower polarizability of the bonds Zr-O-W compared to Zr-O-Mo.

Full text of DOI:10.1016/j.jcat.2014.09.016

Journal of Catalysis 320 (2014) 1–8
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Immobilizing heteropolyacids on zirconia-modified silica as catalysts
for oleochemistry transesterification and esterification reactions
Maryna Kuzminska a,c, Tetyana V. Kovalchuk , Rénal Backov , Eric M. Gaigneaux
b
c
a,
⇑
a
Université Catholique de Louvain, Institute of Condensed Matter and Nanosciences – IMCN, Croix du Sud 2, Box L7.05.17, B-1348 Louvain-la-Neuve, Belgium
National T. Shevchenko University, Vladimirskaya 64, 01601 Kiev, Ukraine
Université de Bordeaux, Centre de Recherche Paul Pascal, CRPP-UPR CNRS 8641, 115 av. du Dr. A. Schweitzer, F-33600 Pessac, France
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
A new method of chemical immobilization of Keggin heteropolyacids (HPAs) was suggested. H PW12O40
,
3
Received 10 May 2014
Revised 29 August 2014
Accepted 18 September 2014
H
4
SiW12
zirconium butoxide. The immobilization method promoted strong interaction HPA-support and yielded
5 wt.% of well-dispersed HPAs, so increasing the density of acid sites. The catalysts were active in the
reaction of transesterification of methyl stearate with n-butanol and esterification of oleic acid with
trimethylolpropane. We demonstrate that, contrary to the immobilized H PMo12 40, the H PW12
and H SiW12 40-based catalysts are stable toward leaching in a non-polar oleic acid medium. A discus-
3
O40, and H PMo12O40 were immobilized on the silica which was previously grafted with
2
3
O
3
O
40
Keywords:
4
O
Keggin HPA
Immobilization
Transesterification
Esterification
Biofuel
sion on circumventing the leaching in non-polar versus polar media is proposed in terms of interaction
strength HPA-support. The stronger interaction (i.e., better resistance for leaching) between the support
and H
and the lower polarizability of the bonds Zr–O–W compared to Zr–O–Mo.
Ó 2014 Elsevier Inc. All rights reserved.
3 4
PW12O40 (or H SiW12O40) is referred to the lower difference of electronegativity between Zr and W
Biolubricant
Oleic acid
Butyl stearate
1
. Introduction
Particularly, some studies showed that zirconia is able to interact
and stabilize dispersed HPAs [13,14]. However, due to high crystal-
Considering the field of oleochemistry, acid catalysts are of big
linity of ZrO
of active sites remains limited. In order to limit this drawback,
the combination of ZrO with an amorphous support could be
advantageous as the high surface area support would provide good
accessibility of active species and the zirconia would chemically
stabilize heteropolyacids.
2
and often low specific surface area, the accessibility
interest for the biofuel-dedicated transesterification and esterifica-
tion reactions as well as biolubricant productions. Contrary to the
alkaline synthetic paths, acidic catalysts can catalyze simulta-
neously both above-mentioned reactions while avoiding soap for-
mation when low oil quality is employed [1]. Different studies
showed that heteropolyacids (HPAs) are active acidic catalysts
for both esterification and transesterification [2–6]. However, bulk
heteropolyacids (HPAs) are soluble in polar media that make their
recovery difficult. Moreover, if used in non-polar medium, the
HPAs are not soluble and marginally active due to low accessibility
of their active sites. In order to circumvent these penalties, differ-
ent approaches were attempted to heterogenize HPA as for
instances: impregnation into porous supports [7–9], entrapment
in solid matrices via sol–gel [10,11], grafting on functionalized
surfaces [12], and so forth. Nevertheless, the preparation of stable
heterogeneous HPAs that do not leach in polar media while being
well dispersed in non-polar media still remains a challenge.
2
In this work, we demonstrate an alternative approach to immo-
bilize chemically HPAs on zirconia/silica support. The novelty of
the suggested immobilization method relies on the use of zirconia
as a linker between silica and HPA. Within the limit of our knowl-
edge, the attempts to combine silica and zirconia with HPAs were
done so far only by impregnation method or by sol–gel technique
[15,16]. The difference in the HPAs immobilization may contribute
positively or negatively to the catalyst texture, stability, and num-
ber of active sites. Moreover, we studied the stability of prepared
materials toward leaching considering two reactions bearing
industrial applications. First, we studied the stability of the
prepared catalyst toward leaching in a model reaction for biofuel
production, namely the methyl stearate transesterification with
n-butanol. This reaction allowed us both to make screening tests
and justify the efficiency of the suggested synthetic HPAs
immobilization. Second, we tested the catalysts activity for the
⇑
Corresponding author.
E-mail addresses: maryna.kuzminska@uclouvain.be (M. Kuzminska), eric.
gaigneaux@uclouvain.be (E.M. Gaigneaux).
http://dx.doi.org/10.1016/j.jcat.2014.09.016
0
021-9517/Ó 2014 Elsevier Inc. All rights reserved.

2
M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
clean ‘‘solvent-free’’ synthesis of biolubricants by oleic acid esteri-
fication with trimethylolpropane. This reaction has not been exten-
sively studied so far, where only few reports addressed some issues
to be overcome [17,18]. The reaction possesses technical and scien-
tific difficulties among which is the intrinsic molecular hindrance
of the reaction components, being furthermore very difficult to
analyze analytically. Besides, the mass transfer and diffusion limi-
tations, due to high viscosity of the reaction medium, undoubtedly
minimize the reaction kinetics. In this work, we report the success-
ful preparation of stable heterogeneous catalysts appearing as
good candidates when employed for the above-mentioned com-
plex industrial processes.
2.4.3. Fourier Transform Infrared Spectroscopy (FTIR)
FTIR spectra were recorded in transmission mode using an
IFS55 Equinox spectrometer (Bruker) equipped with a DTGS detec-
tor. The spectra were recorded with 100 scans between 400 and
À1
À1
4400 cm with a resolution of 4 cm . The samples were analyzed
after dilution in KBr (Janssens Chimica 99%) by a weight factor of
50.
2.4.4. NMR spectroscopy
31P magic angle spinning (MAS) NMR measurements were per-
formed at 202.47 MHz with a Bruker ASX 500 NMR spectrometer
operating in a static field of 11.7 Teslas. The chemical shift was ref-
3 4
erenced with respect to external 85% H PO .
2
. Materials and methods
.1. Chemicals
Silica (Merck 60, 333 m /g), 80% zirconium butoxide solution in
2
.4.5. X-ray diffraction (XRD)
XRD was used to check the crystallinity and dispersion of the
2
active phase in the synthesized catalysts. Measurements were per-
formed with a Siemens D5000 diffractometer equipped with a
2
1
.6 kW source and using the K
The 2h range was scanned between 5° and 75° at a rate of 0.02°/
s. As a control sample, SiO was mechanically mixed with the cor-
a radiation of Cu (k = 1.5418 Å).
n-butanol (Sigma), pentadecane (Sigma, TCI), acetonitrile, ethanol,
n-butanol, toluene, hexane (all used solvents were of analytical
2
grade or HPLC grade). Keggin HPAs (H
PMo12 40) were used as such without any previous treatment
Sigma). Other chemicals were as follows: N,O-Bis(trimethyl-
silyl)trifluoroacetamide (BSTFA) (Sigma, Alfa Aesar), oleic acid
3 4
PW12O40, H SiW12O40, H3-
responding HPA (18–25 wt.%) and the XRD profile was recorded as
described above.
O
(
2.4.6. N
2
physisorption
9
9
0% (Alfa Aesar), trimethylolpropane (Alfa Aesar), methyl stearate
6% (Sigma), butyl stearate 95% (TCI).
Specific surface area and pore size were obtained through nitro-
gen adsorption–desorption experiments in the Micromeritics Tri-
star 3000. Before analysis, the samples were degassed overnight
under vacuum (0.67 Pa) at 150 °C. The measurements were per-
formed at À196 °C and with relative pressures in the range of
2.2. Silica grafting with zirconium n-butoxide (Step I, Fig. 4)
0
.01–1.00 (P/P
the adsorption isotherm in the P/P
method, and denoted afterward SBET
0
). The specific surface area was calculated from
2
2 g of commercial SiO
00 mbar, 150 °C and for 2 h) was dispersed in 180 ml of toluene.
0 ml of zirconium n-butoxide (ca. 1 mmol/g of SiO ) was then
2
(pre-dried in vacuum at around
0
range of 0.05–0.30 using BET
2
1
.
2
added to the silica suspension under inert atmosphere and vigor-
ous stirring. The mixture was stirred overnight (16 h in total) at
2.5. Catalytic studies
1
05 °C under reflux. Then, the reaction mixture was cooled down
Before each catalytic study, the catalysts were pre-dried at 80 °C
to room temperature, and the final product was washed four times
with toluene. The obtained material – denoted ZrO
stored under a layer of toluene.
in vacuum for 2 h to keep them in the same conditions before the
catalytic tests. All catalytic studies were repeated at least twice to
ascertain the reproducibility of the results.
2 2
/SiO – was
2.3. Immobilization of HPA on the ZrO
2
/SiO
2
(Step II, Fig. 4)
2.5.1. Transesterification reaction of methyl stearate (MS) with n-
butanol (BuOH)
2
.5 g of corresponding Keggin HPA was dissolved in 100 ml of
Previously, pre-dried catalyst (0.1 g) was weighed into the reac-
tor. Then, n-butanol (3 ml) and methyl stearate (0.5 g) were added.
The start of the reaction was considered after all components were
mixed and the reactor placed into an oil bath at 100 °C. The reac-
tion was performed under magnetic agitation (400 rpm) in a batch
closed reactor connected to a reflux. Pentadecane was used as
acetonitrile. Then, 5 g of ZrO /SiO (pre-dried in vacuum at 40 °C
2
2
for 1 h) was added to the HPA solution. The mixture was stirred
at 80 °C for 15 h. Finally, the materials were filtered and washed
with acetonitrile in a Soxhlet apparatus to remove unreacted
HPA. The so-obtained catalysts were dried in air at 110 °C over-
night. Depending on the corresponding HPA used for the synthesis,
internal standard for the GC analysis. For sampling, 20 ll of the
the materials were denoted PW–ZrO
PMo–ZrO /SiO
2
/SiO
2
, SiW–ZrO
2
/SiO
2
, and
reaction mixture was taken at certain time intervals and extracted
with 2 ml of hexane and 2 ml of water. The organic phase was ana-
lyzed by GC. The standard deviation of the detection of butyl stea-
rate was in the range of 0.2–0.9 (%), as calculated from three
parallel tests for each GC injection of kinetic measurements.
2
2
.
2
2
.4. Characterization
.4.1. Elemental analysis
The content of Zr, P, W, and Mo was measured by inductively
coupled plasma-atomic emission spectroscopy (ICP AES) on an Iris
Advantage apparatus from Jarrell Ash Corporation. The amount
2.5.2. Control tests
All catalytic studies were compared to a ‘‘blank’’ test where the
reaction was carried out at the same conditions as described above
but without any catalyst and to the reactions catalyzed by the sup-
(
wt.%) of immobilized HPA was calculated by using the formula Hn-
40 (for X = P, n = 3 and for X = Si, n = 4; M = W or W).
2 2 2
ports (0.1 g) – SiO and ZrO /SiO .
XM12
O
2.5.3. Leaching tests
2
.4.2. Acidity measurements
The acidity of prepared materials was measured by potentio-
metric titration with n-butylamine as described elsewhere [19].
The leaching tests were performed similar as described else-
where [20] with some modifications. The catalysts were treated
with hot butanol (100 °C, 2 h, reflux). The ratio catalyst/butanol

M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
3
was the same as in the catalytic tests. The hot filtrated butanol was
n i i
injection was calculated by using the formula K = C /S (where C
then used for the transesterification reaction as a reactant under
the conditions described above but without any catalyst.
is a concentration of the solution and S is a ratio of peak areas
between the i compound and the internal standard). The response
i
factor (K ) of each standard was taken as an average value from all
2
.5.4. Esterification of oleic acid (OA) with trimethylolpropane (TMP)
injections of the corresponding standard. Obtained response fac-
tors for known compounds were extrapolated on the basis of the
number of C atoms to calculate the theoretical response factors
for products of the catalytic reaction (mono-, di-, and tri-esters of
OA and TMP). See calibration curve in the Supplemental informa-
tion (Fig. S3).
2
.70 g of oleic acid and 0.43 g of TMP (molar ratio between OA
and TMP = 3:1) were weighted into the batch reactor. The mixture
was heated for about 1 min to melt TMP, after which the catalyst
(
0.10 g) was added. The start of the reaction was considered after
all components were mixed, and the reactor was placed in an oil
bath at 120 °C. The reaction was performed under magnetic agita-
tion at 400 rpm and in an open reactor. For sampling, 10 ll of the
3
. Results and discussion
reaction mixture was taken at certain time intervals and deriva-
tized as described below (see Section 2.6) followed by gas chroma-
tography (GC). Standard deviations of the GC measurements were
in the range of 0.04–0.20 (mol/l) for the detection of monoester,
3
1
3.1. Efficiency of the HPA immobilization: FTIR and P NMR
spectroscopy
0
.03–0.15 for diester and 0.01–0.11 for the detection of triester.
We studied the efficiency of the HPA immobilization through
3
1
the FTIR and P NMR spectroscopy. Table 1 depicts a summary
of the main observed bands and their bond signatures. FTIR spectra
2.5.5. Control tests
The ‘‘blank’’ test was performed in identical conditions as
À1
in the range of 1200–600 cm (Fig. 1) indicate the most crucial
described above for the esterification catalytic tests but without
the catalyst.
vibrational mode changes when grafting occurs. The spectra of
supported HPAs are shown after the subtraction of the spectrum
of ZrO
2 2
/SiO .
2
2
.6. Analytics
À1
The stretching band of
after silica treatment with zirconium butoxide (Fig. 1a and b) as
well as the band for free terminal silanols at 3746 cm (Table 1)
indicating the chemical interaction of surface silanols with Zr(OC4-
m
Si–OH) at 967 cm almost disappeared
.6.1. Gas chromatography
For both reactions, the separation on the GC system was
À1
performed as described elsewhere [21]. The reaction of transeste-
rification was followed by the evolution of butyl stearate, and
the concentration was calculated based on the calibration curve
for the analytical standard of butyl stearate. For the esterification
reaction, the evolution of mono-, di-, and triesters was followed.
H
9
)
4
. After further HPA immobilization (Fig. 1c, d, e, and Table 1),
the characteristic M–Od) stretching modes of the Keg-
gin structure appear (where M is W or Mo). Other vibrations of
Keggin units, such as asymmetric
corner-sharing oxygen connecting
observed from the initial IR spectra as they overlap with strong
m
M–Ob–M) and m
m
asP–O (or masSi–O) and m
asM–Oc–M
(
3
M O13 units), cannot be
2.6.2. Derivatization of esterification products
À1
À1
stretching modes of silica Si–O–Si (ꢀ1090 cm and ꢀ800 cm ).
For this reason, the subtraction of the spectrum of the support from
the ones of immobilized HPA was performed (Fig. 1c–e). From the
subtracted spectra of PW- and PMo-catalysts, the P–O bands
appeared broadened which indicates a slight change of symmetry
of initial Keggin anions due to the interaction with the support
To improve the GC detection and separation of low volatile
compounds (fatty acid, mono- and di-esters), the sample was
derivatized with BSTFA following the protocol [21] with some
modifications. Thus, 10
l
l of the reaction mixture was first solubi-
l of BSTFA was added
lized with 2 ml of acetonitrile. Then, 500
l
and the final mixture was heated at 60 °C for 30 min with vigorous
magnetic agitation. After the derivatization was finished, the final
products were extracted with 2 ml of hexane followed by GC
injection.
[
14].
It should be indicated also that the symmetric and asymmetric
À1
stretching
m
(C–H) vibrations at 3000 cm
disappeared (Table 1)
/SiO that indicates the
9
H O-groups after the HPA immo-
after the reaction between HPA and ZrO
complete hydrolysis of surface C
bilization on ZrO /SiO
Additional information on the structure of immobilized HPA
was obtained by P NMR. In Fig. 2, the NMR spectra of studied cat-
alysts are compared to the ones of pure HPA. From Fig. 2, the peak
2
2
4
2
.6.3. Identification of the reaction components
Since no standards for the products of esterification of OA with
2
2
.
TMP are available commercially to determine their retention times,
the solution was synthesized from pure methyl oleate (using pro-
tocol [21], with some modifications). This ester reacted with TMP
under sodium methoxide as catalyst. The methyl ester-to-TMP
molar ratio and catalyst (w/w percentage) were fixed at 3:1 and
3
1
at À13.7 ppm of PW–ZrO
2 2
–SiO indicates that the Keggin structure
is distorted due to the strong interactions with the support but the
fundamental Keggin structure is preserved [24]. Additionally, the
chemical shift indicates that there is one proton directly attached
to the heteropolyanion (i.e., this proton is acidic) [25]. Similarly,
0
.8%, respectively, to ensure optimal conversion. The reaction
was carried out at 110 °C in the open flask for 10 h. The reaction
products were neutralized to remove the catalyst and then distilled
under vacuum to remove traces of water. The derivatized transe-
sterification products were then injected to the GC to identify their
retention times. Retention times for oleic acid and trimethylolpro-
pane were identified from their commercial standards.
3
1
the P NMR signal of PMo–ZrO
2 2
–SiO shows a strong interaction
between support-HPA.
To sum up, FTIR and 31P NMR data confirm efficient immobiliza-
tion of HPA onto porous support with preserving the initial struc-
ture of Keggin units which is crucial for their stability and catalytic
performance.
2
.6.4. Response factor prediction for mono-, di-, and triesters
For the prediction of response factors of the products, standard
solutions of 2-ethylhexyl butyrate (EHB), ethyl stearate (ES),
n-butyl stearate (BS), and glycerol trioleate (GTO) were used. For
each compound, at least three different concentrations have been
injected to the GC. At each standard solution, a known amount of
3.2. Acidic properties
The acidity of prepared materials was performed by means of
potentiometric titration with n-butylamine. This method allows
measuring the total number and the strength of acid sites. In such
n
internal standard was added. The response factor K for each

4
M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
Table 1
Summary of the FTIR bands observed for prepared materials. Characteristic absorption bands were assigned based on the literature [22,23].
À1
Material
SiO
Bands,
m
(cm
)
2
460 (Si–O out of plane deformation), 800 (Si–O bending), 967 (free silanols Si–OH stretching), intensive broad band at 1100–1225 (stretching of
siloxanes Si–O–Si), broad ꢀ3500 (hydrogen bonded OH), 3746 (free terminal silanol groups Si–OH)
ꢀ3000 (C–H symmetric and asymmetric stretching modes of butoxy groups), broad ꢀ3500 (hydrogen bonded OH), diminished (almost disappeared)
at 967 (stretching of Si–O–Si or/and Si–O–Zr)
ZrO
2 2
/SiO
2
PW–ZrO /SiO
2
d b
1078 (masP–O, from the subtracted spectrum), 983 (W–O asymmetric ligand vibration, terminal O bonded to W atom), 893 (interligand W–O –W,
edge-sharing O connected to W), broad ꢀ3500 (hydrogen bonded OH); the band for C–H stretching (ꢀ3000) disappeared indicating the complete
hydrolysis of the butoxy groups
SiW–ZrO
PMo–ZrO
2
/SiO
/SiO
2
977 (as. ligand W–O
1088 ( asP–O, from the subtracted spectrum), 953 (as. ligand Mo–O
for C–H stretching (ꢀ3000) disappeared
d
), 923 (interligand W–O
b
–W), broad ꢀ3500 (hydrogen bonded OH); the band for C–H stretching (ꢀ3000) disappeared
2
2
m
d
), 885 (interligand Mo–O
b
–Mo), broad ꢀ3500 (hydrogen bonded OH); the band
strength can be defined according to the following ranges: very
strong site, E > 100 mV; strong site, 0 < E < 100 mV; weak site,
À100 < E < 0 mV; and very weak site, E < À100 mV [26].
The titration curves of the supported HPA (Fig. 3) indicate that
all prepared catalysts have strong acid sites (E > 100 mV) and pres-
ent similar behavior in the potentiometric titration. The acidity
trend of the prepared materials decreases in the order SiW–ZrO
SiO (E /SiO (E 348 mV) > PMo–ZrO /SiO
370 mV) ꢀ PW–ZrO
206 mV).
In the case of SiO
SiO . This last value suggests an interaction of surface silanol
groups of silica and formation of butoxy groups on the surface.
(
a)
i
i
(
b)
i
i
i
(
c)
d)
2
/
2
i
2
2
i
2
2
(
(
2
, the E
i
is 41 mV and it is À39 mV for ZrO
2
/
(
e)
2
(
f)
(
g)
3.3. Content and suggested immobilization scheme
Table 2 shows the content of Zr, P, and W (Mo) of the catalysts
(
h)
obtained by ICP analysis.
From elemental analysis, the data demonstrate that the syn-
thetic procedure yields up to 0.6 mmol/g of Zr (ca. 5 wt.%) and
0
.1 mmol/g of HPA (18–25 wt.%). The molar ratio HPA/Zr allows
to propose the manner on which the Zr species and the HPA are
anchored together. Thus, the maximal theoretical HPA/Zr ratio
would be 0.75 (assuming that zirconium butoxide reacted with
the silanols by only one butoxy group and that the three remaining
1
200 1100 1000 900 800 700 600
Wavenumber (cm-1
)
À1
Fig. 1. FTIR spectra in the range 1200–600 cm of: (a) SiO
2
, (b) ZrO
PW12 40, (g) H
For shown (c)–(e) spectra, the spectrum of ZrO
2
/SiO
2
, (c) PW–
SiW12
/SiO was
ZrO
h)
subtracted.
2
/SiO
2
, (d) SiW–ZrO
2
/SiO
2
, (e) PMo–ZrO
2
/SiO
2
, (f) H
3
O
4
O
40
,
4 9
C H O-groups further reacted with HPA through the hydrolysis
(
H
3
PMo12
O
40
.
2
2
step). Although lower than 0.75, the HPA/Zr value (ca. 0.2, see
Table 2) suggests different outcomes that may take place during
the synthesis.
method, the initial electrode potential (E
acid strength of the surface sites and the range where the plateau is
reached (meq/g solid) shows the total number of acid sites. The
i
) indicates the maximum
Taking into account experimental HPA/Zr molar ratio and steric
hindrance of Zr(OBu)4 and HPAs, we would like to suggest the
scheme of the HPA immobilization (Fig. 4). Thus, for the step I
-
15,2 ppm
- 4,1 ppm
H PW O
H PMo O
3
12 40
3
12 40
1
0
0
-10
-20
-30
-40
-40
10
10
0
-10
-20
-30
-40
-
13,7 ppm
-
2,5 ppm
PW-ZrO /SiO
2
2
PMo-ZrO /SiO
2 2
10
0
-10
-20
-30
0
-10
-20
-30
-40
(ppm)
(ppm)
Fig. 2. 31P NMR of PW–ZrO
and PMo–ZrO
2
/SiO
2
2 2
/SiO compared to of pure HPA.

M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
5
4
3
2
1
00
00
00
00
0
polarizability of Zr–O–Mo bonds is not favoring the stability, espe-
cially in polar environment. And finally, the interaction between Zr
center and more than one heteropolyanion is unlikely due to the
large size of HPA (ca. 1 nm versus 0.072 nm for Zr [29]).
To sum up, the surface Zr species are bipedal and the two
remaining butoxy groups would be replaced by one heteropolyan-
PW-ZrO /SiO
2
2
SiW-ZrO /SiO
4+
2
2
PMo-ZrO /SiO
2
2
SiO2
ZrO /SiO
3
ion. For materials based on triprotic HPAs (H PW12O40 and H3-
2
2
PMo12 40), one proton per polyanion would thus remain on the
O
surface (as shown in Fig. 4, the HPA having to release two protons
to gain two-negative charges as needed to neutralize the two time
positively charged Zr center) whereas for the tetraprotic H
4
SiW12-
O40 – two protons would remain. Moreover, the suggested scheme
of the synthesis leads to 0.25 maximal theoretical HPA/Zr ratio
which is comparable to the experimental value 0.2.
-
100
0
,0
0,4
0,8
1,2
1,6
meq/g
3.4. Textural characteristics
Fig. 3. Potentiometric titration curves of studied materials.
The specific surface area (SBET), mean pore diameter, and pore
volume are shown in Table 3. The results show that after silica
modification with zirconium butoxide and further immobilization
of HPAs, the surface area and pore volume of the materials
decreased to the same extend. The changes of the texture of the
(
Fig. 4), the Zr(OBu)
ence of the highly electronegative C
4
is a very reactive molecule due to the pres-
O-group and has high affin-
4 9
H
materials suggest successful support modification with Zr(OBu)
and further HPA immobilization.
4
ity toward surface silanols [27]. The electrophilic Zr center would
attack surface silanol and thereby would form the Si–O–Zr bond.
The FTIR spectra confirmed this step (see Table 1), as the bands
À1
3
746 and 967 cm (for silanol groups) disappeared after the syn-
3.5. X-ray diffraction
2 2
thesis. Moreover, the TGA analysis of ZrO /SiO showed 6% of the
mass loss after 300 °C which indicates the presence of residual
chemically bonded butoxy groups (data are shown in the Supple-
mental information, Fig. S2). The further reaction between the
adjacent silanol and the second leaving butoxy group (formation
of bipedal Zr species) would be also possible since the conforma-
tional structure of zirconium butoxide allows it (tetrahedral
arrangement of butoxy groups around Zr atom). For the same rea-
son, the reaction between the third butoxy group and the adjacent
silanol (formation of tri-pedal Zr species) would be sterically hin-
dered. For the step II (Fig. 4), due to the well-matched electroneg-
X-ray diffraction profiles of immobilized HPAs (example is
shown for PW–ZrO /SiO , Fig. 5a) are compared to those of the
2
2
supports and to the control sample (Fig. 5a–d). We did not observe
the crystalline phases for the immobilized HPA samples which sug-
gest that HPAs are homogeneously distributed onto the support in
a non-crystalline form or that crystallites are very small. The later
specificity is of importance for the catalytic properties of the syn-
thesized materials as accessibility of the active species and active
surface would be enhanced.
4
+
6+
ativity of Zr and W (1.33 and 1.7 correspondingly [28]), further
reaction between Keggin units and the surface Zr oxide species will
go through the formation of Zr–O–W covalent bonds at the inter-
3.6. Catalytic studies
3.6.1. Transesterification of methyl stearate with n-butanol
The control catalytic tests (Fig. 6A) showed that the supports
SiO and ZrO /SiO do not catalyze the reaction.
face of the two components [16]. Similarly, for the H
3 40
PMo12O
acid, the Zr–O–Mo bonds will form. However, in this case, the dif-
2
2
2
ference in electronegativity between Zr4 and Mo is much higher
+
6+
The lower activity of the SiO (compared to the blank test) likely
2
(
1.33 and 2.16 correspondingly) which will lead to a higher ionic
comes from internal diffusion limitations. This phenomenon can be
referred to the bulky reagents which retain decreased mobility
character of Zr–O–Mo bond compared to the Zr–O–W. The higher
Table 2
Elemental analysis (ICP AES data).
Material
SiO
ZrO
PW–ZrO
SiW–ZrO
PMo–ZrO
Zr, mmol/g
P,
l
mol/g
W (Mo), mmol/g
HPA, mmol/g (calculated)
HPA/Zr (calculated)
HPA, wt.% (calculated)
2
–
–
–
0.10
–
–
–
1.03
0.97
1.16
–
–
0.09
0.08
0.10
–
–
0.17
0.15
0.18
–
–
25
23
18
2
/SiO
2
0.58
0.54
0.54
0.57
2
/SiO
/SiO
/SiO
2
2
2
2
2
0.10
Fig. 4. Hypothetical model of the reaction between silica and zirconium butoxide and further HPA interaction with the modified support. The number of Zr atoms per Si–OH
group as well as per HPA unit was taken arbitrarily.

6
M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
Table 3
(
A) 0,0050
e)
Textural properties of prepared materials (N
2
physisorption data).
BJH, Å
0
0
,0049
,0048
2
3
Material
SiO
ZrO
PW–ZrO
SiW–ZrO
PMo–ZrO
S
BET, m /g
Ø
Vp(BJH), cm /g
2
443
332
230
216
256
50
50
56
55
58
0.76
0.60
0.41
0.38
0.46
2
/SiO
2
2
/SiO
/SiO
/SiO
2
2
2
0
0
0
,002
,001
,000
2
2
(c)
b)
(
(
a)
b)
(a)
0
250
500
750
1000 1250 1500
Reaction time (min)
(
(
B) 0,005
(
e)
c)
(c)
(
0
0
0
,004
,003
,002
(
d)
1
0
20
30
40
50
60
70
2Θ (Degree)
(
b)
Fig. 5. XRD patterns for PW–ZrO
SiO mixed with 25% w/w H PW12
2
/SiO
2
(a), SiO
2
(b), ZrO
2
/SiO
2
(c), control sample –
(a)
2
3
O
40 (d).
0,001
0
,000
inside the small pores of the support. The latter is not observed
with ZrO /SiO likely due to reduced size of the pores by zirco-
nia-precursor which prevents the reagents from entering the
pores. The catalytic data of the immobilized HPAs showed that
they are catalytically active for transesterification reaction
0
250
500
750
1000 1250 1500
2
2
Reaction time (min)
(
C) 0,0050
,0049
0,0048
e)
0
(
Fig. 6B). After 24 h of the reaction, the yield of butyl stearate
reached 22%, 30%, and 85% for PW–ZrO /SiO , SiW–ZrO /SiO
and PMo–ZrO /SiO correspondingly. The activity of the immobi-
2
2
2
2
,
2
2
lized HPAs is reversed to the known acidity of bulk HPAs as it
was also observed for the homogeneous HPAs in the transesterifi-
cation [2]. Thus, it is known that the acidity decreases in the order
(c)
0
0
0
,002
,001
,000
3 4 3
H PW12O40 > H SiW12O40 > H PMo12O40. In our case, the reversed
activity trend of heterogenized HPAs (compared to the acidity
order) could be referred (a) to a partly homogeneous character of
the catalytic process; (b) to a higher accessibility of the active spe-
cies in the heterogeneous catalysts. To validate the leaching sce-
nario into the homogeneous phase, we treated the catalysts with
hot butanol and used that butanol farther as a reagent in the
transesterification reaction without any additional catalyst.
Fig. 6C demonstrates the results of leaching tests. The kinetic
curves for the tests with BuOH used for the treatment of PW–
(
d)
(b)
(a)
0
250
500
750
1000 1250 1500
Reaction time (min)
Fig. 6. (A) Control catalytic tests of the transesterification of MS with BuOH: (a)
SiO as a catalyst; (b) blank test; (c) test with ZrO /SiO . (B) Kinetic curves of the
transesterification of MS catalyzed by: (a) PW–ZrO /SiO ; (b) SiW–ZrO /SiO ; (c)
PMo–ZrO /SiO . (C) Leaching tests: transesterification of MS performed with BuOH
2
treatment; (b) test
treatment; (D) blank
test. On each graph (e) dot curves show the maximal product yield determined from
the catalytic test with homogeneous HPA (see Supplemental information for
details).
2
2
2
ZrO
with the blank test (curve d) within the error range, whereas the
curve for the reaction with BuOH after PMo–ZrO /SiO (c curve)
is above the curve d. These observations suggest that PMo–ZrO
SiO showed strong leaching of corresponding HPA into the reac-
2 2 2 2
/SiO and SiW–ZrO /SiO (curves a and b) are superimposing
2
2
2
2
2
2
used for the catalyst treatment. (a) Test after SiW–ZrO
after PW–ZrO /SiO treatment; (c) test after PMo–ZrO /SiO
2
/SiO
2
2
2
2
2
2
2
/
2
tion medium. Additional information on the leaching was obtained
by ICP analysis of the materials before and after catalytic tests. In
Table 4, the concentration of tungsten (molybdenum) within the
catalysts is compared before and after the catalytic tests. It can
be seen that, after the transesterification, the amount of W (Mo)
decreased by 10–17 mol% of the initial amount. The obtained
results indicate that all the materials partly leached into the liquid
medium. The leaching is likely caused by dissolution of the corre-
sponding HPA into polar butanol used in excess as reagent and
solvent.
3.6.2. Esterification of oleic acid with trimethylolpropane
The catalysts were also tested in the reaction which has indus-
trial interest – namely the esterification of oleic acid with trime-
thylolpropane for the production of biolubricant. Fig. 7 shows the
evolution of esters in the esterification of OA.
Comparison of the catalytic performances with the control test
(dash curves in Fig. 7) demonstrates that the materials are catalyt-

M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
7
Table 4
limitations take place, which explains moderate activity of the sup-
ported HPAs. The later suggests again that the support with larger
pores would help to avoid this penalty.
Additional studies on the leaching during the esterification
were performed by elemental analyses. The content of tungsten
(molybdenum) within the materials after esterification showed
Concentration of tungsten (molybdenum) within the catalysts before and after
catalytic tests (ICP AES data).
Material
W (Mo), mmol/g
Before catalytic
test
After
transesterification
After
esterification
PW–ZrO
SiW–ZrO
PMo–ZrO
2
/SiO
/SiO
/SiO
2
1.03
0.97
1.16
0.85
0.87
1.00
1.03
0.98
0.97
that PW–ZrO
of tungsten as before the catalytic test (Table 4). Contrary, the
PMo–ZrO /SiO decreased the amount of molybdenum after the
esterification. The results indicate that in the hydrophobic medium
of oleic acid, PMo–ZrO /SiO leached into the liquid which contrary
to PW–ZrO /SiO and SiW–ZrO /SiO . The latter can be explained
2 2 2 2
/SiO and SiW–ZrO /SiO retain the same amount
2
2
2
2
2
2
2
2
2
2
2
2
PW-ZrO /SiO
2
2
2
1
1
0
0
,5
,0
,5
,0
,5
,0
2
by the different stability of the corresponding HPAs on the surface.
The hypothesis is that reaction between the support and
SiW-ZrO /SiO
2
2
PMo-ZrO /SiO
2
2
3
H PW12
O
4
40 (or H SiW12
O40) yields strong interactions support-
Monoester
Blank
HPA whereas the interaction support-H
3
PMo12 40 is much weaker.
O
The latter is likely due to the higher polarizability of the Zr–O–Mo
bonds as a direct consequence of the difference in electronegativity
between Zr and Mo (see Section 3.3).
Diester
Triester
Interesting to mention that even after the materials were used
in the catalytic reactions, their FTIR spectra still contain the char-
acteristic HPA vibrations (data are shown in Supplemental infor-
mation on Fig. S4; for the FTIR measurements, the used materials
were washed first with hexane and acetone after catalytic tests
and dried at room temperature). The latter suggests that immobi-
lized HPAs retain their Keggin structure even after being treated at
the conditions of catalytic measurements.
Diester
Triester
0
250
500
750
1000 1250 1500
Reaction time (min)
To conclude, the catalytic and elemental analysis results indi-
cate that PMo–ZrO
non-polar medium compared to PW–ZrO
SiO . All three materials showed leaching in the model reaction
of transesterification induced by polar butanol. However, in the
industrial reaction of biolubricant production, the PW–ZrO /SiO
and SiW–ZrO /SiO do not leach, showing thereby their potential
toward industrial applications. We can conclude that the applied
synthetic procedure is efficient for immobilization of H
and H SiW12 40 for applications in non-polar media. It allows
obtaining stable-toward-leaching catalysts that catalyze complex
step-by-step reactions in the field of oleochemistry. The benefit
of heterogenized heteropolyacids compared to the homogeneous
ones lies in the higher dispersion of the active species and in the
practical and economical aspects where separation and recyclabil-
ity of immobilized HPAs become much easier from the liquid
medium.
2
/SiO
2
is less stable toward leaching in the
1
1
0
0
0
0
,25
,00
,75
,50
,25
,00
Diester
Diester
2
/SiO and SiW–ZrO
2
2
/
Monoester
2
2
2
2
2
3
PW12O
40
4
O
Triester
0
50
100 150 200 250 300 350
Reaction time (min)
Fig. 7. Evolution of mono-, di-, and triesters concentration in the esterification of
oleic acid with trimethylolpropane. (a) ( ) Test with PW–ZrO /SiO ; ( ) test with
SiW–ZrO /SiO ; ( ) test with PMo–ZrO /SiO . Dash curves show the evolution of
2
2
2
2
2
2
the product in the blank test. Lower figure is the same as the upper one but focused
in the lower reaction time and product concentrations.
4
. Conclusion
ically active, notwithstanding that the activity is moderate and
does not differ significantly between the three studied catalysts.
Only PMo–ZrO /SiO appears slightly more active in the beginning
2 2
of the reaction (see Fig. 7, zoomed image). The kinetic curves for
the evolution of esters show that the catalysts accelerate signifi-
cantly the formation of mono- and diesters.
The reaction rate for triester formation in the beginning of the
test remains similar to the blank test and can be explained by steric
hindrance of the diester – the bulky molecule bears the last hydro-
xyl-group which is less accessible for the acid to attack. It is also
possible that the pores of the catalyst support are too small and
the formation of triester inside the pores is not favorable. The latter
hypothesis suggests that the use of a support with more suitable
texture (i.e., bigger pores) would facilitate the formation of bulky
triester. The heterogeneous catalytic performance is close to the
homogeneous HPAs (the data for the homogeneous HPA are
addressed in the Supporting information). In the current case, the
size of the reactants and products is too high and internal diffusion
In this study, we proposed a new method for chemical immobi-
lization of Keggin HPAs at silica through the link of zirconia. The
method yields up to 25 wt.% of immobilized HPA and increased
density of acid sites compared with the silica-based initial sup-
ports. The reaction between silica and zirconium butoxide likely
leads to bipedal zirconium species on the surface and 0.2 M ratio
of HPA/Zr. The catalytic tests demonstrate that the synthesized
materials are active for both transesterification and esterification
reactions of bulky molecules. On the one hand, three catalysts
address partly a homogeneous catalytic behavior in the transeste-
rification reaction due to the leaching of active species into the
liquid phase. On the another hand, in the oleic acid esterification,
immobilized H
to leaching contrary to PMo–ZrO
3
PW12
O
40 and H
4
SiW12
/SiO
O
40 exhibit clear resistance
that leaches even in non-
2
2
polar medium. The latter is explained by difference in interaction
strength between the HPAs and the zirconia-modified silica sup-
port. The stronger interaction between the support and
3 4
H PW12O40 (or H SiW12O40) is explained with the lower difference

8
M. Kuzminska et al. / Journal of Catalysis 320 (2014) 1–8
of electronegativity between Zr and W that minimizes polarizabil-
ity of the bonds Zr–O–W when compared with the Zr–O–Mo ones.
For future work, we believe that by increasing the pore diameters
of the catalysts as well as by changing catalytic conditions, it could
be possible to increase the performance of the explored catalysts.
[6] A. Zieba, L. Matachowski, J. Gurgul, E. Biela n´ ska, A. Drelinkiewicz, J. Mol. Catal.
A: Chem. 316 (2010) 30–44.
[
[
7] B.B. Bardin, R.J. Davis, Appl. Catal., A 200 (2000) 219–231.
8] S. Benadji, P. Eloy, A. Léonard, B.L. Su, K. Bachari, C. Rabia, E.M. Gaigneaux,
Microporous Mesoporous Mater. 130 (2010) 103–114.
[
9] E. Grinenval, X. Rozanska, A. Baudouin, E. Berrier, F.o. Delbecq, P. Sautet, J.-M.
Basset, F.d.r. Lefebvre, J. Phys. Chem. C 114 (2010) 19024–19034.
[
[
10] Á. Molnár, C. Keresszegi, B. Török, Appl. Catal., A 189 (1999) 217–224.
11] I. Dosuna-Rodriguez, E.M. Gaigneaux, in: Proceedings Book of the 1st
International Congress on Catalysis for Biorefineries, 2nd–5th Oct 2011, pp.
Acknowledgments
4
58–463.
12] Y. Hanada, M. Kamada, K. Umemoto, H. Kominami, Y. Kera, Catal. Lett. 37
1996) 229–233.
13] T. Rajkumar, G. Ranga Rao, J. Mol. Catal. A: Chem. 295 (2008) 1–9.
[
[
The authors acknowledge the financial support from the
‘DGTRE, Région Wallonne’’ for the project ‘‘CATHESTOL’’ (Conven-
tion number 1117464) and the company ‘‘Mosselman s.a.’’ for con-
structive interaction and the support. The authors also would like
to thank Christel Gervais (Professor of UPMC, Paris) for the solid
P NMR analysis. The work was performed in the framework of
IDS-FunMat doctoral program supported by the Erasmus Mundus,
Program of EU-FP7.
(
‘
[14] E. López-Salinas, J.G. Hernández-Cortéz, I. Schifter, E. Torres-Garcıa, J.
´
Navarrete, A. Gutiérrez-Carrillo, T. López, P.P. Lottici, D. Bersani, Appl. Catal.,
A 193 (2000) 215–225.
[
15] L.M. Gomez Sainero, S. Damyanova, J.L.G. Fierro, Appl. Catal., A 208 (2001) 63–
75.
16] F. Su, L. Ma, Y. Guo, W. Li, Catal. Sci. Technol. 2 (2012) 2367–2374.
17] C.O. Åkerman, Y. Gaber, N.A. Ghani, M. Lämsä, R. Hatti-Kaul, J. Mol. Catal. B:
Enzym. 72 (2011) 263–269.
3
1
[
[
[
[
[
18] C.O. Åkerman, A.E.V. Hagström, M.A. Mollaahmad, S. Karlsson, R. Hatti-Kaul,
Process Biochem. 46 (2011) 2225–2231.
19] M. del Rosario Torviso, E.N. Alesso, G.Y. Moltrasio, P.G. Vázquez, L.R. Pizzio, C.V.
Cáceres, M.N. Blanco, Appl. Catal., A 301 (2006) 25–31.
20] M. Di Serio, R. Tesser, L. Casale, A. D’Angelo, M. Trifuoggi, E. Santacesaria, Top.
Catal. 53 (2010) 811–819.
21] R. Yunus, O.T. Lye, A. Fakhru’l-Razi, S. Basri, JAOCS 79 (2002) 1075–1080.
22] R. Thouvenot, M. Fournier, R. Franck, C. Rocchiccioli-Deltcheff, Inorg. Chem. 23
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.09.016.
[
[
(
1984) 598–605.
References
[23] P.L. Launer, Silicon Compounds: Register and Review, in: B. Arkles (Ed.),
Petrarch Systems, 1987.
[
24] L.F. Chen, J.A. Wang, L.E. Noreña, J. Aguilar, J. Navarrete, P. Salas, J.A. Montoya,
P. Del Ángel, J. Solid State Chem. 180 (2007) 2958–2972.
25] A. Micek-Ilnicka, J. Mol. Catal. A: Chem. 308 (2009) 1–14.
26] R. Cid, G. Pecchi, Appl. Catal. 14 (1985) 15–21.
27] C.J. Brinker, G.W. Scherer, The Physics and Chemistry of Sol–Gel Processing,
Academic Press, New York, USA, 1990.
[
[
1] E. Santacesaria, G.M. Vicente, M. Di Serio, R. Tesser, Catal. Today 195 (2012) 2–
3.
2] P. Morin, B. Hamad, G. Sapaly, M.G. Carneiro Rocha, P.G. Pries de Oliveira, W.A.
Gonzalez, E. Andrade Sales, N. Essayem, Appl. Catal., A 330 (2007) 69–76.
3] M.N. Timofeeva, Appl. Catal., A 256 (2003) 19–35.
4] O.d.S. Lacerda, R.M. Cavalcanti, T.M.d. Matos, R.S. Angélica, G.N. da Rocha Filho,
I.d.C.L. Barros, Fuel 108 (2013) 604–611.
5] A.I. Tropecêlo, M.H. Casimiro, I.M. Fonseca, A.M. Ramos, J. Vital, J.E.
Castanheiro, Appl. Catal., A 390 (2010) 183–189.
1
[
[
[
[
[
[
[
28] http://bilbo.chm.uri.edu/PeriodicTable/.
29] I.K. Song, M.S. Kaba, M.A. Barteau, Langmuir: ACS J. Surf. Colloids 18 (2002)
[
2358–2362.