Ionic Liquid–Silicotungstic Acid Composites as Efficient and Recyclable Catalysts for the Selective Esterification of Glycerol with Lauric Acid to Monolaurin

Article abstract of DOI:10.1002/cctc.201600788

The synthesis of glycerol monolaurate (GML) by the esterification of glycerol (GL) with lauric acid (LA) over a series of propyl sulfonic acid-functionalized ionic liquids (SAFILs)-modified silicotungstic acid (STA; H₄SiW₁₂O₄₀) composite catalysts has been investigated. The synthesized organic–inorganic hybrid catalysts were characterized by different physicochemical techniques. In particular, their acidic properties were studied by solid-state ³¹P magic angle spinning (MAS) NMR spectroscopy by using adsorbed trimethylphosphine oxide (TMPO) as a probe. The effects of key reaction parameters, such as glycerol/lauric acid molar ratio, amount of catalyst, reaction time, and reaction temperature on LA conversion and GML product yield were elucidated and optimized with response surface methodology (RSM). The N,N-dimethyl(benzyl)ammonium propyl sulfobetaine (DMBPS)-modified STA [DMBPSH]H₃SiW₁₂O₄₀ exhibited the optimal catalytic activity and was exploited for process optimization. A highest GML yield of 79.1 % was achieved with the optimized reaction conditions. The high catalytic activity of these hybrid catalysts were attributed to strong acidity, low transport resistance, and their “pseudoliquid” characteristics. A kinetic study was made based on a second-order irreversible model of the esterification reaction, which resulted in an activation energy of 39.49 kJ mol^?1 for [DMBPSH]H₃SiW₁₂O₄₀ under optimized reaction conditions.

Full text of DOI:10.1002/cctc.201600788

DOI: 10.1002/cctc.201600788
Full Papers
Ionic Liquid–Silicotungstic Acid Composites as Efficient
and Recyclable Catalysts for the Selective Esterification of
Glycerol with Lauric Acid to Monolaurin
Xiaoxiang Han,*[a] Xiaofang Zhang,[a] Guangqi Zhu,[a] Juanjuan Liang,[a] Xianghui Cao,[a]
Renjun Kan,[a] Chin-Te Hung,[b] Li-Li Liu,[b] and Shang-Bin Liu*[b, c]
The synthesis of glycerol monolaurate (GML) by the esterifica-
tion of glycerol (GL) with lauric acid (LA) over a series of propyl
sulfonic acid-functionalized ionic liquids (SAFILs)-modified sili-
cotungstic acid (STA; H4SiW12O40) composite catalysts has been
investigated. The synthesized organic–inorganic hybrid cata-
lysts were characterized by different physicochemical tech-
niques. In particular, their acidic properties were studied by
solid-state 31P magic angle spinning (MAS) NMR spectroscopy
by using adsorbed trimethylphosphine oxide (TMPO) as
a probe. The effects of key reaction parameters, such as glycer-
ol/lauric acid molar ratio, amount of catalyst, reaction time,
and reaction temperature on LA conversion and GML product
yield were elucidated and optimized with response surface
methodology (RSM). The N,N-dimethyl(benzyl)ammonium
propyl sulfobetaine (DMBPS)-modified STA
[DMBPSH]H3SiW12O40 exhibited the optimal catalytic activity
and was exploited for process optimization. A highest GML
yield of 79.1% was achieved with the optimized reaction con-
ditions. The high catalytic activity of these hybrid catalysts
were attributed to strong acidity, low transport resistance, and
their “pseudoliquid” characteristics. A kinetic study was made
based on a second-order irreversible model of the esterifica-
tion reaction, which resulted in an activation energy of
39.49 kJmolꢀ1 for [DMBPSH]H3SiW12O40 under optimized reac-
tion conditions.
Introduction
As the primary surplus byproduct of biodiesel production, glyc-
erol (GL; also called glycerine), has glutted the global market
and has seen the development of relevant industries, such as
oleo chemical and soap and other cleaning compound manu-
facturing.[1] Despite the fact that GL is a promising low-cost
raw material for the productions of food additives, surfactants,
soap, medicines, and cosmetics,[2] its utilization towards more
value-added chemicals remains a challenging and demanding
task.[3–6] On the other hand, glycerol monolaurate (GML; also
known as glyceryl laurate or monolaurin) has been widely used
as emulsifiers or an antibacterial agent in the food, pharma-
ceutical, and cosmetic industries.[7] GML is also one of the gen-
erally recognized as safe (GRAS) food additives certified by the
US Federal Drug Administration (FDA) and is commonly syn-
thesized by the direct esterification of GL with lauric acid (LA;
or dodecanoic acid) over acidic catalysts.
Conventionally, homogeneous catalysts, such as H2SO4, HCl,
H3PO4, p-toluenesulfonic acid, or acidic ionic liquids, are com-
monly exploited for esterification reactions.[8,9] However, the
applications of these mineral acids are hampered by severe en-
vironmental and recovery cost issues. Their inherent shortcom-
ings include their highly corrosive, toxic, non-dissociable, intri-
cate separation and purification, making recovery/reuse rather
impractical.[10] To unravel these problems, heterogeneous acid
catalysts, such as heteropolyacids (HPAs),[2,11,12] zeolites,[13]
organo-montmorillonites,[14] metal oxides,[15,16] organic/inorgan-
ic acid-modified mesoporous silicas,[17–19] ion-exchanged
resins,[20] and enzymes,[21] have been employed for esterifica-
tion reactions. Nevertheless, these solid acid catalysts normally
have the drawbacks of high energy consumption, being vul-
nerable to deactivation, needing longer reaction times, requir-
ing formidable separation and recovery, and so on.[9,10] Hence,
it is highly desirable to seek alternative environmentally
benign and cost-effective acid catalysts that are not only
highly active for esterification/transesterification reactions, but
also possess durability and recyclability. In this context, hybridi-
zation of organic acids (such as ionic liquids; ILs) with inorgan-
ic acids (such as heteropolyacids; HPAs) has been shown to be
a justifiable tactic.[10,22–25]
[a] Dr. X. Han, X. Zhang, G. Zhu, J. Liang, X. Cao, R. Kan
Department of Applied Chemistry
Zhejiang Gongshang University
Hangzhou 310035 (China)
E-mail: hxx74@126.com
[b] Dr. C.-T. Hung, L.-L. Liu, Prof. Dr. S.-B. Liu
Institute of Atom and Molecular Sciences
Academia Sinica
P.O. Box 23-166
Taipei 10617 (Taiwan)
[c] Prof. Dr. S.-B. Liu
Department of Chemistry
National Taiwan Normal University
Taipei 11677 (Taiwan)
Supporting information for this article can be found under:
http://dx.doi.org/10.1002/cctc.201600788.
This manuscript is part of a Special Issue on the “Catalytic Conversion of
Biomass”. A link to the Table of Contents will appear here once the Spe-
cial Issue is assembled.
ChemCatChem 2016, 8, 1 – 13
1
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Being considered as environmentally friendly solvents and
catalysts, ILs have received considerable R&D attention in vari-
ous fields. For example, ILs have been extensively used in
chemical synthesis owing to their unique characteristics, such
as low melting points, excellent thermal stability, negligible
volatility, remarkable solubility, and adjustable physical and
chemical properties, recyclability, and reusability.[26–29] ILs have
also been widely exploited as homogeneous catalysts, espe-
cially for the esterification reaction system owing to their
unique self-separation properties,[30] forming a liquid–liquid bi-
phase after the reaction to facilitate separation and recycling
of the catalyst. Nevertheless, IL catalyst systems still suffer from
high costs and cumbersome water removal procedures.[10] Re-
cently, these issues have been largely resolved by incorporat-
ing ILs with HPA superacids to form organic–inorganic compo-
sites and a variety of such IL-based HPA catalysts have been
utilized for esterification reactions and conversions of bio-
mass.[10,22–25] It is noteworthy that, by taking the advantages
from the respective organic (IL) and inorganic (HPA) constitu-
ents, these novel IL–HPA catalysts have tunable acidity and are
mostly water soluble, hence, most suitable for esterification/
transesterification reactions. Moreover, depending on the reac-
tion system, the IL–HPA catalysts may be exploited for hetero-
geneous as well as homogeneous reactions. For the latter, re-
action-induced self-separation characteristics normally prevail,
thus, facilitating product separation and catalyst recycling.
Herein, we report the syntheses of a series of propyl sulfonic
acid-functionalized ionic liquid (SAFIL)-modified silicotungstic
acid (STA; H4SiW12O40) composite catalysts. These novel organ-
ic–inorganic hybrid catalysts were thoroughly characterized by
a variety of different analytical and spectroscopic techniques.
The catalytic performances of these SAFIL–STA composite salts
for the esterification of glycerol (GL) with lauric acid (LA) to
glycerol monolaurate (GML) were investigated and compared
with other solid acid catalysts. Moreover, the effects of key re-
action variables, such as GL/LA molar ratio, catalyst amount,
and reaction time and temperature, on GML product yield and
selectivity were elucidated and optimized with response sur-
face methodology (RSM) based on a Box–Behnken design
(BBD).[31] A kinetic study of the esterification reaction was also
established and evaluated under the optimized conditions.
tively.[32,33] The broad absorption band centered at around
3438 cmꢀ1 should be associated with the stretching vibrations
of OꢀH species. Likewise, similar absorption bands were ob-
served for the DMBPS-modified STA composites in spite of
marginal variations in peak intensities and positions and addi-
tional bands in the 1100–1500 cmꢀ1 region. In particular, the
band at approximately 1472 cmꢀ1 may be ascribed to bending
vibrations of -CH2 and -CH3 groups, whereas the bands at ap-
proximately 1224 and 1170 cmꢀ1 are due to asymmetric and
symmetric stretching vibrations of S=O arising from the sulfon-
ic groups of the DMBPS IL. For the latter, an extra absorption
peak at approximately 2966 cmꢀ1 was observed, which may be
assigned to the stretching vibrations of -CH2 species. The
above results verify the presence of sulfonic groups, hence,
the successful anchoring of the DMBPS IL onto the STA. More-
over, the Keggin structure of the pristine STA remained practi-
cally unchanged even after linking with the SAFIL.
The surface properties of the SAFIL–STA composites were
also verified by X-ray photoelectron spectroscopy (XPS), as il-
lustrated in Figure 1a for the [DMBPSH]H3SiW12O40 catalyst. The
peaks with the binding energies (BE) of approximately 530.9,
401.9, and 166.7 eV were due to the O1s, N1s, and S2s core
levels, respectively. The presence of the N1s signal clearly indi-
cates the presence of nitrogen species (-N+(CH3)2CH2-) in the
ammonium sulfobetaine, confirming the successful incorpora-
tion of DMBPS onto the STA.[34] In addition, the high-resolution
W4f spectrum (Figure 1b) revealed the presence of a well-re-
solved spin-orbit doublet corresponding to W4f5/2 (BE=
38.0 eV) and W4f7/2 (35.8 eV) core levels, respectively. The
value of BE difference (2.2 eV) so observed is typical for WVI.[35]
It is noteworthy that a similar W4f spectrum was observed for
pure STA, which revealed two prominent peaks at 37.7 and
35.7 eV.[36] The higher binding energies observed for the W4f
peaks in the DMBPS-modified STA may be attributed to switch-
ing of the W=O bonds to weaker WꢀO bonds upon incorporat-
ing the SAFIL.[12,36] Likewise, the peaks with binding energies at
101.9 and 102.4 eV may be ascribed to the Si2p core level of
STA.[37] The broad overlapping signal observed for the C1s core
level may be deconvoluted into three bands with BE of 286.8,
285.9, and 284.6 eV (Figure 1d), corresponding to the presence
of CꢀN, CꢀS, and CꢀC bonds, respectively.[34] Again, the above
results indicate that the preservation of the Keggin structure
may be inferred for the modified organic STA salt.
Results and Discussion
The structural integrity of various [DMBPSH]xH4ꢀxSiW12O40
(x=0–4) samples were further examined by XRD, as shown in
Figure S2 (in the Supporting Information). Unlike the profile of
DMBPS (Figure S2b), which showed diffraction peaks mostly in
the 2q range of 5–358, the pristine STA revealed peaks over
a much wider range with three main peaks at 10.3, 25.3, and
34.6, which may account for the secondary structure of STA
formed by the body centered cubic (bcc) packing of the pri-
mary structure (i.e., the Keggin SiW12O404ꢀ).[37] Upon incorporat-
ing the DMBPS IL, XRD patterns similar to that of the pristine
STA were observed (Figure S2c–f), except for the marginal shift
of the diffraction peaks toward higher 2q values and slight de-
crease in peak intensities. The presence of DMBPS on STA in-
duced an additional broad peak in the 2q region 5.0–10.0, the
Catalyst characterization
For elucidation of the chemical composition and configuration
of the various samples, FTIR spectra of the parent STA
(H4SiW12O40) and various DMBPS-modified STA catalysts,
namely [DMBPSH]xH4ꢀxSiW12O40 (x=1–4), are depicted in Fig-
ure S1 of the Supporting Information together with the raw
DMBPS IL. For the pristine STA, four distinct characteristic
4ꢀ
bands corresponding to the Keggin SiW12O40 polyanions
were observed at 978, 925, 881, and 792 cmꢀ1, which may be
assigned to asymmetric stretching vibrations of terminal W=Ot,
4ꢀ
SiꢀO (centering SiO4 tetrahedron), edge-sharing WꢀObꢀW,
6ꢀ
and corner-sharing WꢀOcꢀW of the WO6 octahedra, respec-
&
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 1. XPS survey spectrum (a), and the corresponding high-resolution W4f (b), Si2p (c), and C 1s (d) core level spectra of the [DMBPSH]H3SiW12O40
catalyst.
featured diffraction peaks observed for the pure DMBPS (Fig-
ure S2b) were no longer visible, indicating the formation of
a new organic salt phase with good crystallinity. Nevertheless,
a gradual decrease in crystallinity with increasing DMBPS load-
ing (i.e., increasing x) may be inferred. The above results,
which are complementary to the FTIR and XPS data, again
verify the successful incorporation of DMBPS IL onto the STA
while retaining the Keggin structure with good crystallinity.[11]
The thermal stability of the organic STA salts was also veri-
fied by means of thermogravimetric/differential thermogravi-
metric (TG-DTG) measurements, as exemplified in Figure S3 (in
the Supporting Information). As shown in Figure S3a, the pris-
tine STA showed three distinct weight-loss peaks at 321, 468,
and 673–805 K, which may be ascribed to desorption of physi-
sorbed water, loss of crystalline water, and collapse of the
Keggin structure, respectively. On the other hand, two weight-
loss peaks (at 575 and 595 K) were observed for DMBPS owing
to decomposition of the organic IL (Figure S3b). Four main
weight-loss peaks were observed for the [DMBPSH]H3SiW12O40
sample (Figure S3c). The weight losses occurred in the temper-
ature range 303–385 and 437 K should be due to desorption
of surface-bound and crystalline water, respectively. Whereas
the weight-loss peak at 538 K should be associated with de-
49.64, 62.38, 68.30, 127.02, 129.25, 130.09, 132.93 ppm.
[DMBPSH]H3SiW12O40: 1H NMR (500 MHz, D2O); d=2.218 (m,
2H), 2.887 (t, J=7.0 Hz, 2H), 2.984 (s, 6H), 3.351 (t, J=7.6 Hz,
2H), 4.455 (s, 2H), 7.469–7.493 ppm (m, 5H); 13C NMR
(125 MHz, D2O); d=18.44, 47.46, 49.71, 62.55, 68.78, 127.02,
129.37, 131.00, 133.06 ppm.
Additional study by N2 physisorption (Figure S4 in the Sup-
porting Information) confirms that the pristine STA, parent
DMBPS IL, and DMBPS-STA composite catalysts all exhibited
the anticipated low surface area characteristics. The pristine
STA gave rise to a Brunauer–Emmett–Teller (BET) surface area
of approximately 3.0 m2 gꢀ1, and the incorporation of DMBPS IL
onto the STA appeared to have little influence on the surface
area of the [DMBPSH]H3SiW12O40 catalyst. In this context, the
surface area of the catalyst appeared to be irrelevant to the
observed catalytic activity. The excellent solubility of the
DMBPS–STA composite catalyst in water and polar solvents as
well as the unique pseudoliquid characteristics of the STA
should be responsible for the excellent catalytic activity and
GML selectivity observed.[22,24]
In
addition,
the
acidic
properties
of
various
[DMBPSH]xH4ꢀxSiW12O40 (x=0–4) catalysts and the pure DMBPS
were characterized by using the solid-state 31P TMPO NMR ap-
proach.[38,39] It is noteworthy that, herein, the 31P chemical shift
(d31P) of the adsorbed trimethylphosphine oxide (TMPO) is
used as a quantitative scale for acidic strength, which is made
possible by the prevailing linear correlation between the ob-
served d31P and the proton affinity (i.e., acidic strength) of
Brønsted acid sites.[38–41] Unlike the phosphotungstic acid (PTA;
H3PW12O40), which exhibited 31P NMR signals for the resonance
of adsorbed TMPO at 31P chemical shifts (d31P) of 92.1, 89.4,
87.7, ꢀ10.9, and ꢀ15.2 ppm,[42] the pristine STA (H4SiW12O40) ex-
hibited only overlapping resonance peaks at 90 and 88 ppm
(Figure 2a). It has been shown[42] that the resonances with
d31P=ꢀ10.9 and ꢀ15.2 ppm, which may also be observed for
composition
of
the
organic
moieties
on
the
[DMBPSH]H3SiW12O40 and the broad feature centering at ap-
proximately 715 K is due to collapse of the Keggin structure.
Again, these results further verify the successful anchoring of
DMBPS on STA and such SAFIL–STA organic salt composites
remain stable above 700 K, a temperature far beyond the reac-
1
tion temperature (typically, <450 K) studied. Moreover, the H
and 13C NMR spectra obtained from the DMBPS and the
[DMBPSH]H3SiW12O40 further confirmed the anticipated chemi-
cal compositions with the following NMR data: DMBPS:
1H NMR (500 MHz, D2O); d=2.213 (m, 2H), 2.866 (t, J=7.5 Hz,
2H), 2.951 (s, 6H), 3.348 (t, J=8.0 Hz, 2H), 4.414 (s, 2H), 7.436–
7.474 ppm (m, 5H); 13C NMR (125 MHz, D2O); d=18.33, 47.39,
3ꢀ
the bare PTA in the absence of TMPO, arise from the PW12O40
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
3
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
tion of STA Keggin units, which is favorable for molecular
transport.[23]
Comparisons of catalytic performance over various solid
acid catalysts
Table 1 displays the catalytic performances of a total of 24 acid
catalysts for esterification of glycerol (GL) with lauric acid (LA).
The catalysts examined include Keggin-type heteropolyacids,
such as PTA (entry 1), phosphomolybdic acid (PMA) (entry 2),
STA (entry 4), and
a series of SAFIL–STAs, denoted as
[XILH]xH4ꢀxSiW12O40 (x=1–4; XIL =DMBPS, DMPPS, TEAPS,
MIMPS, and PPS; entries 5–12). In addition, the catalytic activity
of two [DMBPS]-modified PTA and PMA salts (entries 13 and
14) and partially ion-exchanged (Fe3+, Ag+, Ce3+) STAs (en-
tries 15–17) were also compared together with conventional
solid catalysts, such as zeolite, sulfonic polymer resin, ionic
liquid, and active solid oxides (see Table 1). The reactions were
carried out under the same reaction conditions, that is, GL/
LA=3:1 (mol/mol), catalyst amount (relative to LA) 4 wt%, re-
action time 1 h, and reaction temperature 423 K. As shown in
Table 1, the esterification over the various catalysts yielded
mainly glycerol monolaurate (GML) and glycerol dilaurate
(GDL) as products. The fact that no glycerol trilaurate (GTL) or
other dehydration products of GL were detected in the prod-
uct mixture may be attributed to the presence of water and
the relatively mild reaction conditions applied.[43]
Figure 2. Solid-state 31P MAS NMR spectra of TMPO adsorbed on (a) the
pristine H4SiW12O40, (b) [DMBPSH]H3SiW12O40, (c) [DMBPSH]2H2SiW12O40,
(d) [DMBPSH]3HSiW12O40, (e) [DMBPSH]4SiW12O40, and (f) pure DMBPS. The
asterisks denote spinning sidebands.
Keggin polyanions alone. Conversely, the three resonances
(d31P=92.1, 89.4, and 87.7 ppm), which exceed the chemical
shift threshold for superacidity (86 ppm),[38–41] may be attribut-
ed to TMPO adsorbed on the three Brønsted superacidic
proton (H+) sites (i.e., signals from TMPOH+ complexes). Clear-
ly, the substitution of the Si element for P in PTA creates an
extra protonic acid site as a result of charge balance, meaning
that the Si-substituted STA in fact possesses Brønsted acid sites
with lower acidic strengths compared with PTA. Nevertheless,
the two main resonances (d31P=90 and 88 ppm; Figure 2a)
may be inferred as being due to the presence of superacidic
active sites (d31Pꢁ86 ppm). On the other hand, DMBPS ionic
liquid alone, showed practically no acidity in the presence of
adsorbed TMPO (Figure 2 f); the 31P signals at 42 ppm ob-
served for the DMBPS is mainly due to bulk TMPO.[38,39]
Table 1. Catalytic performances of esterification of glycerol (GL) with
lauric acid (LA) over various solid acid catalysts.[a]
Entry Catalyst
Selectivity [%][b] Conversion GML yield
GML GDL GTL of LA [%]
[%]
1
2
3
4
5
6
7
8
H3PW12O40
H3PMo12O40
DMBPS
81.3 18.7 nil
80.1
45.9
27.9
76.7
88.1
79.2
70.5
62.7
81.5
87.1
78.4
78.7
84.5
55.0
82.6
80.2
84.1
15.9
35.4
17.4
18.4
17.4
20.3
16.5
65.1
43.7
20.1
69.2
78.4
69.7
62.6
56.8
72.3
67.5
70.2
70.5
73.8
52.1
66.1
63.6
71.6
14.7
31.7
16.1
16.6
15.9
18.6
15.3
95.2 4.8 nil
72.1 17.9 nil
90.2 9.8 nil
H4SiW12O40
[DMBPSH]H3SiW12O40 89.0 11.0 nil
[DMBPSH]2H2SiW12O40 88.5 11.5 nil
[DMBPSH]3HSiW12O40 88.7 11.3 nil
[DMBPSH]4SiW12O40
[DMPPSH]H3SiW12O40 88.7 11.3 nil
[TEAPSH]H3SiW12O40 77.5 22.5 nil
[MIMPSH]H3SiW12O40 89.5 10.5 nil
[PPSH]H3SiW12O40 89.6 10.4 nil
[DMBPSH]H2PW12O40 87.3 12.7 nil
[DMBPSH]H2PMo12O40 94.8 5.2 nil
Fe1/3H3SiW12O40
AgH3SiW12O40
Ce1/3H3SiW12O40
ZSM-5
Amberlyst-15
CeMnTiO
Upon incorporating DMBPS IL onto STA, a progressive de-
crease in acidic strength (i.e., 31P value) from beyond 86 ppm
to high-field is readily observed (xꢁ1; Figure 2b–e). Eventually,
as all available proton sites were intercalated with DMBPS (x=
4; Figure 2e), a broad resonance centered at d31P=83 ppm
with lower peak area was observed, indicating an overall de-
crease in both the acidic strength and concentration of acid
sites compared with the pristine STA. Nevertheless, the acidic
strength so observed for the [DMBPSH]4SiW12O40 is still consid-
erably higher than typical solid acid catalysts, such as zeo-
lites.[38,39] Moreover, the additional peak at approximately
67 ppm observed for the organic composite salts with x=1–3
(Figure 2b–d) may be ascribed to the presence of (TMPO)nH+
(n>1) adducts, in which an acid site was associated with more
than one probe molecules.[42] Clearly, the incorporation of
DMBPS tends to provoke an effective averaging of Brønsted
acidic strength, leading to lowering of the overall acidity. Thus,
a progressive increase in DMBPS loading (x) not only facilitates
elimination of available protic sites but also promotes segrega-
90.6
9.4 nil
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
80.0 20.0 nil
79.3 20.7 nil
85.1 14.9 nil
92.9
7.1 nil
89.6 10.4 nil
92.6
90.4
91.5
91.7
93.0
7.4 nil
9.6 nil
8.5 nil
8.3 nil
7.0 nil
CeFeTiO
CeCuTiO
CeZnTiO
CeAgTiO
[a] Reaction conditions: GL/LA=3 (mol/mol); catalyst amount (relative to
LA) 4 wt%; reaction time 1 h; reaction temperature 423 K. [b] GML=glyc-
erol monolaurate; GDL=glycerol dilaurate; GTL=glycerol trilaurate.
&
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
4
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
The PTA catalyst, which is anticipated to possess super
strong Brønsted acidity among the various solid acids exam-
ined, showed comparable catalytic activity with the pristine
STA. Although the PTA gave slightly better LA conversion com-
pared with STA, it resulted in lower GML yield and selectivity.
By comparing the catalytic performances of various Keggin-
type heteropolyacids, the following trends may be inferred: LA
conversion: Ce1/3H3SiW12O40 > Fe1/3H3SiW12O40 > AgH3SiW12O40 ꢂ
H3PW12O40 > H4SiW12O40 @ H3PMo12O40; GML yield: Ce1/
3H3SiW12O40 > H4SiW12O40 > Fe1/3H3SiW12O40 > H3PW12O40 >
AgH3SiW12O40 @ H3PMo12O40. Note that although PMA possess-
es higher acidity (d31Pꢂ81–90 ppm by the 31P TMPO approach)
than STA,[44] it exhibited the worst catalytic performance during
the esterification reaction. Nonetheless, rare-earth ion-ex-
changed STAs, which tend to introduce Lewis metal centers
onto the Brønsted acidic STA, showed comparable performan-
ces with its pristine counterpart particularly with considerable
increase in LA conversion. This may be readily seen by compar-
ing the results obtained from the Ce1/3H3SiW12O40 catalyst
(Table 1; entry 17) with STA (entry 4). Clearly, the features
(strength and amount) of the Brønsted acid sites are not the
only key factors dictating the esterification reaction. In this
context, other factors such as Lewis acidity, diffusivity of reac-
tant/product, and mass transport resistance may also affect
the catalytic activity. One important drawback of these Keggin-
type heteropolyacid catalysts is that the reaction mandatorily
proceeds homogeneously, which makes product separation
and catalyst recovery very difficult.
ucts (top layer) and the colloidal IL–STA catalyst (bottom layer).
As such, the precipitated organic composite salt catalyst may
be recycled after filtering and washing treatments.[22–25]
Among the various IL–STA catalysts examined, the
[DMBPSH]H3SiW12O40 catalyst (Table 1; entry 5) was found to ex-
hibit the best catalytic activity with 88.1% LA conversion and
78.4% GML yield, corresponding to an optimal GML selectivity
of 89%. As the incorporation of the DMBPS IL tends to elimi-
nate the available protic sites from STA on a one-to-one basis,
the effect of its loading on catalytic activity was also investigat-
ed. It is indicative that a notable decrease in LA conversion
and monoglyceride yield was observed with increasing the rel-
ative loading of constituents (x) in the [DMBPSH]xH4ꢀxSiW12O40
salt from 1 to 4. It is anticipated that, while a gradual increase
in x should lead to progressive lowering of the amount and
strength of protic sites, it also serves to promote segregation
4ꢀ
of SiW12O40 polyanions in the secondary structure of STA,
which in turn renders better molecular transport during the
esterification reaction.[23–25] This may be attributed to the pseu-
doliquid nature of the Keggin-type heteropolyacids.[46] Al-
though DMBPS IL itself is a weak acidic catalyst (cf. Figure 2 f)
that showed inferior catalytic performance for the esterification
reaction, it has an affinity for Brønsted acid protons, leading to
the formation of DMBPSH+ complexes in the presence of STA.
In this context, it is indicative that pristine STA possesses
a higher amount of Brønsted acid sites with strongest acidic
strengths (cf. Figure 2a), however, it is handicapped by trans-
port resistance. As such, the four available Brønsted acid pro-
tons (H+) may not be fully accessible to the reactants during
the esterification reaction. The modest catalytic activity ob-
served (Table 1; entry 4). Upon incorporating the SAFIL onto
the STA, the presence of DMBPS favors fast transport of reac-
tants/products during the reaction, leading to the higher activ-
ity observed for the [DMBPSH]H3SiW12O40 catalyst (entry 5).
However, as the loading of DMBPS was further increased, the
continuous lowering of acidity (cf. Figure 2b–e) led to a pro-
gressive decrease in catalytic activity. By the same token, varia-
tions in the catalytic activities observed for the two DMBPS-
modified composite salts prepared by substituting STA for TPA
and TMA (entries 13 and 14) may be realized. The variations in
catalytic activities observed for the series of [XILH]xH4ꢀxSiW12O40
(x=1–4; XIL =DMBPS, DMPPS, TEAPS, MIMPS, and PPS) cata-
lysts during esterification of LA with GL may be attributed to
a subtle balance between catalyst acidity and molecular trans-
port resistance, as exemplified by the optimal performance
found over the [DMBPSH]xH4ꢀxSiW12O40 catalyst with x=1.0. As
the [DMBPSH]H3SiW12O40 catalyst affords the best catalytic per-
formance, it is exploited as a standard test catalyst for the sub-
sequent process optimization and kinetic studies.
Moreover, conventional acid catalysts such as zeolite ZSM-5
(entry 18), Amberlyst-15 (entry 19), and active solid oxides
(CeMTiO; M=Mn, Fe, Cu, Zn, and Ag; entries 20–24) were also
examined. These catalysts showed poor activity during the
esterification reaction and they gave rise to inferior catalytic
performances (typically with LA conversion ꢃ35% and GMA
yield ꢃ32%). Although LA conversions as high as 94–96%
have been reported for sulfonic acid-functionalized SBA-15 cat-
alysts[19] with GML selectivities of 60–70%, the reactions were
carried out with a much longer reaction time (20 h; under
reflux). In addition, although a high GML selectivity (>98%)
was found over a zeolitic metal–organic framework (MOF) cata-
lyst,[45] the observed monoglyceride yield was only about 40%.
By comparison, the series of IL–STA catalysts (Table 1; en-
tries 5–12) exhibit satisfactory catalytic performance for esterifi-
cation of LA with GL. For example, for the [XILH]H3SiW12O40 (i.e.,
x=1) series catalysts with XIL =DMBPS, DMPPS, TEAPS, MIMPS,
and PPS, yields of GML of 78.4%, 72.3%, 67.5%, 70.2%, and
70.5% may be achieved, respectively. Most importantly, as
these organic–inorganic composites are highly soluble in water
and strong polar solvents but immiscible with apolar esters,
such reaction systems possess dual homogeneous/heterogene-
ous characteristics. Although the system is strictly homogene-
ous during the early stage of the esterification reaction owing
to complete dissolution of the catalyst in GL and water, the
mixed solution gradually turns turbid and milky during the re-
action. As the reaction reaches its final stage, the high yield of
product esters (GML and GDL) provokes the occurrence of
spontaneous self-separation, forming a biphasic layer of prod-
Effects of reaction parameters on the esterification reaction
The influence of experimental variables, such as reactant molar
ratio, catalyst amount, reaction time, and reaction temperature,
during esterification of LA with GL were further investigated
by using the [DMBPSH]H3SiW12O40 organic composite salt as
the model catalyst. The experiments were carried out by vary-
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Figure 3. Variations of LA conversion and GML yield during esterification of lauric acid (LA) with glycerol (GL) over the [DMBPSH]H3SiW12O40 catalyst as a func-
tion of (a) GL/LA reactant ratio, (b) catalyst amount, (c) reaction time, and (d) reaction temperature. While varying each experimental variable, the other pa-
rameters were kept constant at their predicted values: GL/LA=3.0, catalyst amount=4 wt%, reaction time=1.0 h, and reaction temperature=423 K.
ing one variable while keeping the other variables fixed at the
following values: GL/LA=3.0, catalyst amount=4 wt% (rela-
tive to LA), reaction time=1.0 h, and reaction temperature=
423 K. The catalytic results are depicted in Figure 3.
reached at 1 h. Further prolonging the reaction time to 4.0 h,
the LA conversion gradually increase to 96.1% while the GML
yield dropped to approximately 72.0%. As esterification of GL
with LA is a consecutive reaction, it may be expressed as:[16]
As esterification is a reversible reaction, an excess of GL is
usually added to promote the reaction rate and shift the reac-
tion equilibrium towards the formation of esters based on Le
Chatelier’s principle.[17] As such, a molar ratio of alcohol (GL) to
fatty acid (LA) of 2.0 to 4.0 was examined. As shown in Fig-
ure 3a, both the LA conversion and GML yield were found to
increase initially with increasing GL/LA ratio, reaching a maxi-
mum of 88.1% and 78.4%, respectively, at GL/LA=3.0. After-
wards, both conversion and yield decreased consistently as the
reactant ratio further increased as a result of the extended di-
lution of the catalyst in the reaction mixture.
GL þ LA ! GML þ H2O;
GML þ LA ! GDL þ H2O;
GDL þ LA ! GTL þ H2O
ð1Þ
Thus, the formation of GDL and GTL are responsible for the
decrease in GML yield and selectivity. In other words, a pro-
longed reaction time does not achieve an optimal yield of the
more valuable GML product. For the reaction conditions em-
ployed for the esterification reaction, it is indicative that exten-
sion of reaction time is unfavorable for GML yield and selectivi-
ty.
The effect of catalyst amount on catalytic activity during the
esterification reaction was also studied. As shown in Figure 3b,
the LA conversion gradually increased from approximately 68.2
to 91.6% as the amount of catalyst (relative to LA) in the reac-
tion mixture increased from 2 to 6 wt%. This may be attribut-
ed to the progressive increase in active acid sites available for
catalyzing the esterification reaction. Meanwhile, the yield of
GML also increased initially with increasing catalyst loading; it
reached a maximum (78.4%) at a catalyst loading of 4 wt%.
Further increasing the amount of catalyst resulted in a notable
decrease in GML yield while increasing the LA conversion, indi-
cating that the decrease in GML selectivity is due to formation
of di- and triglycerides (GDL and GTL).
Finally, the effect of reaction temperature on catalytic activi-
ty was investigated by fixing the other reaction parameters,
that is, GL/LA=3.0, catalyst amount=4 wt%, and reaction
time=1.0 h, while changing the reaction temperature from
403–443 K. In this case, a paradox between the reaction rate
and GML selectivity with reaction temperature is anticipated.
At lower reaction temperatures, the reaction rate should be
lower, whereas undesirable side reactions may be promoted at
higher reaction temperatures. Thus, notable increases in LA
conversion as well as GML yield were observed at lower tem-
peratures (<423 K; Figure 3d). A maximum GML yield of
78.4% and 88.1% LA conversion were observed at 423 K. How-
ever, further increasing the temperature led to a drastic de-
crease in GML yield while the LA conversion increased gradual-
ly. Again, this may be ascribed to formation of undesirable by-
products at elevated temperatures. On the other hand, at
The effect of reaction time on LA conversion and GML yield
was examined under the reaction conditions of GL/LA=3.0,
catalyst amount=4 wt%, and reaction temperature=423 K
while varying the reaction time from 0.5 h to 4.0 h. As shown
in Figure 3c, both the conversion and monolaurin yield in-
crease rapidly with reaction time and a good yield (78.4%) was
&
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
lower reaction temperatures (<423 K), a lower GML yield is an-
ticipated owing to lowering of the LA conversion.
Furthermore, the quality of regression fitting with Equa-
tion (2) was evaluated by the analysis of variance (ANOVA) and
the results are summarized in Table 4. The fact that the model
The results obtained from the above studies therefore reveal
the following optimized conditions for esterification of LA with
GL over the [DMBPSH]H3SiW12O40 catalyst: GL/LA reactant
molar ratio 3.0, catalyst amount 4 wt% (relative to LA loading),
reaction time 60 min, and reaction temperature 423 K.
Table 4. ANOVA table for the proposed quadratic model.
Source
Sum of squares DF Mean square F value Prob.>F Sign.
Model
x1
x2
x3
x12
194.23
24.62
12.41
69.48
20.64
4.04
9
1
1
1
1
21.58
24.62
12.41
69.48
20.64
4.04
294.45 <0.0001 **
335.91 <0.0001 **
169.32 <0.0001 **
948.00 <0.0001 **
281.61 <0.0001 **
Process optimization
The RSM is used for optimization of process variables and
mutual interactions between them and for prediction of the re-
sponse (GML yield), Y, based on an empirical model:
x22
1
55.09
0.0001 **
x32
x1ꢁ2
x1ꢁ3
x2ꢁ3
Residual
Lack-of-fit
Pure error
Cor total
48.13
0.41
0.011
7.96
0.51
0.37
1
1
1
1
7
3
4
16
48.13
0.41
0.011
7.96
0.073
0.12
0.35
656.67 <0.0001 **
5.61
0.15
0.0497
0.706
*
2
2
Y ¼78:40 þ 1:75x1 þ 1:25x2 þ 2:95x3ꢀ2:21x1 ꢀ0:98x2
108.61 <0.0001 **
ð2Þ
2
ꢀ3:38x3 ꢀ0:32x1x2ꢀ0:053x1x3ꢀ1:41x2x3
3.61 0.1232 NS
0.14
194.74
For which x1, x2, and x3 are the coded values of the process var-
iables specified in Table 2. To verify the model, up to 17 experi-
mental runs were performed and the response value corre-
sponding to each experimental set was obtained by multiple
regression analysis, as depicted in Table 3. It is clear that no
significant differences between the experimental values and
predicted data were observed, indicating the suitability of the
quadratic model.
R2 =0.9974, Adeq precision: 49.844. [**] Represents highly significant; [*]
represents significant; [NS] not significant.
F value (294.45) was much greater than its tabular counterpart
(F0.01,9,7 =6.71) was observed, implies that the model was signif-
icant, as revealed by the small “Prob.>F” values (<0.0001). In
2
2
2
this context, x1, x2, x3, x1 , x2 , x3 , x1x2, and x2x3 terms were sig-
nificant. Clearly, the significance of the three independent vari-
ables follows the order: reaction temperature (x3)> alcohol/
acid molar ratio (x1)> amount of catalyst (x2). In addition, a rela-
tively low coefficient of variation (CV=0.36) also justifies the
precision of the model and the reliability of the process. More-
over, the satisfactory sample correlation coefficient (R2 =
0.9974) observed ensured the validity of the chosen model
[Eq. (2)]. An observed “lack-of-fit F value” of 3.61 also implied
that it was not significant relative to the pure error.
Table 2. List of symbols for different process variables and corresponding
coded levels and ranges used in the experimental designs.
Variable (unit)
Symbol
Level and range
0
ꢀ1
2.5
3.0
413
1
3.5
5.0
Glycerol/lauric acid ratio (mol/mol)
Amount of catalyst (wt%)
Reaction temperature (K)
x1
x2
x3
3.0
4.0
423
433
In terms of mutual interactions between a pair of variables
during the esterification reaction, interactions between
amount of catalyst (x2) and reaction temperature (x3) were
found to be much more significant than those between the
GL/LA molar ratio (x1) and amount of catalyst (x2), as can be
seen from the observed “Prob.>F” value in Table 4. It may also
be readily seen from the three-dimensional (3D) response sur-
face and contour plots shown in Figure 4. As can be seen from
Figure 4(a,d), the influence of the reactant molar ratio on GML
yield is indeed more significant than the amount of catalyst.
The same notion may also be applied for reaction temperature
versus reactant molar ratio (Figure 4(b,e)); the influence of the
latter is clearly less significant. In other words, reaction temper-
ature (x3) is the most crucial variable for esterification, com-
pared with the reactant ratio (x1) and amount of catalyst (x2),
which is in excellent agreement with the ANOVA data in
Table 4. By the same token, interactions between the variable
pairs of x3 and x2 on product yield showed highly significant ef-
fects compared with that of x3 and x1, as revealed by the ellip-
tical mound shape of the contour plots in Figure 4e,f, respec-
Table 3. List of experimental designs and response values during esterifi-
cation of glycerol with lauric acid over the [DMBPSH]H3SiW12O40 catalyst.
Run
Variable and value
GML yield [%]
x1
x2
x3
Experimental
Predicted
1
2
3
4
5
6
7
8
2.5
3.5
2.5
3.5
2.5
3.5
2.5
3.5
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
5.0
5.0
4.0
4.0
4.0
4.0
3.0
5.0
3.0
5.0
4.0
4.0
4.0
4.0
4.0
423
423
423
423
413
413
433
433
413
413
433
433
423
423
423
423
423
71.63
76.18
74.88
78.14
68.19
71.40
74.32
77.32
68.56
73.76
77.15
76.70
78.33
78.37
78.66
78.16
78.49
71.89
76.04
75.02
77.89
68.05
71.67
74.05
77.46
68.44
73.75
77.15
76.82
78.40
78.40
78.40
78.40
78.40
9
10
11
12
13
14
15
16
17
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Figure 4. (a–c) 3D response surface and (d–f) contour plots showing variations between a pair of experimental variables (see Table 2) on predicted GML yields
while keeping other variables at a constant level of 0. Assignments: interactions between (a, d) GL/LA ratio (x1) and catalyst amount (x2); (b, e) x1 and reaction
temperature (x3); (c, f) x2 and x3.
tively. In this regard, the interactions between x1 and x2 (Fig-
ure 4d) are less significant for the GML yield.
optimal conditions as above. After each run, the catalyst was
recovered from the system by filtration, followed by washing
with diethyl ether and drying under vacuum for 8 h at 343 K.
As shown in Figure 5, the [DMBPSH]H3SiW12O40 catalyst
showed excellent stability and recyclability. The LA conversion
decreased from the initial run of 90.0 to 78.8% after six con-
secutive cycles, whereas the GML product yield slightly de-
creased from 79.1 to 72.9%. Additional measurements by FTIR
spectroscopy revealed that, except for the decrease in peak in-
tensities, the positions of the four characteristic bands from
Based on the 17 experimental runs and response from the
RSM, the experimental and predicted GML yield values may
readily be compared (Table 3). As pointed out earlier, the
values predicted based on the proposed model closely resem-
ble the actual values. As a result, an optimum GML yield of
79.18% was predicted for the esterification of GL with LA over
the [DMBPSH]H3SiW12O40 catalyst under the optimal process
conditions of glycerol/lauric acid molar ratio=3.32 (mol/mol),
amount of catalyst=3.91 wt%, and reaction temperature=
426.5 K with a reaction time of 1.0 h. To verify the validity of
these predicted data, three parallel experiments were carried
out under the optimized conditions. For convenience, the fol-
lowing values of process parameters were adopted: GL/LA
ratio=3.3 (mol/mol), catalyst amount=3.9 wt%, reaction tem-
perature=427 K, and reaction time=1.0 h. Consequently, an
average GML yield of 79.1% was obtained, in excellent agree-
ment with the predicted value. The above results clearly indi-
cate the validity of the RSM model.
4ꢀ
the Keggin SiW12O40 polyanions remained practically un-
Catalyst stability and recyclability
Figure 5. Stability of the [DMBPSH]H3SiW12O40 catalyst during esterification
of glycerol (GL) with lauric acid (LA). Reaction conditions: GL/LA ratio=3.3;
amount of catalyst=3.9 wt%; reaction time=1.0 h; reaction temperatur-
e=427 K.
The stability and recyclability of the [DMBPSH]H3SiW12O40 cata-
lyst for esterification of glycerol with lauric acid were also ex-
amined. The test experiments were performed under the same
&
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
8
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
changed even after six runs (Figure S5; in the Supporting Infor-
mation). Further elemental analyses by inductively coupled
plasma (ICP) confirmed that the concentration of Si and W
atoms in the catalyst decreased from 0.89 and 5.85 wt% in the
fresh catalyst to 0.65 and 4.25 wt% in the regenerated spent
catalyst, respectively. Thus, we attribute the decreases in cata-
lytic activity to the loss of catalyst during the recovery and re-
generation process.
Conclusions
A series of propyl sulfonic acid-functionalized ionic liquids
modified silicotungstic acid (SAFIL–STA) composite catalysts,
namely [XILH]xH4ꢀxSiW12O40 (x=1–4; XIL =DMBPS, DMPPS,
TEAPS, MIMPS, and PPS) were successfully synthesized and ex-
amined. These SAFIL–STA composites were found to be highly
efficient for catalyzing esterification of lauric acid with glycerol
for selective production of glycerol monolaurate surpassing
ion-exchanged and/or heteroatom-substituted heteropolyacids
and conventional catalysts, such as zeolite, Amberlyst-15, and
active solid oxides. An optimum GML yield of 79.1% over the
[DMBPSH]H3SiW12O40 catalyst under reaction conditions of
a GL/LA ratio of 3.3 (mol/mol), a catalyst amount of 3.9 wt%,
and a reaction temperature of 427 K within a reaction time of
1.0 h. These results were in excellent agreement with that pre-
dicted by the response surface methodology. The superior cat-
alytic activity observed for these SAFIL–STA composites are at-
tributed to the high acidity, low transport resistance, and
“pseudoliquid” characteristics of the catalysts. The esterification
of GL with LA over the [DMBPSH]H3SiW12O40 catalyst was found
to be a second-order reaction with an activation energy of
39.49 kJmolꢀ1. This together with the self-separation property
of the catalyst system, the high activity as well as the excellent
stability and recyclability found herein, indicates that the
SAFIL–STA composites are truly efficient and reusable catalysts
suitable for practical industrial applications, particularly for the
selective production of GML.
Kinetic study
The kinetics involved in the esterification of GL with LA to GML
over the [DMBPSH]H3SiW12O40 catalyst were also examined. The
reactions were carried out at different temperatures by using
a GL/LA ratio of 3.3 (mol/mol) and a catalyst amount of
3.9 wt% while withdrawing approximately 1 mL of the reaction
mixture for reactant analysis at regular time intervals. Figure 6a
displays the variations of ln(CA/CB) with reaction time at differ-
ent temperatures. Based on the experimental design [Experi-
mental Section; Eq. (8)], the linear fitting of the results at the
temperatures 418, 423, 428, and 433 K corresponds to R2
values of 0.99, 0.96, 0.95, and 0.96, respectively. The results
clearly indicate that esterification of GL with LA is indeed
a second-order reaction. The reaction rate constant (k) so ob-
tained at different temperatures were further analyzed by
using the Arrhenius equation [see Experimental Section;
Eq. (9)], as shown in Figure 6b. Accordingly, an activation
energy (Ea) and a pre-exponential factor (k0) of 39.49 kJmolꢀ1
and 399 Lmolꢀ1 minꢀ1, respectively, were deduced. The Ea value
so obtained for the [DMBPSH]H3SiW12O40 catalyst during
esterification of GL with LA was lower than that found for the
same reaction over the HSO3-functionalized SBA-15
Experimental Section
Catalyst preparation
All the chemicals were analytical reagents and were used without
further purification unless specified otherwise. The SAFIL–HPA cata-
lysts were prepared by mixing the desired amounts of calcined
heteropolyacids (HPAs), such as silicotungstic acid (STA), phospho-
tungstic acid (PTA), and phosphomolybdic acid (PMA), with various
propyl sulfonic acid-functionalized ionic liquids (SAFILs), including
N,N-dimethyl(benzyl)ammonium propyl sulfobetaine (DMBPS), N,N-
dimethyl(phenyl)ammonium propyl sulfobetaine (DMPPS), triethy-
lammonium propyl sulfobetaine (TEAPS), methylimidazolium
propyl sulfobetaine (MIMPS), and pyridinium propyl sulfobetaine
(42 kJmolꢀ1 [18] and the zinc carboxylate (51 kJmolꢀ1 [47]
) cata-
)
lysts. The lower Ea value observed herein for the SAFIL–STA
composite catalyst may be ascribed to the high acidity, low
molecular transport resistance, and the “pseudoliquid” charac-
teristics of the [DMBPSH]H3SiW12O40 catalyst. Thus, the
[DMBPSH]H3SiW12O40 composite is truly a highly effective cata-
lyst for the production of GML by esterification of GL with LA.
Figure 6. (a) Variations of ln(CA/CB) versus time and (b) Arrhenius plot for esterification of GL with LA over the [DMBPSH]H3SiW12O40 catalyst. Reaction condi-
tions: GL/LA ratio, 3.3 (mol/mol); catalyst amount, 3.9 wt%; reaction time, 15–75 min; reaction temperature, 418–433 K.
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
9
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
(PPS). The syntheses of these organic–inorganic composite salts
are similar to the procedures reported elsewhere.[23] In brief, varied
stoichiometric amounts of HPAs, SAFILs, and water were put into
a three-necked flask, then, the mixture was kept at 363 K while stir-
ring continuously for 24 h. After removing the water, the obtained
solid product was washed with diethyl ether followed by drying
under vacuum. As a result, a series of organic STA salts may be ob-
tained, for example, for the DMBPS IL series, the composite cata-
lysts were expressed as [DMBPSH]xH4ꢀxSiW12O40, where x=1–4 rep-
resents the DMBPS/STA molar ratio. For comparison purposes,
Esterification reaction
The catalytic activity of about two dozen different catalysts (see
Table 1) were assessed and compared by esterification of glycerol
(GL) with lauric acid (LA), which is anticipated to form a distribution
of glycerine laurates as the main products, including glycerol mon-
olaurate (GML), glycerol dilaurate (GDL), and glycerol trilaurate
(GTL), the selectivity of which depends on the type of catalyst
used as well as the reaction conditions. Typically, a fixed amount of
LA (0.05 mol) was added with the desired amounts of GL and the
solid catalyst in a three-necked flask (100 mL) equipped with a ther-
mometer and a mechanical stirrer. The reaction mixture was first
heated to 423 K for a desired period of time in an oil bath under
continuous stirring, then, cooled to room temperature. At the end
of the esterification reaction, the high yield of esters (mostly GML
and GDL) tend to self-separate spontaneously to form biphasic
layers of products (upper layer) and colloidal IL–HPA catalyst
(lower layer). After withdrawing the products from the reaction
bath, the precipitated IL–HPAs catalyst was filtered and washed
with diethyl ether for recycling. The reaction products were ana-
lyzed by gas chromatography (GC; Agilent 7890B) by using a HP-5
capillary column equipped with a flame-ionization detector (FID)
and using methyl laurate as the internal standard. The product
yields were derived by:
other
SAFIL–STA
composite
salts,
for
example,
[DMPPSH]H3SiW12O40, [TEAPSH]H3SiW12O40, [MIMPSH]H3SiW12O40,
and [PPSH]H3SiW12O40 (i.e., with x=1), and various ion-exchanged
and/or heteroatom-substituted HPA[48] were also prepared together
with a series of active solid oxides (CeMTiO; M=Mn, Fe, Cu, Zn,
and Ag) synthesized by using a sol-gel method,[15] as listed in
Table 1.
Catalyst characterization
FTIR spectra were acquired with a Bruker IFS-28 spectrometer by
using anhydrous KBr as the standard. Each spectrum was recorded
in the range 4000–400 cmꢀ1 with a resolution of 1 cmꢀ1. The XRD
measurements were conducted with a DX-2700 diffractometer
(Dandong Kemait NDT Co., Ltd., China) by using CuK radiation at
40 kV and 30 mA. Each XRD profile was scanned over a 2q angle of
5–808 with an interval of 0.048. Thermogravimetric and differential
thermogravimetric (TG-DTG) analyses were recorded with a TGA/
DSC-1 (Mettler Toledo) thermal analyzer. Typically, each sample was
heated from room temperature (RT; 298 K) to 873 K at a heating
rate of 10 Kminꢀ1 under a dynamic N2 atmosphere. XPS spectra
were measured with an ESCALab220i-XL electron spectrometer (VG
Scientific) equipped with a 300 W AlKa (1486.6 eV) monochromatic
radiation source. Elemental analyses were carried out with an iCAP
6500 ICP-OES analyzer (Thermo Scientific) equipped with a unique
charge injection device (CID) detector. High-resolution 1H and
13C NMR spectra were recorded with a Bruker-Biospin AV-500 spec-
Yield ð%Þ ¼ Conversion ð%Þ ꢄ Selectivity ð%Þ
ð3Þ
Experimental design and mathematical model
Response surface methodology (RSM) was exploited for optimiza-
tion of the experimental variables based on a Box–Behnken design
(BBD).[31] The experimental designs invoked investigated the effects
of three independent process variables, namely the reactant (GL/
LA) molar ratio (x1), amount of catalyst (x2), and reaction tempera-
ture (x3), on LA conversion and yield of GML, as summarized in
Table 2. As such, a 33 full-factorial central composite design with
three coded levels was used, leading to a total of 17 experimental
sets, including 12 factorial points and 5 centering points. The three
experimental variables above were tested at coded levels and
ranges expressed by either a plus sign (+1; high level/value), zero
(0; center level/value), or a minus sign (ꢀ1; lower level/value).
Design-Expert 6.0.5 (Stat-Ease, USA) software was used for the
design of experiments and analyses of experimental data. The
coded values of these factors were obtained by the equation:
trometer operating at
a Larmor frequency of 500.13 and
125.76 MHz, respectively. D2O was used as solvent as well as the
source signal for field lock. Solid-state 31P magic angle spinning
(MAS) NMR experiments were conducted with a Bruker Biospin
AVANCE III 500 spectrometer operating at 202.46 MHz. All spectra
were recorded by using a single pulse sequence under the follow-
ing conditions: pulse width, 1.5 ms (p/6); recycle delay, 10 s;
sample spinning rate, 12 kHz. The 31P NMR chemical shift (d31P)
was referenced to 85% H3PO4 aqueous solution as 0 ppm.
Xi ꢀ X0
ð4Þ
xi ¼
DXi
where xi, Xi, and X0 (i=1–3) represent the coded, real, and central
values of the independent variables, respectively, and Xi =(high lev-
elꢀlow level)/2 denotes the step-change values of the associated
variable. A second-order model equation given by RSM was used
to reveal the interactive effects between process variables, to opti-
mize the reaction process, and to predict the yield of the primary
product (namely, GML; denoted as Y). The responses of the experi-
mental design may be expressed by the quadratic equation:
Here, the sample preparation procedures invoked for acidity char-
acterization by this unique approach are described in brief below:
prior to the adsorption of the TMPO probe molecule, the catalyst
sample was first subjected to a thorough dehydration treatment at
423 K for 12 h under vacuum (<10ꢀ5 Torr). Subsequently, a known
amount of TMPO (dissolved in anhydrous CH2Cl2) was introduced
into a vessel containing the dehydrated solid sample in a N2 glove-
box, followed by removal of the CH2Cl2 solvent by evacuation at
323 K for 0.5 h. To ensure a uniform adsorption of TMPO adsorbate
on the catalyst substrate, the sealed sample vessel was further sub-
jected to a thermal treatment at 393 K for at least 24 h. Finally, the
TMPO-loaded sample was transferred into a MAS rotor in a N2 glo-
vebox and the rotor was then sealed with a gas-tight Kel-F cap.
3
3
3
X
X
X
Y ¼ b0 þ
bixi þ
biixi2 þ
bijxixj
ð5Þ
i¼1
i¼1
i<j
where b0, bi, bii, bij denotes the regression coefficients of the offset,
linear, quadratic, and interactive terms for the variables, respective-
&
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
10
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
ly, and xi and xj are the uncoded independent variables.[49] Design-
Expert 6.0.5 (Stat-Ease, USA) software was used for the design of
experiments and analyses of experimental data. The coefficient of
determination (R2) was used to evaluate the accuracy of regression
fittings and reliability of the model.
Keywords: glycerol monolaurate
optimization · propyl sulfonic ionic liquid · silicotungstic acid
·
kinetic
·
process
[1] M. S. Ardi, M. K. Aroua, N. A. Hashim, Renewable Sustainable Energy Rev.
2015, 42, 1164–1173.
[2] S. Zhu, Y. Zhu, X. Gao, T. Mo, Y. Zhu, Y. Li, Bioresour. Technol. 2013, 130,
45–51.
[3] C. Crotti, E. Farnetti, J. Mol. Catal. A 2015, 396, 353–359.
[4] G. V. Waghmare, M. D. Vetal, V. K. Rathod, Ultrason. Sonochem. 2015, 22,
311–316.
[5] D. Sun, Y. Yamada, S. Sato, Appl. Catal. B 2015, 174, 13–20.
[6] G. S. Kumar, Y. Wee, I. Lee, H. J. Sun, X. Zhao, S. Xia, S. Kim, J. Lee, P.
Wang, J. Kim, Chem. Eng. J. 2015, 276, 283–288.
[7] Q. Li, J. D. Estes, P. M. Schlievert, L. Duan, A. J. Brosnahan, P. J. Southern,
C. S. Reilly, M. L. Peterson, N. Schultz-Darken, K. G. Brunner, K. R.
Nephew, S. Pambuccian, J. D. Lifson, J. V. Carlis, A. T. Haase, Nature
2009, 458, 1034–1038.
Kinetic study
Kinetic studies during esterification of GL with LA over the
[DMBPSH]H3SiW12O40 catalyst, which exhibited the best catalytic
performance compared with the other solid acids (see Table 1),
were carried out by using a set of variables obtained from the
multi-objective optimization, namely under varied reaction time
(0–75 min) and temperatures (418–433 K) with fixed GL/LA=3.3
and catalyst amount 3.9 wt%. The reaction rate, r, for esterification
of GL and LA to GML may be expressed by Equation (6):
[8] J. F. Puna, J. F. Gomes, J. M. N. Correia, A. P. S. Dias, J. C. Bordado, Fuel
2010, 89, 3602–3606.
[9] X. X. Han, H. Du, C. T. Hung, L. L. Liu, P. H. Wu, D. H. Ren, S. J. Huang,
S. B. Liu, Green Chem. 2015, 17, 499–508.
dCA
dt
¼ k CaCb ꢀ k CgCh
ð6Þ
r ¼ ꢀ
þ
ꢀ
B
C
D
A
[10] Y. Leng, J. Wang, D. R. Zhu, X. Q. Ren, H. Q. Ge, L. Shen, Angew. Chem.
Int. Ed. 2009, 48, 168–171; Angew. Chem. 2009, 121, 174–177.
[11] S. Zhu, X. Gao, F. Dong, Y. Zhu, H. Zheng, Y. Li, J. Catal. 2013, 306, 155–
163.
[12] X. X. Han, K. K. Chen, W. Yan, C. T. Hung, L. L. Liu, P. H. Wu, K. C. Lin, S. B.
Liu, Fuel 2016, 165, 115–122.
[13] M. D. Machado, J. Pꢂrez-Pariente, E. Sastre, D. Cardoso, A. M. de Guere-
Çu, Appl. Catal. A 2000, 203, 321–328.
[14] T. Y. Wibowo, A. Z. Abdullah, R. Zakaria, Appl. Clay Sci. 2010, 50, 280–
281.
where k+ and k are the forward and reverse rate constants, re-
ꢀ
spectively, CA, CB, CC, and CD represent the concentrations of LA,
GL, GML, and H2O, respectively, and a, b, g, and h denote their cor-
responding reaction order. Given that the reaction was carried out
under excess GL, the esterification reaction could be considered as
an irreversible process. In this context, the second term in Equa-
tion (6) may be cancelled, and the rate equation may be simplified
as:
[15] M. A. Olutoye, C. P. Wong, L. H. Chin, B. H. Hameed, Fuel Process. Tech-
nol. 2014, 124, 54–60.
[16] L. Zhou, E. Al-Zaini, A. A. Adesina, Fuel 2013, 103, 617–625.
[17] P. Y. Hoo, A. Z. Abdullah, Chem. Eng. J. 2014, 250, 274–287.
[18] L. Hermida, A. Z. Abdullah, A. R. Mohamed, Chem. Eng. J. 2011, 174,
668–676.
[19] L. Hermida, A. Z. Abdullah, A. R. Mohamed, J. Appl. Sci. 2010, 10, 3199–
3206.
[20] I. Dosuna-Rodrꢃguez, E. M. Gaigneaux, Catal. Today 2012, 195, 14–21.
[21] J. B. Monteiro, M. G. Nascimento, J. L. Ninow, Biotechnol. Lett. 2003, 25,
641–644.
dCA
dt
¼ kCAaCBb
ð7Þ
r ¼ ꢀ
For simplicity, the reaction orders a and b were hypothesized as
1 (first order). Accordingly, the values of k for the esterification re-
action at different temperatures may be obtained by linear fitting
of logarithmic ln(CA/CB) versus time, which may be defined as:
CA
CB
ð8Þ
ln ¼ ðCAO ꢀ CBOÞkt þ C
[22] Y. Leng, J. Wang, D. Zhu, Y. Wu, P. Zhao, J. Mol. Catal. A 2009, 313, 1–6.
[23] X. X. Han, Y. F. He, C. T. Hung, L. L. Liu, S. J. Huang, S. B. Liu, Chem. Eng.
Sci. 2013, 104, 64–72.
where CAO and CBO refer to the difference between the initial con-
centrations of LA and GL, whereas the variations of the reaction
rate with temperature may be expressed by the Arrhenius equa-
tion:
[24] X. Han, W. Yan, K. Chen, C. T. Hung, L. L. Liu, P. H. Wu, S. J. Huang, S. B.
Liu, Appl. Catal. A 2014, 485, 149–156.
[25] M. Y. Huang, X. X. Han, C. T. Hung, J. C. Lin, P. H. Wu, J. C. Wu, S. B. Liu, J.
Catal. 2014, 320, 42–51.
[26] E. D. Bates, R. D. Mayton, I. Ntai, J. H. Davis, Jr., J. Am. Chem. Soc. 2002,
124, 926–927.
[27] A. E. Visser, J. D. Holbrey, R. D. Rogers, Chem. Commun. 2001, 2484–
2485.
Ea 1
ð9Þ
ln k ¼ ln k0 ꢀ
R T
[28] K. Aliouane, M. Hamadꢄne, A. Guehria-La¨ıdoudi, A. Simon, J. Ravez, J.
Fluorine Chem. 2000, 105, 71–76.
where R represents the gas constant, T is the reaction temperature,
and Ea and k0 denote the activation energy and pre-exponential
factor, respectively.
[29] T. Welton, Chem. Rev. 1999, 99, 2071–2084.
[30] C. Xie, H. Li, L. Li, S. Yu, F. Liu, J. Hazard. Mater. 2008, 151, 847–850.
[31] K. Hinkelmann, J. Jo, J. Stat. Planning Inference 1998, 72, 347–354.
[32] T. Okuhara, N. Mizuno, M. Misono, Adv. Catal. 1996, 41, 113–252.
[33] K. M. Parida, S. Mallick, J. Mol. Catal. A 2007, 275, 77–83.
[34] J. Yuan, Q. F. Hou, B. L. Liu, J. Shen, S. C. Lin, Colloids Surf. B 2004, 36,
19–26.
Acknowledgments
[35] G. Li, Y. Ding, J. Wang, X. Wang, J. Suo, J. Mol. Catal. A 2007, 262, 67–
76.
[36] F. J. Berry, G. R. Derrick, J. F. Marco, M. Mortimer, Mater. Chem. Phys.
2009, 114, 1000–1003.
[37] A. H. Jadhav, H. Kim, I. T. Hwang, Bioresour. Technol. 2013, 132, 342–350.
[38] A. Zheng, S. J. Huang, S. B. Liu, F. Deng, Phys. Chem. Chem. Phys. 2011,
13, 14889–14901.
Support of this work by the Project of Science and Technology
Department of Zhejiang Province (No. 2016C31016), the Program
for Zhejiang Leading Team of S & T Innovation (No. 2013TD07),
and the Ministry of Science and Technology, Taiwan (MOST, 104-
2113-M-001-019) is gratefully acknowledged.
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
11
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[39] A. Zheng, S. Li, S. B. Liu, F. Deng, Acc. Chem. Res. 2016, 49, 655–663.
[40] A. Zheng, S. J. Huang, W. H. Chen, P. H. Wu, H. Zhang, H. K. Lee, L. C. de
Mꢂnorval, F. Deng, S. B. Liu, J. Phys. Chem. A 2008, 112, 7349–7356.
[41] A. Zheng, H. Zhang, X. Lu, S. B. Liu, F. Deng, J. Phys. Chem. B 2008, 112,
4496–4505.
[42] S. J. Huang, C. Y. Yang, A. Zheng, N. Feng, N. Yu, P. H. Wu, Y. C. Chang,
Y. C. Lin, F. Deng, S. B. Liu, Chem. Asian J. 2011, 6, 137–148.
[43] S. Zhu, Y. Zhu, S. Hao, H. Zheng, T. Mo, Y. Li, Green Chem. 2012, 14,
2607–2616.
[45] L. H. Wee, T. Lescouet, J. Fritsch, F. Bonino, M. Rose, Z. Sui, E. Garrier, D.
Packet, S. Bordiga, S. Kaskel, M. Herskowitz, D. Farrusseng, J. A. Martens,
Catal. Lett. 2013, 143, 356–363.
[46] M. Misono, Catal. Today 2009, 144, 285–291.
[47] A. Macierzanka, H. Szelag, Ind. Eng. Chem. Res. 2004, 43, 7744–7753.
[48] X. Han, K. Chen, H. Du, X. J. Tang, C. T. Hung, K. C. Lin, S. B. Liu, J. Taiwan
Inst. Chem. Eng. 2016, 58, 203–209.
[49] Istadi, N. A. S. Amin, Fuel Process. Technol. 2006, 87, 449–459.
[44] N. Feng, A. Zheng, S. J. Huang, H. Zhang, N. Yu, C. Y. Yang, S. B. Liu, F.
Deng, J. Phys. Chem. C 2010, 114, 15464–15472.
Received: June 28, 2016Published online on && &&, 0000
&
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
12
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
X. Han,* X. Zhang, G. Zhu, J. Liang,
X. Cao, R. Kan, C.-T. Hung, L.-L. Liu,
S.-B. Liu*
&& – &&
Ionic Liquid–Silicotungstic Acid
Composites as Efficient and Recyclable
Catalysts for the Selective
Esterification of Glycerol with Lauric
Acid to Monolaurin
Ionic liquid/silicotungstic acid compo-
sites: Eco-friendly [DMBPSH]H3SiW12O40
(DMBPS=N,N-dimethyl(benzyl)ammoni-
um propyl sulfobetaine) catalyst shows
superior activity and a unique self-sepa-
ration property for facile recovery and
recycling in the synthesis of glycerol
monolaurate.
ChemCatChem 2016, 8, 1 – 13
www.chemcatchem.org
13
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ