Understanding the role of Keggin type heteropolyacid catalysts for glycerol acetylation using toluene as an entrainer

Article abstract of DOI:10.1016/j.apcata.2016.08.016

The heterogeneously catalyzed esterification (acetylation) of glycerol toward triacetin in batch mode in presence of toluene as entrainer was studied. Silicotungstic acid, tungstophosphoric acid and phosphomolybdic acid as heteropolyacids (HPAs) supported on silica, alumina or silica-alumina were used as catalysts. The course of the reaction was found to be very sensitive to the nature of the HPA as well as the support. Solid characterization by Raman spectroscopy, XRD, and pyridine-FTIR revealed that only combinations of tungsten-based HPAs and silica support were able to preserve the structure of active component throughout the preparation process, which was essential to obtain active and selective catalysts. The interaction between HPA and support was decisive for stability and dispersion of the catalytically active species. With the best performing catalyst H₄SiW₁₂O₄₀/SiO₂, selectivity to triacetin reached 71% at complete conversion within 24?h. The high selectivity to triacetin is attributed the Br?nsted acidic sites originated from stabilized Keggin structure and continuous removal of water during course of reaction. Toluene is able to form azeotropic mixtures with water and acetic acid and keeps the reaction temperature below the boiling point of acetic acid. Thus, water-free reaction conditions can be established. The catalyst was reusable; however, the activity and selectivity towards triacetin slightly decreased in a repetition run due to loss of active sites.

Full text of DOI:10.1016/j.apcata.2016.08.016

Accepted Manuscript
Title: Understanding the role of Keggin type heteropolyacid
catalysts for glycerol acetylation using toluene as an entrainer
Author: S.S. Kale U. Armbruster R. Eckelt U. Bentrup S.B.
Umbarkar M.K. Dongare A. Martin
PII:
S0926-860X(16)30423-9
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2016.08.016
APCATA 15979
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
27-4-2016
28-7-2016
17-8-2016
Please cite this article as: S.S.Kale, U.Armbruster, R.Eckelt, U.Bentrup, S.B.Umbarkar,
M.K.Dongare, A.Martin, Understanding the role of Keggin type heteropolyacid
catalysts for glycerol acetylation using toluene as an entrainer, Applied Catalysis A,
General http://dx.doi.org/10.1016/j.apcata.2016.08.016
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Understanding the role of Keggin type heteropolyacid catalysts for glycerol acetylation
using toluene as an entrainer
S. S. Kale a, U. Armbruster a, R. Eckelt a, U. Bentrup a, S. B. Umbarkar b, M. K. Dongare b, c
A. Martin a,*
,
a
Leibniz-Institut für Katalyse e. V. an der Universität Rostock, Albert-Einstein-Str. 29a, D-
18059 Rostock, Germany
b Catalysis Division, CSIR-National Chemical Laboratory, Pune 411008, India
c Mojj Engineering Systems Ltd., 15-81/B, MIDC, Bhosari, Pune 411026, India
*Corresponding author. Phone: +49(381)1281-246, Fax: +49 381 1281 51246.
E-mail address: andreas.martin@catalysis.de (A. Martin).
Keywords: Glycerol, Acetylation, Azeotropes, Heteropolyacids, Triacetin, Heterogeneous
catalyst

Graphical abstract
a)
100
80
60
40
20
0
% (Conversion)
supported HPA catalyst
% (MAG)
% (DAG)
% (TAG)
% (DGTA)
0
5
10
15
20
25
Time / h

Highlights
High Brønsted acidity is required to achieve good performance.
Preservation of the Keggin structure of the HPAs is mandatory.
Toluene as entrainer enables high triacetin yields.
H4SiW12O40/SiO2 reached selectivity to triacetin of 71% at complete glycerol conversion.

Abstract
The heterogeneously catalyzed esterification (acetylation) of glycerol toward triacetin in
batch mode in presence of toluene as entrainer was studied. Silicotungstic acid,
tungstophosphoric acid and phosphomolybdic acid as heteropolyacids (HPAs) supported on
silica, alumina or silica-alumina were used as catalysts. The course of the reaction was found
to be very sensitive to the nature of the HPA as well as the support. Solid characterization by
Raman spectroscopy, XRD, and pyridine-FTIR revealed that only combinations of tungsten-
based HPAs and silica support were able to preserve the structure of active component
throughout the preparation process, which was essential to obtain active and selective
catalysts. The interaction between HPA and support was decisive for stability and dispersion
of the catalytically active species. With the best performing catalyst H4SiW12O40/SiO2,
selectivity to triacetin reached 71% at complete conversion within 24 h. The high selectivity
to triacetin is attributed the Bronsted acidic sites originated from stabilized Keggin structure
and continuous removal of water during course of reaction. Toluene is able to form
azeotropic mixtures with water and acetic acid and keeps the reaction temperature below the
boiling point of acetic acid. Thus, water-free reaction conditions can be established. The
catalyst was reusable; however, the activity and selectivity towards triacetin slightly
decreased in a repetition run due to loss of active sites.

1. Introduction
Glycerol is inevitably produced in the biodiesel production from transesterification of
vegetable oil or animal fats and has led to a drastic surplus in chemical market. Global
production of biodiesel market is estimated to reach 36.9 million metric tons in 2020, which
will give approximately 3.7 million metric tons of crude glycerol [1, 2, 3]. The current use of
glycerol in pharmaceuticals, food and cosmetics consumes only a small part and hence,
demand is somehow limited. To sustain chemical market and industry, it is important to
convert glycerol into value added chemicals by several catalytic processes, such as selective
oxidation to glyceric acid or hydroxyacetone [ 4 ], dehydration to acrolein [ 5 - 7 ],
hydrogenolysis to 1,2- or 1,3-propanediol [ 8 ], etherification to alkyl ether [ 9 , 10 ],
condensation to dimers or higher oligomers [11] and many others [3, 12]. Another interesting
approach to convert glycerol into monoacetyl glycerol (MAG, monoacetin), diacetyl glycerol
(DAG, diacetin) and triacetyl glycerol (TAG, triacetin) by means of esterification with acetic
acid (acetylation). These products have application in food and leather industry, as
plasticizers and also they may serve as solvent and fuel additive [13, 14, 15].
Glycerol acetylation is an acid catalyzed reaction with equimolar formation of water as by-
product at every consecutive step and the chemical equilibrium limits the extent of the
esterification. In addition, Gibbs free energies of the first two acetylation steps (MAG and
DAG) are 19.15 and 17.80 kJ/mol, respectively, whereas this value for the third step (TAG)
is relatively high (55.58 kJ/mol) and thus, the third step should be the most difficult one [16].
Beyond that, elevated temperatures are necessary to obtain high reaction rates. Generally the
reaction is performed using homogeneous catalysts like sulfuric acid and paratoluene sulfonic
acid [17] or heterogeneous catalysts such as sulphated mesoporous silica [18], sulphated
zirconia [19, 20], sulphated activated carbon [21, 22], double SO3H-functionalized ionic
liquids [23] and acid ion exchange resins like Amberlyst-15 or Amberlyst-36 [24-26]. On the
other side, use of acetic anhydride as acetylation agent instead of acetic acid boosts the
selectivity to TAG close to 100% as no water is formed, and the reaction rates are high even
in the absence of catalysts. However, acetic anhydride is more expensive and hazardous than
acetic acid, and it is not environmentally friendly [27, 28].
Furthermore, supported heteropolyacids have been employed for glycerol acetylation
reaction. Phosphomolybdic acid encaged in a zeolite catalyst showed 68% glycerol
conversion with only 2% selectivity to TAG [29]. Very recently, Patel et al. used MCM-41
and ZrO2 supported tungstophosphoric acid catalysts but the selectivity to TAG was found to
be rather low with 15% and 4%, respectively [30]. The same group also used 30 wt%
tungstophosphoric acid impregnated on MCM-48 as catalyst with 30% selectivity to TAG
using an acetic acid/glycerol molar ratio of 6 [31]. Silicotungstic acid supported on zirconia
showed 32.3% selectivity to TAG with complete conversion of glycerol using acetic
acid/glycerol molar ratio of 10 [32]. Huang et al. used tungstophosphoric acid immobilized

on ionic liquid as a homogeneous catalyst which showed 98% selectivity to TAG using a
high acetic acid/glycerol molar ratio of 10 and toluene as an entrainer under nitrogen flow of
150 ml/h [33].
In general, glycerol acetylation with acetic acid strongly depends on the acidic properties
of the catalysts, but it is equilibrium controlled due to formation of water. A recent study
from our group has shown that the use of toluene as an entrainer allows to remove formed
water immediately by azeotropic distillation and to shift the equilibrium almost completely
towards TAG using ion exchange resins Amberlyst-15 or Amberlyst-70 as catalysts [34].
The aim of this present work is to evaluate the catalytic activity of different HPA
supported on oxides with large surface area under such water-free conditions with respect to
catalyst structure and amount of acid sites. Silica, alumina and silica-alumina supported HPA
(STA, TPA and PMA) catalysts were synthesized, characterized in fresh and spent state and
evaluated in glycerol acetylation reaction. The results were compared with reference
experiments either without catalyst (blank tests) or homogeneously catalyzed (pure HPAs are
dissolved in acetic acid) under same reaction conditions and correlated with physiochemical
properties of the catalysts.
2. Experimental
2.1.Materials
The organic compounds glycerol (Alfa Aesar, Germany, > 99%), acetic acid (Fisher
chemicals, Analytical reagent grade), toluene (Acros Organics, analytical reagent grade) and
the silicotungstic acid (STA, Fluka), tungstophosphoric acid (TPA, Fluka), phosphomolybdic
acid (PMA, Merck) and the supports silica-alumina (SIRAL® 40; SiO2/Al2O3 = 40: 60),
dispersible alumina (Disperal® / SASOL) as precursor for γ-alumina (γ-Al2O3) and silica
(SS61138 Chempur / SiO2) were used as received. Furthermore, trimethylchlorosilane and
hexamethyl disilazane (Aldrich), dodecane (TCI Europe), hexadecane (Aldrich) and pyridine
(ACROS) were used for analytical purposes.
2.2.Catalyst preparation
While the silica and silica-alumina supports were not treated before use, dispersible
alumina Disperal P2® was used as a precursor to prepare mesoporous γ-Al2O3 with tailor-
made properties. In a typical procedure, 70 g Disperal P2® was added to 350 g distilled water
and stirred for 1 h; to this solution, 78 g Triton X 100 template and 30 g aqueous ammonium
acetate solution (33 wt%) were added to form a white gel. This gel was further dried in air at
110 °C and calcined in air at 600 °C for 4 h. The other supports SiO2 and SiO2-Al2O3 were
calcined in air at 500 °C for 5h before impregnation.
The heteropolyacids STA, TPA, and PMA were supported onto the three supports by wet
impregnation method with a nominal HPA loading of 20 wt%. In the typical procedure for
STA/SiO2, an aqueous solution (10 ml) of STA (2 g) and some rinsing water (8 ml) was

added to the silica (8 g) under constant stirring overnight. In total 18 ml of water was used for
synthesis. Excess water was removed using a rotary evaporator. The obtained solid was dried
at 120 °C for 2 h and then calcined at 300 °C with air in a muffle for 4 h to obtain the
powdery catalyst.
2.3.Characterization of catalyst
BET surface area, pore volume and pore diameter of the catalysts were determined from
the nitrogen adsorption/desorption isotherms at -196 °C on a Micromeritics ASAP 2010
instrument. Before the measurement, each sample was evacuated at 200 °C for 4 h. The
average pore diameters were calculated according to the BJH method.
Powder X-ray diffraction (XRD) patterns were measured on a theta/theta diffractometer
(X’Pert Pro from Panalytical, Almelo, Netherlands) with CuKα radiation (λ= 0.015418 nm,
40 kV, 40 mA) and an X’Celerator RTMS detector. The alignment was checked by means of
silicon standard. The data were collected in the 2 theta range from 5-70°. The phase
composition of the samples was determined using the program suite WinXPOW by
STOE&CIE with inclusion of the Powder Diffraction File (PDF) data base of the ICDD
(International Centre of Diffraction Data).
The Raman spectra were recorded using a fiber optical RXN-Spectrometer (Kaiser
Optical System) equipped with a 10-400 mW diode laser for excitation at a wavelength of
785 nm and a Mk II Filtered Probe Head with a non-contact optics. The measurements were
performed with 70 mW. The spectra were recorded with 3 accumulations and 3 s exposure
time.
For surface acidity measurement by means of IR spectroscopy pyridine was used as
probe molecule. The measurements in transmission mode were carried out on a Bruker
Tensor 27 spectrometer equipped with a heatable homemade reaction cell with CaF2 windows
connected to a gas-dosing and evacuation system. The sample powders were pressed into
self-supporting wafers with a diameter of 20 mm and a weight of 50 mg. Before pyridine
adsorption, the samples were pretreated by heating in synthetic air to 300 °C for 10 min,
subsequent cooling to room temperature and evacuation. Pyridine was adsorbed at room
temperature by means of a carrier gas passing through a saturator filled with pyridine until
saturation of sample surface. Then the reaction cell was evacuated to remove physisorbed
pyridine. The desorption of pyridine chemisorbed on acid sites was followed by heating the
sample in vacuum to 300 °C and recording spectra every 50 °C.
For the determination of the composition by means of ICP-OES technique, a Varian 715-
ES ICP-Emission-Spectrometer was used. Approximately 20 mg of the sample was mixed
with 8 ml of aqua regia and 2 ml of hydrofluoric acid. The digestion was performed in a
microwave-assisted sample preparation system "MULTIWAVE" from Anton Paar/Perkin-
Elmer at ~200 °C and ~65 bar pressure. The digested solution was filled up to 100 ml and
analyzed. The data analysis was performed on the Varian 715-ES software "ICP Expert".

Tungsten and phosphor leached into the reaction mixture were analyzed qualitatively
by XRF performed on a BRUKER S2 RANGER instrument.
2.4.Catalyst test
Glycerol acetylation was performed in a three-necked 250 ml glass flask equipped with a
fractionating Vigreux column, a modified Dean-Stark apparatus equipped with thermometer
to measure the temperature of the boiling azeotropic mixture and a reflux condenser as
described in detail elsewhere [34]. The flask was also equipped with a glass tube (with
thermocouple) to monitor directly the temperature of the reaction mixture. The reaction
mixture was heated using an oil bath (oil bath temperature 150 °C; reaction mixture
temperature 105 °C) and both reaction mixture and oil bath were stirred using a magnetic
stirrer (1200 rpm) for intensive mixing and to improve heat transfer. Typically the flask was
charged with 10 g glycerol (0.108 mol) and 39.16 g acetic acid (0.652 mol), representing an
acetic acid to glycerol ratio of 6: 1, 60 g toluene (0.652 mol) and 1 g dodecane (internal
standard to evaluate liquid volume change). After heating the reaction mixture to desired
temperature, 500 mg catalyst was added. This was defined as starting time of the reaction.
During the running test, samples of 100 μl volume were withdrawn periodically. Besides the
regular catalyst tests, homogeneous reaction with dissolved STA, TPA and PMA as well as a
blank test in absence of any catalyst were carried out under similar reaction conditions. The
total catalyst weight used in the latter reactions with pure HPA (0.1 g) was equal to the HPA
load (20 wt%) in the supported catalysts to allow comparison of the results.
With some catalysts, reusability experiments and stop experiments were conducted by
removing catalyst from a regular experiment after a defined reaction time, addition of fresh
glycerol-acetic acid feed and continuation of the run under identical reaction conditions as
before. Such experiments also served as a cross-check of HPA leaching into the bulk liquid
and any contribution of homogeneous catalysis.
2.5.Analytical procedure
For proper quantification of the highly polar compounds, the OH groups of unconverted
glycerol, MAG and DAG were silylated before gas chromatographic analysis. In the typical
procedure, 100 μl of product samples were silylated using hexamethyl disilazane and
trimethylchlorosilane with pyridine as solvent. In addition, hexadecane was always used as an
internal standard for the GC analysis (this means that two internal standards were used to
cover the complete analysis procedure). Thereby prepared product samples in close vial were
kept at 70 °C in a drying oven for 45 min and then analyzed using a gas chromatograph (HP
5890 series II) equipped with a CP-Sil 13 CB column (25 m × 0.32 mm) and FID detector.
The temperature program was as follows: 50 °C for 1 min hold, heating at 20 K/min to 310
°C, 2 min hold at 310 °C. The educts glycerol and acetic acid as well as the silylated products
MAG, DAG and TAG were clearly eluted. Peak assignment and calibration of GC detector
were carried out by means of authentic compounds that were prepared at high purity by

fractionated distillation. For DAG, two isomers were detected in the chromatograms, but
separation was incomplete and individual quantification was impossible. The relative error of
this analytical procedure was below 1%. All reported conversion data refer to glycerol.
3. Results and discussion
3.1.Catalyst characterization
3.1.1. Textural properties
Table 1 represents the BET surface areas, pore volumes and average pore diameters for
the supported HPA catalysts prepared with 20 wt% loading in addition to bare supports. The
data indicate a remarkable decrease in the BET surface area (by 6-35%) as well as in the pore
volume after impregnation due to the blockage of pores of the support with the HPA
components; however, this is no surprise when regarding the high HPA loading of 20 wt%.
The pore diameters range from 7.8-12.0 nm which is typical for mesoporous material.
Interestingly, only a marginal decrease in the average pore diameter was observed after
impregnation, and the data for catalysts with the same support are almost the same.
3.1.2. Raman spectroscopy
Raman spectroscopy is a well-known technique to investigate Keggin structures of HPAs
on the surface of a support. Raman spectra of all fresh supported catalysts and additionally of
pure bulk STA, TPA, PMA and also MoO3 were recorded and compared as shown in Fig. 2.
Pure bulk PMA shows Raman bands at 996 cm-1 and 983 cm-1, which can be assigned to
the stretching vibration of terminal Mo=O, and additionally a less intense band appears at 883
cm-1 for bending vibration for Mo-O-Mo units [35, 36] (Fig. 2a). The spectrum of PMA/SiO2
reveals bands at 995 cm-1 (low intensity), 903 cm-1 (very low intensity), 849 cm-1 (moderate
intensity), 820 cm-1 (high intensity, main band), 776 cm-1 (moderate intensity) and 667 cm-1
(low intensity). Comparing the spectrum of PMA/SiO2 with the bulk MoO3 spectrum, the
bands at 820 and 667 cm-1 might be assigned to orthorhombic α-MoO3 species formed by
decomposition of the Keggin structure PMA. This has been reported also by other authors
[37]. PMA/SiO2 shows additional bands at 903, 849 and 776 cm-1 which can be assigned to
monoclinic β-MoO3 surface species [38]. The spectra of PMA/SiO2-Al2O3 and PMA/Al2O3
show a broad band at 984 cm-1, indicating a high dispersion of PMA on the surface.
The Raman spectrum of bulk STA reveals an intense band at 1000 cm-1, a shoulder at 973
cm-1 (stretching vibration of W=O) and a weak band at 920 cm-1 for the bending vibration of
W-O-W (Fig. 2b). Comparing the spectrum of supported STA/SiO2 with bulk STA,
preservation of the bands at 1000 cm-1 and 979 cm-1 can be recognized. This clearly indicates
that the Keggin structure [SiW12O40]4- of STA is preserved on the surface of SiO2. The
spectrum of STA/SiO2-Al2O3 shows a broad band at 984 cm-1 with additional shoulders at

1002 and 961 cm-1. This broad band might stem from partially deteriorated lacunary species
[SiW11O39]8- [39], formed by interaction of STA with surface acid sites.
Similar to STA, the Raman spectrum of bulk TPA exhibits the intense bands at 1011 and
993 cm-1 for stretching vibrations of W=O and the band at 928 cm-1 for bending vibration of
W–O–W (Fig. 2c). The spectrum of TPA/SiO2 also reveals bands at 1010, 993 and 925 cm-1
indicating that the primary structure of TPA (Keggin [PW12O40]3-) is retained on the surface
of SiO2. The spectrum of TPA/SiO2-Al2O3 shows a broad band at 970 cm-1 indicating the
formation of either lacunary species [PW11O39]-7 or points to well dispersed Keggin species.
[40]. The Raman spectrum of TPA/Al2O3 exhibits similar structure as TPA/SiO2-Al2O3,
represented by a broad band at 970 cm-1, most possibly due to interaction of Keggin TPA
with Al2O3 support.
The presented Raman results reveal that STA put on any support seems to be the most
stable Keggin structure among the selected supports. Regarding the effect of nature of the
support on HPA stability, a ranking in the order SiO2 > SiO2-Al2O3 > Al2O3 can be
established.
3.1.3. XRD
The XRD patterns of bare SiO2 and SiO2 supported HPA catalysts are depicted in Figure
3. The bare SiO2 shows a broad reflection at 24° being typical for amorphous silica, which is
also present in the supported samples (PDF 00-029-0085). STA/SiO2 indicates weak and
broad (halo) reflections at 8°, 22° and 27°, which might point to the high dispersion of STA
over the surface of SiO2. Similarly, the pattern for TPA/SiO2 indicates a low crystallinity due
to high dispersion of the HPA. The related pattern reveals reflections at 10°, 27°, and 35° that
can be assigned to the Keggin structure of H3PW12O40·6H2O (PDF 00-050-0304). In contrast
to the first two samples, PMA/SiO2 shows sharp peaks at 13°, 23°, 25° and 27°, indicating the
presence of MoO3 molybdate phase (PDF 00-089-7112). The MoO3 species most likely were
formed by decomposition of heteropoly anions at elevated temperature. By the way, here we
have a nice conformity with Raman information on MoO3 formation (cf. Fig. 2a). From the
XRD analysis of STA/SiO2 and TPA/SiO2 samples, it can be concluded that heteropoly
anions are preserved at least partly, but at high dispersion, and thus this technique is near to
detection limit to identify the crystalline domains of respective Keggin structures. Both the -
alumina and silica-alumina supported catalysts exhibit only broad support reflections,
therefore, the XRD patterns are not shown here. None of the reflections can be assigned to
the expected HPA structures.
3.1.4. FTIR spectra of adsorbed pyridine
The acidic properties of bare supports and the STA, TPA and PMA containing catalysts
using SiO2, SiO2-Al2O3 and Al2O3 as supports were investigated by recording the FTIR
spectra of adsorbed pyridine, as shown in Fig. 4. Generally, pyridine is coordinatively
adsorbed on Lewis acid sites (L-Py, LS) showing bands at 1445-1460 cm-1 and 1610-1620

cm-1, while Brønsted acid sites (PyH+, BS) are characterized by bands at 1540-1548 cm-1 and
1635-1640 cm-1 [41, 42]. The typical bands at 1445-1460 cm-1 for LS (L-Py) and 1540-1548
cm-1 for BS (PyH+) were used to calculate the acid site concentration from the band
intensities and the absorption coefficients given in [43].
Table 2 represents the acidic properties in terms of band intensities and concentration of
BS and LS. The bare silica support did not contribute to any surface acidity. As expected
from Raman results, STA and TPA supported on silica showed higher BS concentration than
Lewis acidity, as the Keggin structures seem to be preserved. It also has to be pointed out that
STA possesses one proton more than the other HPAs. On the other side, PMA/SiO2 shows
lower BS than LS concentration (Fig. 4a) as thermal decomposition of PMA into MoO3
species was observed in XRD and Raman spectroscopic studies. It is evident that the
Brønsted acidity of the catalysts on this support correlates well with the stability of the HPAs
in the order STA > TPA > PMA.
In contrast to the pure SiO2 support, the acid site concentration of bare Al2O3 support
reveals very strong Lewis acidity without any Brønsted acidity (Fig. 4b). The samples
STA/Al2O3 and TPA/Al2O3 show less Brønsted acidity than PMA/Al2O3 might be due to
stronger interaction with the support. On the other hand, PMA/Al2O3 exhibits strong Brønsted
acidity compared to STA/Al2O3 and TPA/Al2O3 catalysts, most likely due to formed MoO3
species which show Brønsted acidic sites.
Interestingly, the acid site concentration of bare silica-alumina support reveals high
Lewis acidity and a moderate contribution from BS (Fig. 4c). Surprisingly, the SiO2-Al2O3
supported HPAs showed the opposite trend in the concentration of BS as compared to SiO2
supported samples: STA < TPA < PMA. At the same time, the concentration of LS stays
rather unaffected. This indicates that none of the tungsten containing HPAs seems to
decompose with simultaneous formation of new LS as observed for the silica supported
samples. A possible explanation for the observed trend in Brønsted acidity might be that this
SiO2-Al2O3 support offers plenty of reactive sites and a considerable amount of BS from
HPA might interact with those via dehydration or condensation during calcination. It is also
known that LS can be transformed into BS in the presence of water at elevated temperatures
[44].
3.2.Catalytic test runs
Glycerol acetylation is a consecutive reaction with stepwise formation of MAG, DAG
and TAG together with their respective isomers along with water as an unavoidable side-
product. In the presence of water, each reaction step is controlled by chemical equilibrium.
The equilibrium can be shifted towards the products by adding acetic acid in excess (to
operate at high glycerol conversion) or by removing one of the products (preferably water)
permanently during the reaction, e.g. by means of azeotropic distillation using an entrainer
like toluene. In a recent study with ion exchange resins as catalysts at such reaction
conditions, we could maintain a very low stationary water concentration of 0.05% which was

shown to have a significant positive impact on reaction rates, selectivities and catalyst
stability [34]. We reported that without removal of water during reaction under identical
conditions, the selectivity to triacetin was significantly low.
Originally, the reaction mixture was biphasic due to the limited miscibility of polar
reactants in toluene, but starts to boil at 100 °C, and conversion of glycerol to MAG takes
place and at the same time the reaction mixture becomes homogeneous. As the reaction
proceeds, the temperature in the reaction mixture and on the top of the modified Dean Stark
apparatus increased. The temperature inside the flask remained at 105 °C for a long time and
finally reached 110 °C (boiling point of toluene) after desired product (TAG) was formed in
high quantity. Toluene is able to form an azeotropic mixture with water at 84 ºC as well as
with acetic acid at 104 ºC [45], and hence, depending on the composition of the reaction
mixture, two azeotropic mixtures may form during glycerol acetylation reaction. This
explains the observed course of the reaction temperature.
For comparison, the reaction was initially performed in the absence of catalyst (blank
test; molar ratio of acetic acid: glycerol = 6, temperature = 105 °C, 4 h). The glycerol
conversion reached 52% after 4 h and the selectivities for MAG and DAG were 85% and
15%, respectively, without formation of TAG. The course of glycerol conversion and product
selectivities is given in Table 3, which follows the well-known behavior of a consecutive
reaction.
Fig. 5 gives an overview of glycerol acetylation using supported STA, TPA and PMA
catalysts, bare supports SiO2, SiO2-Al2O3 and Al2O3 and bulk STA, TPA and PMA solids.
The results of the catalytic runs after 4 hours of reaction time can be clearly separated into
two groups according to the nature of the solids. The catalysts STA/SiO2 and TPA/SiO2 gave
almost 100% glycerol conversion with a maximum selectivity of 70 and 75% to DAG and
11% and 7% selectivity to TAG, respectively. Surprisingly, the results obtained with the
corresponding bulk STA and TPA solids showed almost similar results despite their much
lower specific BET surface area, with almost 100% glycerol conversion after 4 h and
selectivity to TAG of 12.4% and 16%, respectively. On the other hand, all catalysts using
either SiO2-Al2O3 or Al2O3 as support or PMA as active compound showed poor performance
compared to tungsten-containing materials. In the related runs, the glycerol conversion never
exceeded 65% and the product selectivities (MAG: 80-85%, DAG: 12-18%, TAG: < 3%)
were comparable to the reaction without catalyst or with bare supports SiO2, SiO2-Al2O3 or
Al2O3 (Fig. 5).
It is known that bulk TPA and STA are more stable than PMA and these HPAs also
follow the same order of their acid strength [46], which might explain the observed high
activity of tungsten containing bulk HPAs. Although TPA has higher acid strength, the
activity of SiO2 supported STA was found to be slightly higher, which can be explained by
the hydrolytic stability of STA (more resistant to water) and the higher total number of

protons [46, 47]. In the case of PMA, the obtained conversion was not higher than in the
blank test, this may be due to the lower acid strength and thermal stability [48].
The high catalytic activity of STA/SiO2 and TPA/SiO2 compared to all other supported
HPA catalysts used in the present study can be correlated first of all to the high Brønsted
acidity (see Table 2) with simultaneous retention of HPA structure on silica support as seen
in FTIR, XRD and Raman analyses. SiO2-Al2O3 supported HPAs showed high Brønsted as
well as Lewis acidity, whereas Al2O3 supported catalysts revealed high Lewis acidity with
very low Brønsted acidity and silica supported catalysts (except PMA/SiO2) showed high
Brønsted acidity with very low Lewis acidity. In the latter case, the limited stability of PMA
is the decisive factor and makes it a poor catalyst for acetylation reaction. The two best
performing heterogeneous catalysts STA/SiO2 and TPA/SiO2 have by far the lowest number
of Lewis sites. A high Lewis acidity does in no case allow high conversion even when the
number of Brønsted sites is high (as found in STA/SiO2-Al2O3). It can be clearly stated that if
one of these features (Brønsted acidity, structural integrity of HPA, low Lewis acidity) is
changed, neither active nor selective catalyst is obtained.
On the other side, it is somewhat surprising that the much higher specific BET surface
area as measured for supported catalysts has no measurable effect on conversion (and product
distribution) as compared to parent bulk HPAs. This clearly indicates that activity and
selectivity (product formation) seems to strongly depend on acid strength of the catalysts. To
elucidate the leaching of active components from the tungsten-containing catalysts
contributing to the possible homogeneous catalyzed reaction, separate experiments were
carried out and discussed below.
3.3.Reaction progress monitoring
STA/SiO2 and TPA/SiO2 (besides the corresponding bulk HPA) showed highest catalytic
activity compared to all other supported catalysts and supports. Hence, the performance of
these catalysts was monitored over elongated run times (up to 24 h) to study the change in
product selectivity toward TAG under the same reaction conditions as described above (Fig.
6a and b). As expected for a consecutive product formation, with longer run time an increase
in TAG selectivity was observed at the cost of MAG and DAG presence.
The product composition follows consequently the behavior of a consecutive reaction.
Glycerol conversion reached more than 95% within the first 2 h of reaction time. The
maximum selectivities of 36% and 48% for MAG were obtained with both catalysts within 1
h. The slightly better catalytic performance of STA/SiO2 compared to TPA/SiO2 is evident
from the time to reach the maximum selectivity to DAG (STA/SiO2: 75% after 4 h;
TPA/SiO2: 71% after 6 h). After 24 h, the selectivity to TAG was 71% with STA/SiO2 and
67% with TPA/SiO2.
As the reaction proceeds, the by-product diglycerol tetraacetate (DGTA) was detected.
We have reported the formation of DGTA already in studies with ion exchange resins as

catalysts [34]. The formation of this by-product seems to be a special feature of glycerol
acetylation under water-free conditions and high conversion. The DGTA selectivity after 24 h
was 11.9% and 11.1% with STA/SiO2 and TPA/SiO2, respectively. DGTA is expected to
form most likely by oligomerization of MAG and/or DAG as we never observed diglycerol
(oligomerized product) in GC analysis. However, it is well known that oligomerization of
glycerol takes place at high temperature in the presence of strong acid catalyst [11]. On the
other hand, formation of DGTA is significant when the concentration of DAG is high (e.g.
after 4 h), which indicates that DGTA is the oligomer of DAG (see Fig. 5). Formation of
DGTA was confirmed by GC-MS using authentic compound. Interestingly, except DGTA
any other by-product (e.g. dehydrated product acrolein) was not observed.
3.4.Reusability, stability and homogeneous catalyst tests
Reusability and stability of the catalysts STA/SiO2 and TPA/SiO2 was tested by filtering
off the catalyst after 10 h of reaction time and re-using it in another cycle under same reaction
conditions (Fig. 7). The results from the 1st catalytic run with both the catalysts are very
similar to the results obtained after 10 h in the runs described above (see Fig. 5), which
proves that results are reproducible. In both runs, marginal decreases in glycerol conversion
were observed for STA/SiO2 (100% to 97%) and TPA/SiO2 (100% to 94%), however, the
high selectivity to TAG was not maintained during the recycle for both the catalysts. There
was a considerable drop in TAG selectivity with STA/SiO2 (35.8% to 8%) and TPA/SiO2
(22.4% to 5.7%), indicating a lower overall turnover number in esterification. Thus, the
selectivity toward TAG serves as a measure for the catalyst performance rather than the
initial glycerol conversion. It is important to note that, formation of DGTA was not observed
in the reusable test with both catalysts. In addition, the turnover number (TON) for all the
products obtained over these two catalysts was calculated considering Bronsted acid sites and
representing an individual esterification step. After 1st catalytic run with STA/SiO2, TON for
MAG, DAG, TAG and DGTA were 8, 437, 281 and 60 and after recycle test TON changed
to 65, 502, 220 and 0, respectively. Similar for TPA/SiO2 after 1st catalytic run, TON for
MAG, DAG, TAG and DGTA were 70, 982, 314 and 35 and after recycle test 555, 767, 79
and 0.
Table 4 summarizes the results from ICP analyses of fresh and spent catalysts after 2nd
run (STA/SiO2 and TPA/SiO2) and illustrates the quantity of active component (tungsten)
being leached into the reaction mixture leading to a decrease in the catalytic activity. The
tungsten content of the catalysts decreased from 14.7% to 8.2% and 14.4% to 10.4%,
respectively.
In order to clarify whether the leached part still contributes to glycerol acetylation via
homogeneous catalysis, a separate stop experiment was performed with best performing
catalyst STA/SiO2 under identical conditions as depicted in Fig. 8. The catalyst was filtered
off from the reaction mixture after 4 h when the selectivities for MAG, DAG and TAG were
12.9%, 75.7% and 11.4%, respectively, at almost complete conversion of glycerol.

After replenishing the consumed reactants (19.56 g of acetic acid, 5 g of glycerol, molar
ratio = 6), the reaction was continued for another 4 h under the same conditions (except a
slight increase in liquid volume) but in absence of the solid catalyst. The conversion of
glycerol over the same time period of 4 h was slightly lower in the second stage as well as the
selectivity to DAG. The selectivity to MAG was somehow higher but selectivity to TAG
didn’t increase further. Interestingly, the resulting conversion and selectivities (for MAG and
DAG) were very similar to those of the blank test (cf. Table 3) showing a continuously
increasing glycerol conversion; the initial very high selectivity to MAG decreased with time,
the consecutive product DAG appears and TAG formation was never seen. May be leached
species were disintegrated products of the active STA supported on silica, but did not longer
exist as a Keggin anion structure. Literature reports on the interaction of a silica surface
(silanols) and deposited TPA give some proofs on the formation of such surface anions [49].
Moreover, liquid phase reactions in acidic media lead to further decay to lacunary species
[50]. However, the species formed by interaction of intact STA with silanols and the leached
species were not able to catalyze the acetylation reaction efficiently (in contrast to the
corresponding intact bulk HPA). Therefore, most likely the leached active species do not act
as homogeneous catalyst. This means that the glycerol acetylation under given condition
seems to be running on solid catalyst surfaces, i.e. intact HPA structure. It still has to be
clarified whether the catalyst deactivation is due to simple loss of active species or if other
effects like deposition of high-molecular oligomers (possibly formed in a similar way as
DGTA) play a role.
4. Conclusions
The combination of various HPAs and common support materials led only to few
supported catalysts with remarkable performance that were able to produce triacetin (TAG) at
high yield (71%) within acceptable time periods (24 h). Such catalysts were exclusively
composed of tungsten-containing HPAs supported on silica. To achieve good performance,
high Brønsted acidity is required, which seems to be linked first of all to the preservation of
the Keggin structure of the HPAs with its high number of reactive protons. Other
prerequisites to get effective catalysts of this type are the thermal stability of HPAs and a
tuned strength of their interaction with the support. This is ensured if supports like silica
without any Brønsted and Lewis sites are used and the deposited amount of HPA is rather
high, otherwise, surface reactions of silanol groups with Keggin units lead to undesired
surface species. Beyond that, it seems to be detrimental to use supports that possess
remarkable intrinsic acidity, independent whether this is of Brønsted or Lewis type. This is
typical for the alumina containing supports in this study. In such case, the acid sites of HPA
and support may react with each other partially during the calcination at elevated
temperature.
Some of these features of the fresh catalysts (low bonding strength between intact
heteropolyacid crystallites and supports and/or disintegration of Keggin units due to surface

reactions with silanol groups) in combination with the high polarity and hydrophilicity of
unreacted glycerol and acetic acid are most likely responsible for the partial leaching of
active component into the liquid phase and catalyst deactivation. Surprisingly, such leachate
does not much contribute to catalytic activity in esterification. This might be due to the use of
the entrainer toluene to establish water-free conditions, which suppresses the formation of
Brønsted acids away from the solid supported catalysts. The use of an entrainer is definitely
beneficial for the overall process as the chemical equilibrium of the esterification is shifted
toward the target product triacetin. To make the catalysts more stable, a proper balance has to
be found between the described intra- and intercrystallite interactions to preserve the HPAs
structure and simultaneously to strengthen the bonding to the support to avoid leaching. A
principal drawback of the applied reaction system is the missing option to adjust the reaction
temperature; this is solely controlled by the boiling point of the reaction mixture, which itself
changes during the progress of the reaction. To overcome this limitation for further
optimization, a simultaneous control of reaction pressure would be necessary.
Acknowledgements
The authors would like to thank Mrs. A. Simmula for ICP analyses. The authors gratefully
acknowledge the financial support by DLR within the frame of the BMBF-CSIR Cooperative
Science Program (project number IND 11/045).

Figure captions
Figure 1. Reaction scheme for consecutive glycerol acetylation.
Figure 2. Raman spectra of a) STA/SiO2, STA/SiO2-Al2O3 and STA/Al2O3 and, b) TPA/SiO2,
TPA/SiO2-Al2O3 and TPA/Al2O3 and c) PMA/SiO2, PMA/SiO2-Al2O3, PMA/Al2O3 and
reference materials (bulk HPA and MoO3).
Figure 3. X-Ray diffractograms of used SiO2, TPA/SiO2, STA/SiO2 and PMA/SiO2.
Figure 4. FTIR spectra of adsorbed pyridine on a) SiO2, TPA/SiO2, STA/SiO2 and PMA/SiO2
b) Al2O3, TPA/Al2O3, STA/Al2O3 and PMA/Al2O3 and c) SiO2-Al2O3, TPA/SiO2-Al2O3,
STA/SiO2-Al2O3, PMA/SiO2-Al2O3 at 150 °C
Figure 5. Glycerol esterification with acetic acid and toluene as an entrainer (glycerol = 10 g,
acetic acid = 39.16 g, molar ratio acetic acid: glycerol = 6: 1, toluene = 60 g, 105 °C, 4 h;
catalyst mass = 0.5 g (supported) or 0.1 g (bulk HPA)).
Figure 6. Glycerol conversion and selectivities to products over a) STA/SiO2 and b)
TPA/SiO2 during 24 h reaction (glycerol = 10 g, acetic acid = 39.16 g, molar ratio acetic acid:
glycerol = 6: 1, toluene = 60 g, catalyst = 0.5 g, 105 °C).
Figure 7. Reusability results on glycerol acetylation in the presence of toluene as an entrainer
over STA/SiO2 and TPA/SiO2 (glycerol = 10 g, acetic acid = 39.16 g, molar ratio acetic acid:
glycerol = 6: 1, toluene = 60 g, catalyst = 0.5 g, 105 °C, time = 10 h).
Figure 8. Glycerol conversion and selectivities with 0.5 g of STA/SiO2 in stop-experiment
(reaction condition same as Fig. 6). The dotted lines represent the time period for catalyst
removal and addition of fresh feed.

Table 1: Textural properties of supports and catalysts
Catalyst
ABET
Deviation relative
to support (%)
Pore volume
(cm³/g)
Average pore
diameter (nm)
(m²/g)
SiO2
244
206
206
229
-
0.84
0.66
0.62
0.76
12
STA/SiO2
TPA/SiO2
PMA/SiO2
-15.6
-15.6
-6.1
11.9
11.7
11.9
Al2O3
303
228
267
228
-
1.32
0.68
0.84
0.79
11.4
12.1
11.8
11.8
STA/Al2O3
TPA/Al2O3
PMA/Al2O3
-24.7
-11.9
-24.7
SiO2-Al2O3
520
350
341
377
-
1.00
0.73
0.67
0.81
8.4
8.4
7.4
7.8
STA/SiO2-Al2O3
TPA/SiO2-Al2O3
PMA/SiO2-Al2O3
-32.7
-34.4
-27.5

Table 2: Acidic properties of pure carriers and supported heteropolyacids
Catalyst
Intensity (150 °C) Brønsted Intensity (150 °C) L- Lewis acidity
PyH+ (1545 cm-1)
acidity
Py (1545 cm-1)
(μmol/g)
(μmol/g)
SiO2
0
0
0
0
STA/SiO2
TPA/SiO2
PMA/SiO2
3.7
2.1
1.1
274.5
154.9
82.0
0.4
0.6
1.9
24.3
31.1
105.8
Al2O3
0.0
0.5
0.5
0.9
0.0
3.0
5.0
5.3
3.0
171.5
280.7
297.7
168.7
STA/Al2O3
TPA/Al2O3
PMA/Al2O3
40.6
38.4
65.4
SiO2-Al2O3
1.1
1.9
2.0
3.0
82.0
145.9
150.4
222.6
4.2
4.9
5.1
4.1
237.2
278.5
289.8
231.5
STA/SiO2-Al2O3
TPA/SiO2-Al2O3
PMA/SiO2-Al2O3

Table 3. Glycerol conversion and acetins formation in blank test run
Time (h)
Glycerol
Selectivity
Selectivity
Selectivity
conversion (%)
to MAG (%)
to DAG (%)
to TAG (%)
1
2
3
4
28
31
44
52
99
92
88
85
1
8
0
0
0
0
12
15
(glycerol = 10 g, acetic acid = 39.16 g, molar ratio acetic acid: glycerol = 6, toluene = 60 g,
105 °C, time = 4 h)

Table 4. ICP-OES analysis of the fresh and spent catalysts STA/SiO2 and TPA/SiO2.
Catalyst
Element load, wt%
W
P
-
#
STA/SiO2 Fresh
14.7
8.2
Spent
TPA/SiO2 Fresh
Spent
-
14.4
10.4
0.2
0.13
#: Due to the silica support used, amount of leaching of Si from STA was not estimated

Figure 1

820
a
820
849
995
776
996
995
667
667
983
PMA/SiO2
903
MoO3
883
PMA pure
PMA/SiO2-Al2O3
PMA/Al2O3
953
953
1000
900
800
700
Raman shift / cm-1
b
1000
1000
972
979
STA pure
STA/SiO2
920
984
STA/SiO2-Al2O3
STA/Al2O3
961
1002
985
1050
1000
950
900
Raman shift / cm-1
c
1011
993
1010
TPA pure
TPA/SiO2
993
TPA/SiO2-Al2O3
970
970
TPA/Al2O3
1050
1000
950
900
Raman shift / cm-1
Figure 2

1
1- MoO3
2- H4SiW12O40. 14H12O
1
3- H3PW12O40. 14H2O
1
3
1
1
1
1
2
PMA/SiO2
STA/SiO2
2
2
3
3
3
3
TPA/SiO2
SiO2
10
20
30
40
50
60
70
2/
Figure 3

a
1488
1610
1539
1638
0.1
STA/SiO2
TPA/SiO2
1445
PMA/SiO2
SiO2
1700
1600
1500
Wavenumbers (cm-1)
1448
b
0.1
1616
1487
1540
STA/Al2O3
TPA/Al2O3
PMA/Al2O3
Al2O3
1700
1600
1500
Wavenumbers (cm-1)
1451
1494
c
1618
1618
1617
1613
0.1
1541
1619
PMA/SiO2-Al2O3
TPA/SiO2-Al2O3
STA/SiO2-Al2O3
1544
SiO2-Al2O3
1700
1600
1500
Wavenumbers (cm-1)
Figure 4

MAG
DAG
TAG
Conversion
100
90
80
70
60
50
40
30
20
10
0
Figure 5

a)
100
80
60
40
20
0
% (Conversion)
% (MAG)
% (DAG)
% (TAG)
% (DGTA)
0
5
10
15
20
25
Time / h
b)
100
80
60
40
20
0
% (Conversion)
% (MAG)
% (DAG)
% (TAG)
% (DGTA)
0
5
10
15
20
25
Time / h
Figure 6

MAG
DAG
TAG
DGTA
Conversion
100
90
80
70
60
50
40
30
20
10
0
STA/SiO2 1st Run STA/SiO2 2nd Run
TPA/SiO2 1st Run TPA/SiO2 2nd Run
Figure 7

100
80
60
40
20
0
% (Conversion)
% (MAG)
% (DAG)
% (TAG)
0
1
2
3
4
5
6
7
8
9
10
Time / h
Figure 8