Acid-activated and WOx-loaded montmorillonite catalysts and their catalytic behaviors in glycerol dehydration

Article abstract of DOI:10.1016/S1872-2067(17)62813-4

The use of H₂SO₄-, HCl-, H₃PO₄-, and CH₃COOH-activated montmorillonite (Mt) and WO_x/H₃PO₄-activated Mt as catalysts for the gas-phase dehydration of glycerol was investigated. The WO_x/H₃PO₄-activated Mt catalysts were prepared by an impregnation method using H₃PO₄-activated Mt (Mt-P) as the support. The catalysts were characterized using powder X-ray diffraction, Fourier-transform infrared spectroscopy, N₂ adsorption-desorption, diffuse reflectance ultraviolet-visible spectroscopy, temperature-programmed desorption of NH₃, and thermogravimetric analysis. The acid activation of Mt and WO_x loaded on Mt-P affected the strength and number of acid sites arising from H⁺ exchange, the leaching of octahedral Al³⁺ cations from Mt octahedral sheets, and the types of WO_x (2.7 ≤ x ≤ 3) species (i.e., isolated WO₄/WO₆-containing clusters, two-dimensional [WO₆] polytungstates, or three-dimensional WO₃ crystals). The strong acid sites were weakened, and the weak and medium acid sites were strengthened when the W loading on Mt-P was 12 wt% (12%W/Mt-P). The 12%W/Mt-P catalyst showed the highest catalytic activity. It gave a glycerol conversion of 89.6% and an acrolein selectivity of 81.8% at 320 °C. Coke deposition on the surface of the catalyst led to deactivation.

Full text of DOI:10.1016/S1872-2067(17)62813-4

Chinese Journal of Catalysis 38 (2017) 1087–1100
a v a i l a b l e a t w w w . s c i e n c e d i r e c t . c o m
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c h n j c
Article (Special Issue on Nanoscience and Catalysis)
Acid‐activated and WOx‐loaded montmorillonite catalysts and their
catalytic behaviors in glycerol dehydration
Weihua Yu a,b, Pengpeng Wang b, Chunhui Zhou b,c,d,*, Hanbin Zhao b, Dongshen Tong b, Hao Zhang b,
Huimin Yang e, Shengfu Ji f, Hao Wang c
a Zhijiang College, Zhejiang University of Technology, Hangzhou 310024, Zhejiang, China
b Research Group for Advanced Materials & Sustainable Catalysis (AMSC), Research Center for Clay Minerals, State Key Laboratory Breeding Base of Green
Chemistry Synthesis Technology, Discipline of Industrial Catalysis, College of Chemical Engineering, Zhejiang University of Technology, Hangzhou
310032, Zhejiang, China
c Centre for Future Materials, University of Southern Queensland, Toowoomba, Queensland 4350, Australia
d Engineering Research Center of Non‐metallic Minerals of Zhejiang Province, Zhejiang Institute of Geology and Mineral Resource, Hangzhou 310007,
Zhejiang, China
e Key Laboratory of High Efficient Processing of Bamboo of Zhejiang Province, China National Bamboo Research Center, Hangzhou 310012, Zhejiang,
China
f State Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology, Beijing 100029, China
A R T I C L E I N F O
A B S T R A C T
Article history:
The use of H2SO4‐, HCl‐, H3PO4‐, and CH3COOH‐activated montmorillonite (Mt) and WOx/H3PO4‐
activated Mt as catalysts for the gas‐phase dehydration of glycerol was investigated. The
WOx/H3PO4‐activated Mt catalysts were prepared by an impregnation method using
H3PO4‐activated Mt (Mt‐P) as the support. The catalysts were characterized using powder X‐ray
diffraction, Fourier‐transform infrared spectroscopy, N2 adsorption‐desorption, diffuse reflectance
ultraviolet‐visible spectroscopy, temperature‐programmed desorption of NH3, and thermogravi‐
metric analysis. The acid activation of Mt and WOx loaded on Mt‐P affected the strength and number
of acid sites arising from H+ exchange, the leaching of octahedral Al3+ cations from Mt octahedral
sheets, and the types of WOx (2.7 ≤ x ≤ 3) species (i.e., isolated WO4/WO6‐containing clusters,
two‐dimensional [WO6] polytungstates, or three‐dimensional WO3 crystals). The strong acid sites
were weakened, and the weak and medium acid sites were strengthened when the W loading on
Mt‐P was 12 wt% (12%W/Mt‐P). The 12%W/Mt‐P catalyst showed the highest catalytic activity. It
gave a glycerol conversion of 89.6% and an acrolein selectivity of 81.8% at 320 °C. Coke deposition
on the surface of the catalyst led to deactivation.
Received 13 January 2017
Accepted 10 March 2017
Published 5 June 2017
Keywords:
Glycerol
Acrolein
Dehydration
WOx
Acid‐activated nanoclay
Catalyst
© 2017, Dalian Institute of Chemical Physics, Chinese Academy of Sciences.
Published by Elsevier B.V. All rights reserved.
1. Introduction
The use of biodiesel is being developed to address increas‐
ing global concerns regarding the environment and sustainable
* Corresponding author. Tel/Fax: +86‐571‐88320062; E‐mail: clay@zjut.edu.cn
The work was supported by the National Natural Science Foundation of China (21373185, 41672033, 21506188, 21404090), the Open Project Pro‐
grams of Engineering Research Center of Non‐metallic Minerals of Zhejiang Province (ZD2015k07), of State Key Laboratory Breeding Base of Green
Chemistry‐Synthesis Technology (GCTKF2014006), of Key Laboratory of High Efficient Processing of Bamboo of Zhejiang Province (2016), and of
State Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology (CRE‐2016‐C‐303).
DOI: 10.1016/S1872‐2067(17)62813‐4 | http://www.sciencedirect.com/science/journal/18722067 | Chin. J. Catal., Vol. 38, No. 6, June 2017

1088
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
development [1−3]. In biodiesel production, a large quantity of
glycerol is produced as a by‐product [4,5]. Value‐added use of
glycerol is important in improving the economic viability of the
biodiesel industry [6]. One attractive route is the catalytic de‐
hydration of glycerol to the value‐added intermediate acrolein
[7], which is widely used in the synthesis of pharmaceuticals,
detergents, and polymers [8,9]. Acrolein is currently produced
by oxidation of propylene, which is derived from
non‐renewable petroleum, using complex multi‐component
BiMoOx‐based catalysts [10]. The gas‐ or liquid‐phase dehydra‐
tion of glycerol using acid catalysts is a more cost‐effective and
eco‐friendly method for acrolein production [3,11].
composed of a central Al2O3 sheet sandwiched between two
tetrahedral SiO2 sheets [28,29]; it has unique attributes such as
nanometer‐scale platelets, interlayer spaces, a mesoporous
structure, and a higher hydrothermal stability than mesopo‐
rous SiO2 supports [30,31]. In addition, Mt can be modified as
solid acid catalysts. The Brönsted and Lewis acid sites of Mt can
be tailored using surface adsorption [32], ion‐exchange reac‐
tions [30], pillaring, surface grafting, and impregnation [33,34].
To the best of our knowledge, the present paper is the first
report of the use of heat‐resistant Mt as a support in the prep‐
aration of acid‐activated and WOx‐loaded Mt catalysts for glyc‐
erol dehydration. The low selectivity for acrolein and catalyst
deactivation caused by coking limited large‐scale industrial
applications of catalytic dehydration of glycerol to acrolein over
acid catalysts [3]. The present study furthers our understand‐
ing of the effects of the acidic properties on catalyst selectivity
and deactivation.
In this work, we prepared the supported WOx catalysts by
acid‐activated Mt and impregnation with (NH4)10W12O41·5H2O.
The catalytic activities of the WOx‐loaded catalysts in the
gas‐phase dehydration of glycerol were studied using a
fixed‐bed microreactor. The effects of the WOx loading, the type
of acid used for Mt activation, and the concentration of the
glycerol feedstock on glycerol dehydration were investigated.
Acid catalysts such as supported heteropoly acids [12−14],
metal oxides [15], zeolites [16,17], and acidic clay minerals
[18,19] have been tested in the dehydration of glycerol to pro‐
duce acrolein. Among these, supported Keggin‐type tung‐
stophosphoric acids (H3PW12O40) on various supports are
commonly used for this conversion. Acrolein yields of
75%–86% have been achieved on H3PW12O40 catalysts using
mesoporous silicas such as Cs‐SBA‐15 [20], W‐SBA‐15 [6], and
HMS [19] as supports. Tsukuda et al. [21] prepared a 30 wt%
H3PW12O40 catalyst (Q6‐PW‐30) supported on SiO2 with a mean
pore diameter of 6 nm. The catalyst gave an acrolein yield of
64.9% at 325 °C. TiO2, Al2O3, and SiO2‐Al2O3 have also been
used as supports and gave acrolein yields of 49.6% [22], 51.3%
[23], and 19.9% [2], respectively. Yadav et al. [19] reported a
44% acrolein yield using 20 wt% H3PW12O40 supported on
K‐10 clay. These catalysts gave inconsistent catalytic perfor‐
mances and were greatly affected by the type of support used.
High surface acidities and appropriate pore structures of the
catalyst and support are critical factors in achieving high acro‐
lein yields [24]. Although the use of supported H3PW12O40 cat‐
alysts in glycerol dehydration has attracted much attention,
their strong acid sites can cause coke formation, and their low
thermal stabilities can result in decomposition at the catalyst
regeneration temperature (>500 °C), leading to significant de‐
creases in the acidity and activity of the catalyst [20]. Support‐
ed H3PW12O40 catalysts decompose at about 500 °C to P2O5 and
bulky WO3, which provide strong Lewis acid sites and are not
beneficial for the target acrolein‐forming reaction [25].
H3PW12O40 catalysts are therefore usually prepared by low
temperature calcination at 300−400 °C.
2. Experimental
2.1. Materials
Ca2+‐Mt with a cation‐exchange capacity of 66 mmol/100 g
was obtained from a deposit in Gansu, China. Ammonium
paratungstate ((NH4)10W12O41·5H2O, analytical grade, WO3
85%−90%) was purchased from the Sinopharm Chemical Rea‐
gent Co., Ltd., China. H3PO4 and CH3COOH (both analytical
grade) were obtained from the Shanghai Lingfeng Chemical
Reagent Co., Ltd., China. HCl (analytical grade, 36%−48%) and
glycerol were purchased from the Hangzhou Shuanglin Chemi‐
cal Reagent Co., Ltd., China. H2SO4 (analytical grade, 95%−98%)
was obtained from the Xilong Chemical Co., Ltd., China.
H2C2O4·2H2O (analytical grade) was obtained from the Shang‐
hai Meixing Chemical Co., Ltd., China. All reagents were used
without further purification.
WOx‐based materials are promising solid acid catalysts for
the gas‐phase dehydration of glycerol [26]. Chai et al. [7] pre‐
pared a supported WOx catalyst by impregnation of Al2O3 with
(NH4)6H2W12O40, followed by calcination at 800 °C. The catalyst
consisting of 30 wt% WO3/Al2O3 gave a 61% yield of acrolein in
10 h. Supported WOx catalysts were similarly prepared by im‐
pregnation of ZrO2, Al2O3, and SiO2 with (NH4)10W12O41·7H2O,
followed by calcination at 600 °C [14]. The WOx/ZrO2,
WOx/Al2O3, and WOx/SiO2 catalysts gave acrolein yields of 78%,
66%, and 14%, respectively. Supported WOx catalysts have
higher thermal stabilities and stronger acidities than supported
H3PW12O40 catalysts, and they have strong activities in various
catalytic reactions [27]. The catalyst acidity depends on the
WOx loading, the preparation method, and the support used.
Montmorillonite (Mt) is a three‐layer aluminosilicate clay
2.2. Catalyst preparation
2.2.1. Preparation of acid‐activated Mt
Ca2+‐Mt was activated with H2SO4, HCl, H3PO4, or CH3COOH.
In all cases, the Ca2+‐Mt was first dried at 120 °C for 12 h. The
dried Ca2+‐Mt (5 g) was added to 20 wt% acid solution (20 mL)
in a 100‐mL beaker. This mixture was heated at 80 °C for 4 h
with constant stirring and then cooled to room temperature.
The supernatant was discarded and the solid residue, i.e., ac‐
id‐activated Mt, was washed with deionized water until the pH
reached 5−6 and then dried in an air oven at 120 °C for 12 h.
The acid‐activated Mt samples were denoted by Mt‐S, Mt‐Cl,
Mt‐P, and Mt‐Ac, for H2SO4, HCl, H3PO4, and CH3COOH, respec‐
tively.

Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
1089
2.2.2. Preparation of W‐impregnated Mt catalysts
energy [36,37].
W/Mt‐P catalysts were prepared by impregnating the Mt‐P
support (3 g) with an aqueous solution of (NH4)10W12O41·5H2O
and H2C2O4·2H2O. The W concentrations of the impregnating
solutions in the final catalysts were 4, 8, 12, and 16 wt%. The
preparation procedure, using the 12 wt% catalyst as an exam‐
ple, was as follows. (NH4)10W12O41·5H2O (0.60 g) and
H2C2O4·2H2O (0.12 g) were dissolved in deionized water (6 mL)
at 80 °C. Mt‐P (3 g) was added to the solution and the resulting
mixture was kept static for 10 h at room temperature. After
impregnation, the solid product was dried at 12 °C for 12 h, and
calcined in a flow of air at 350 °C for 3 h. The samples were
labeled as xW/Mt‐P, where x is the weight percentage of W in
the final catalyst (i.e., x = 4%, 8%, 12%, and 16%).
The amount of free acid in the catalyst was determined us‐
ing a previously reported titration method [18,38]. The catalyst
(0.04 g) was dispersed in aqueous NaOH solution (0.01 mol/L,
20 mL) under stirring for 120 min at room temperature. The
resulting slurry was centrifuged and the supernatant was ti‐
trated with a standard HCl solution (0.01 mol/L) until the me‐
thyl orange indicator turned red. The amount of free acid was
calculated as
Free acid amount (mmol/g) = (C1V1  C2V2)/m
where C1 and C2 are the concentrations of the aqueous NaOH
and HCl solutions, respectively; V1 is the volume of the aqueous
NaOH solution (20 mL); V2 is the consumed volume of aqueous
HCl solution; and m is the mass of the sample (0.04 g).
2.3. Catalyst characterization
2.4. Catalytic reaction and catalyst reuse tests
Powder X‐ray diffraction (XRD) patterns were obtained us‐
ing a PANalytical X’Pert PRO diffractometer (40 kV, 40 mA)
with Ni‐filtered Cu Kα radiation (λ = 0.154 nm). Fouri‐
er‐transform infrared (FT‐IR) spectra were recorded at room
temperature in the region 400−4000 cm−1 using a Nicolet 6700
spectrometer at a spectral resolution of 4 cm−1; the samples
were KBr pellets.
N2 adsorption‐desorption isotherms were recorded at −196
°C using a Micromeritics ASAP 2020 instrument. Prior to analy‐
sis, the samples were degassed at 300 °C and 1.33  10−4 Pa for
12 h. The specific surface areas were calculated from the linear
parts of the Brunauer‐Emmett‐Teller (BET) plots. Pore size
distributions were calculated based on the adsorption branches
of the N2 isotherms using the Barret‐Joyner‐Halenda (BJH)
method.
Thermogravimetric (TG) analysis and differential thermo‐
gravimetry (DTG) were performed using a Mettler Toledo
TGA/DSC 1 instrument. The sample was heated from 30 to 800
°C at a heating rate of 10 °C/min in an air flow at 40 mL/min.
Temperature‐programmed desorption of NH3 (NH3‐TPD)
was used to determine the strengths of the acid sites on the
catalyst. The catalyst (100 mg) was degassed up to 300 °C un‐
der a N2 flow for 2 h. The catalyst was cooled to 50 °C, and NH3
gas was passed over the catalyst for 1 h to adsorb NH3 on the
surface. The physisorbed gas was removed by heating the cat‐
alyst to 800 °C at a heating rate of 10 °C/min. The total acidity
was determined from the amount of NH3 desorbed per gram of
catalyst.
Diffuse reflectance ultraviolet‐visible (DR UV‐vis) spectra
were performed using a Shimadzu UV‐2550 spectrometer in
reflection mode with a 10 Å resolution and BaSO4 as the refer‐
ence. The spectra were recorded at room temperature in the
range 200–800 nm. The Kubelka‐Munk function F(R∞) was
obtained from the DR UV‐vis absorbance and the edge energies
(Eg) were determined from the linear extrapolation of the
straight line of the slope near the absorption edge in the plot of
[F(R∞)·hν]1/η versus hν based on direct allowed transitions
[35], where η is electron transition exponent and η = 1/2 for
direct allowed transitions when monolayer WOx species cov‐
ered on the surface of catalysts, and hν is the incident photon
Catalytic tests were performed in a vertical stainless‐steel
tubular fixed‐bed microreactor of diameter 8.5 mm and height
200 mm. The catalyst (30−40 mesh) was placed in the iso‐
thermal region of the reactor and the redundant space in the
isothermal region beneath the catalyst bed was filled with
quartz and quartz wool. The reaction temperature was moni‐
tored using a thermocouple inserted into the middle of the
catalyst bed. The catalyst was pretreated at 320 °C for 1 h in a
flow of N2 (10 mL/min). A stream of glycerol aqueous solution
(5−20 wt%) was fed into the reactor at a flow rate of 0.1
mL/min using a syringe pump. The reaction products were
condensed in a cold trap kept in liquid nitrogen and collected
for analysis.
The collected products were dissolved and analyzed using a
gas chromatography‐mass spectrometry (GC‐MS; Shimadzu)
system with a capillary column (TC‐FFAP, 60 m × 0.32 mm ×
0.5 μm) and a flame ionization detector. The GC column was
operated from 40 to 100 °C with a ramping rate of 10 °C/min
for 6 min, and then from 100 to 190 °C with a ramping rate of
20 °C/min, and then kept at 190 °C for 11 min. The results of GC
analyses of the liquid phase and gas phase were combined and
used to evaluate the conversion, selectivity, and carbon bal‐
ance. The gaseous by‐products and coke deposited on the cata‐
lyst were not analyzed quantitatively and their percentages
were denoted by “others” to maintain the material balance. The
glycerol conversion, product selectivity, and product yield were
calculated as:
X (%) = (Moles of glycerol reacted)/moles of glycerol fed
into the reactor)  100
S (%) = (Moles of glycerol converted to a specific prod‐
uct)/moles of glycerol reacted)  100
Y (%) = X  S/100
where X (%) is the glycerol conversion, S (%) is the selectivity
for a specific product in the analyzed sample, and Y (%) is the
yield of a specific product.
Catalyst deactivation was investigated by calcining the used
catalyst at 500 °C in air for 4 h in a muffle furnace. The regen‐
erated catalyst was introduced into the reactor for the next run.
3. Results and discussion

1090
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
(a)
(b)
8.8 nm
22.6 nm
300
200
100
0
(1)
(1)
(2)
(2)
(3)
(3)
(4)
(4)
(5)
(5)
0
20
40
60
80
100
0.0
0.2
0.4
0.6
0.8
1.0
Pore size (nm)
Relative pressure (p/p0)
Fig. 1. N2 adsorption–desorption isotherms (a) and corresponding BJH pore size distributions (b) for Mt‐P (1), 4%W/Mt‐P (2), 8%W/Mt‐P (3),
12%W/Mt‐P (4), and 16%W/Mt‐P (5).
3.1. Catalyst characterization
because the WOx particles were distributed on the external
surface of the solid Mt‐P and blocked the pore openings [42].
3.1.1. Physical properties
The N2 adsorption–desorption isotherms and BJH pore size
distributions of Mt‐P and W/Mt‐P are shown in Fig. 1. There
were no substantial differences between the N2 adsorption
characteristics of the W/Mt‐P catalysts and Mt‐P, but the BJH
pore size distributions were broader. The isotherms of all the
catalysts showed type IV profiles and the N2 adsorption hyste‐
resis loops were H3 type. These adsorption properties suggest
that the catalysts consisted of platelet‐like particles, which
were stacked to form slit‐shaped pores [39], and could be con‐
sidered to correspond to layered structures of Mt‐P and
W/Mt‐P. The stacking of platelet particles resulted in the for‐
mation of both micropores and mesopores. The presence of
micropores was indicated by the abrupt monolayer adsorption
of N2 at low relative pressures [40]. Filling of the slit‐shaped
mesopores occurred at higher relative pressures (P/P0 > 0.4)
by multilayer adsorption of N2. Because Mt‐P and W/Mt‐P
showed both microporosity and mesoporosity, the materials
had a wide range of pore size distributions, with major pore
sizes at 8.8–22.6 nm.
3.1.2. XRD study
The XRD patterns of Ca2+‐Mt, acid‐treated Mt, and the
Mt‐P‐supported WOx catalysts are shown in Fig. 2. The un‐
treated Ca2+‐Mt sample showed the typical Mt diffraction peaks
from (001), (003), (020), (006), and (060) at 2θ = 5.89°, 17.66°,
19.90°, 36.14°, and 62.08°, respectively [43–45]. The d001 value
was 1.50 nm (2θ = 5.89°), typical of the lamellar structure of
Ca2+‐Mt [46]. Because Mt is a naturally occurring clay mineral,
the samples usually contained minor impurities such as feld‐
spar (2θ = 22.04° and 27.79°) [18,47].
The (001) diffraction peak for acid‐activated Mt was
stronger and sharper than the corresponding peak for Ca2+‐Mt;
this is in agreement with our previous observations [16]. The
results indicate that treatment of Ca2+‐Mt with 20 wt% acid did
not completely destroy the inherent layered structure of Mt. It
has been reported that concentrated H2SO4 and HNO3 can cause
complete disappearance of the (001) peak for Mt [48–50], and
the use of H3PO4 broadens and weakens the (001) reflection
[45]. The acid strength, acid concentration, and activation
time/temperature caused different degrees of destruction of
the Mt layered structure [45,50]. Acid treatment also increased
the d001 value for Mt from 1.50 to 1.56−1.57 nm because of cat‐
ionic exchange of hydrated Ca2+ cations in the interlayer spaces
of Mt with H+ ions in the aqueous acid solution [51]. Fig. 2
The specific surface areas increased from 97 m2/g for
Ca2+‐Mt to 174 m2/g for Mt‐P after activation of Ca2+‐Mt by
H3PO4 (Table 1). This increase was caused by partial destruc‐
tion of the lamellar structure of Mt [41]. Impregnation of Mt‐P
with WOx decreased the specific surface area and pore volume
Table 1
Catalyst pore structure data and acidities.
W surface density of sample *
(W/nm2)
Total acidity
(mmol NH3/g‐catalyst)
Sample
ABET (m2/g)
Pore volume (cm3/g) BJH pore size (nm)
Ca2+‐Mt
Mt‐P
4%W/Mt‐P
8%W/Mt‐P
12%W/Mt‐P
16%W/Mt‐P
97
174
135
127
121
132
0.29
0.35
0.28
0.28
0.25
0.27
13.6
8.8
22.6
22.9
22.6
23.2
—
—
1.0
2.1
3.2
4.0
—
2.60
2.83
3.24
3.71
3.35
*W surface density of sample = W loading (wt%) × 6.02 × 105/ABET of sample (m2/g)/183.84 (g/mol).

Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
1091
Al2O3 support, no crystalline WO3 particles were detected.
Monotungstate [WO4] and polytungstate [WO6] species were
both present on the Al2O3‐supported WOx catalyst. In the pre‐
sent work, no monoclinic WO3 peaks were detected for the
4%W/Mt‐P and 8%W/Mt‐P samples, probably because of good
dispersion of WOx species on the Mt surface [26].
WO3
WO2.9
Ffeldspar
WO2.72
16%W/Mt-P
12%W/Mt-P
8%W/Mt-P
3.1.3. FT‐IR study
FT‐IR spectroscopy is an effective tool for probing the
chemical and structural properties of Mt. Fig. 3 shows FT‐IR
spectra of Ca2+‐Mt, acid‐treated Mt, and the W/Mt‐P catalysts.
Absorbance bands were observed at 470 cm−1 from Si–O bend‐
ing vibrations, 520 cm−1 from Al–O–Si stretching vibrations,
and 623 cm−1 from the coupling vibrations of Al–O and Si–O.
These bands did not change or shift significantly as a result of
acid treatment and WOx impregnation. The presence of the
Al−O−Si stretching vibration at 520 cm−1 indicates that Mt re‐
tained its layered structure [18].
The extent of the changes was assessed based on the inten‐
sities of the absorbance bands at 845 cm−1 from Mg–OH–Al
bending vibrations, 918 cm−1 from Al–O–Al bending vibrations,
1039 and 1092 cm−1 from Si–O stretching vibrations, 1637
cm−1 from H–OH bending vibrations, 3420 cm−1 from the
stretching vibrations of hydrated water molecules, and 3624
cm−1 from the stretching vibrations of hydroxyl groups bonded
with octahedral Al3+ cations after acid activation [57]. The acid
treatment led to the decrease in the intensities of the bands at
845 and 918 cm−1 for the acid‐activated Mt samples, and the
intensities of these bands for Mt‐P and Mt‐Cl weakened more
4%W/Mt-P
Mt-S
Mt-Cl
Mt-P
Mt-Ac
001
020
F
Ca2+-Mt 060
006
F
003
10
20
30
40
50
60
2/(o )
Fig. 2. XRD patterns of Ca2+‐Mt, acid‐treated Mt, and WOx‐loaded Mt‐P.
shows that the intensity of the (001) peak in acid‐activated Mt
decreased with increasing strength of the acid used, and the
(003), (020), (006), and (060) reflections were weakened to
different extents or even disappeared after acid activation.
These results indicate that the strong acid H2SO4 attacked the
layered structure most (Mt‐S pattern). The strong acids H2SO4
and HCl damaged more layers of Mt than did the moderate acid
(H3PO4) and weak acid (CH3COOH). The intensities of the feld‐
spar peaks (2θ = 22.04° and 27.79°) also decreased because of
dissolution of this impurity during acid treatment.
For the Mt‐P samples impregnated with WOx, the (001) peak
shifted toward higher diffraction angles, indicating a decrease
in the interlayer space. This shift is probably caused by the
removal of water molecules adsorbed in the Mt interlayer
spaces during calcination at 350 °C in the preparation of
W/Mt‐P. The (001) peak of W/Mt‐P also became broader and
less intense after impregnation with WOx. This is attributed to
the deposition of WOx nanoparticles on the external surfaces of
Mt [42] and/or changes in the pore structure of Mt after im‐
pregnation and calcination [52,53].
Al-OH
3624
Si-O
1039
Al-O
W-O-W
880
Si-O
Si-O
623
470
Si-O
1092
H-OH
3420
(9)
(8)
(7)
(6)
(5)
(4)
For W/Mt‐P, new peaks appeared at 2θ = 10.71° and 39.42°
from WO2.72 [54], at 2θ = 15.79°, 18.63°, 19.76°, 27.69°, and
30.66° from WO2.9, and at 2θ = 21.54°, 21.90°, 26.45°, and
36.14° from WO3 [54,55]. The intensities of these WOx peaks
increased with increasing W loading. When the W loading was
increased from 12% to 16%, the W surface density increased
from 3.2 to 4.0 W/nm2 (Table 1). A very weak diffraction peak
was observed at 2θ = 23.57°−24.11°, indicating the formation
of a stable monoclinic crystalline WO3 phase (mono‐WO3) [56].
The type of WOx species present on the supported catalysts
depends on the support used. Kim et al. [37] prepared sup‐
ported WO3 catalysts by impregnation of Al2O3, Nb2O5, and TiO2
supports with (NH4)10W12O41·5H2O. The W surface densities of
the Nb2O5 and TiO2 supports were greater than 4.0 W/nm2 and
isolated crystalline WO3 particles were formed, whereas for the
(3)
(2)
(1)
845
Mg-OH
1480
1637 -OH
H-OH
845
918
520
Al-O
500 960
-Al
918
Al-O-Al
4000
3000
1500
1000
800
Wavenumber (cm1)
Fig. 3. FT‐IR spectra of Ca2+‐Mt (1) and the catalysts Mt‐Ac (2), Mt‐P (3),
Mt‐Cl (4), Mt‐S (5), 4%W/Mt‐P (6), 8%W/Mt‐P (7), 12%W/Mt‐P (8),
and 16%W/Mt‐P (9).

1092
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
obviously than for Mt‐Ac and Mt‐S. The results indicate that
more octahedral Al3+ cations leached from Mt‐Cl and Mt‐P than
from Mt‐S and Mt‐Ac [46]. This was confirmed by the decrease
in the intensity of the band at 3624 cm−1 for Mt‐Cl and Mt‐P,
probably caused by the removal of octahedral Al3+ cations as a
result of loss of water and the hydroxyl groups coordinated
with them [41]. For Mt‐Cl, acid treatment significantly in‐
creased the intensity of the band at 1092 cm−1. This shows that
more Si−O−Si and Al−O−Si bonds were broken and therefore
generation of WOx with various stoichiometric compositions in
the temperature range 150–400 °C [63,64]. The endothermic
peak at 559 °C is ascribed to removal of the structural hydroxyl
groups of Mt [58]. The mass loss for uncalcined 12%W/Mt‐P
was 11.3% in the temperature range 30–630 °C (Table 2),
which is equal to the theoretical value (11.2%) for loss of
(NH4)10W12O41·5H2O. In the temperature range between 350
and 500 °C, the mass loss was only 1.4%. This is similar to the
mass loss of 1.2% for decomposition of pure
(NH4)10W12O41·3.5H2O in air in the temperature range 377–480
°C [63]. The decomposition of supported (NH4)10W12O41·5H2O
on the 12%W/Mt‐P sample was therefore almost complete at
350 °C. Madarász et al. [65] reported that WO3 crystallization
started from the amorphous state during thermal decomposi‐
tion of pure (NH4)10W12O41·4H2O in calcination between 400
and 500 °C in air. To avoid agglomeration of WOx particles and
recrystallization of WO3 caused by calcination at higher tem‐
peratures [63,66], in the present work, 350 °C was used as the
calcination temperature for the WOx‐loaded Mt catalysts. The
decomposition of (NH4)10W12O41·5H2O leads to formation of
WOx, NH3, and H2O:
+
+
additional Si−OH2 and Al−OH2 groups were formed on the
surface of Mt‐Cl [58]. Mt‐Cl therefore had a larger number of
+
+
Brönsted acid sites (Si−OH2 , Al−OH2 ). A larger number of
exposed framework Al3+ ions provide more Lewis acid sites
[59]. For Mt‐S and Mt‐Ac, the intensity of this band showed
little change. Based on these observations, it is reasonable to
conclude that the amount of octahedral Al3+ ions released from
the Mt lamellar structure during acidification decreased in the
order Mt‐Cl > Mt‐P > Mt‐S/Mt‐Ac. The larger the number of
octahedral Al3+ ions that leached from Mt, the more Lewis acid
sites became available on the Mt surface [58].
The intensities of the absorption band at 3420 cm−1 for the
W/Mt‐P samples were lower than the intensity of this band for
the parent Mt‐P. This is because the adsorbed water molecules
in the interlayer spaces of Mt were removed during calcination
at 350 °C. The additional weak band at 880 cm−1 is attributed to
the W−O−W vibration mode [60,61]. These results suggest the
presence of WOx in the W/Mt‐P samples. These findings are in
agreement with the XRD data. For the W/Mt‐P samples, the
absorption band at 918 cm−1 (Al–O–Al) weakened compared
with that for Mt‐P and the band at 845 cm−1 (Mg–OH–Al) dis‐
appeared. These results verify the effectiveness of supporting
WOx on Mt‐P and the role of Mt‐P in creating a coordination
microenvironment for these atoms.
(NH4)10W12O41·5H2O + 6(x − 3)O2 = 12WOx + 10NH3 + 10H2O
3.1.5. DR UV‐vis spectra study
DR UV‐vis spectra can provide information on local struc‐
tures (isolated, polymeric, clusters, and bulk structures) of co‐
valent‐bridged W–O–W bonds on WOx‐loaded catalysts [36].
Fig. 5 shows DR UV‐vis spectra of the WOx‐loaded catalysts. All
the catalysts showed an absorption band at 256 nm, which is
assigned to isolated [WO4] tetrahedral species or
WO6‐containing clusters [27,54]. The absorption band shoulder
at about 280−400 nm can be attributed to the presence of oc‐
tahedral [WO6] polytungstate species and crystalline WO3
[27,36]. The intensity of the absorption shoulder increased
with increasing W loading. The appearance of an absorption
band for crystalline WO3 in the spectrum of the 12%−16%
W‐loaded catalysts confirms the XRD results.
The specific WOx species in the catalysts were investigated
by the Eg values. For 8%W/Mt‐P and 12%W/Mt‐P, the Eg val‐
ues were 3.4 and 3.2 eV, respectively; these originate from pol‐
ytungstate [WO6] species. The Eg value for 4%W/Mt‐P (3.5 eV)
was slightly higher than those for 8%W/Mt‐P and
12%W/Mt‐P; this is because of the presence of alternating
WO4/WO6 clusters [36]. The Eg value of 2.8 eV for 16%W/Mt‐P
corresponds to the characteristic absorption edge energy of
3.1.4. TG‐DTG analysis
The TG‐DTG curves of uncalcined 12%W/Mt‐P are shown in
Fig. 4. The DTG curve shows three endothermic stages in an air
flow. The first endothermic peak, at 61 °C, corresponds to the
removal of physically adsorbed water [62]. The second‐stage
endothermic peaks at 215, 270, and 294 °C represent decom‐
position of the impregnated precursor (NH4)10W12O41·5H2O.
This stage involves release of crystal water and NH3 and the
100
96
92
88
84
0.2
0.0
DTG
TG
Table 2
Mass losses (%) from catalysts before and after reactions.
559 oC
Temperature region (°C)
Sample
-0.2
-0.4
-0.6
30−150 150−350 350−500 500−630 150−500 150−630
Fresh Mt‐P
Used Mt‐P
Uncalcined
12%W/Mt‐P
Fresh
12%W/Mt‐P
Used
12%W/Mt‐P
5.9
0.6
0.8
1.1
0.9
3.8
1.3
5.0
1.7
4.9
3.0
9.9
215 oC
294 oC
270 oC
61 oC
3.3
4.0
1.3
5.0
0.8
1.2
1.4
0.9
2.9
1.6
1.7
3.5
6.4
1.7
4.1
8.0
3.4
7.6
200
400
Temperature (oC)
600
Fig. 4. TG‐DTG curves of uncalcined 12%W/Mt‐P.

Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
1093
(a)
(b)
24
16
8
256 nm

(4)
(1)
(3)
(4)
(3)
(2)
(1)
(2)
0
500
400
300
200
2
3
4
5
Photon energy (eV)
Wavelength (nm)
Fig. 5. DR UV‐vis spectra of WOx‐loaded catalysts (a) plotted as Kubelka‐Munk function vs wavelength and (b) plotted as [F(R∞)·hν]2 vs photon energy
(1) 4%W/Mt‐P; (2) 8%W/Mt‐P; (3) 12%W/Mt‐P; (4) 16%W/Mt‐P.
three‐dimensional crystalline WO3 [67]. These results indicate
that dispersed WOx species formed W–O–W bridging bonds
groups on the external Mt surface. The isoelectric point (pI) of
Mt is about 1.47 [68]. In the acidic range (pH < pI), the Al–OH
and Si–OH groups on the Mt surface can take up protons to
form Mt‐AlOH2 and Mt‐SiOH2 . These ions provide Brönsted
acid sites on Mt.
The amount of free acid in the acid‐activated Mt decreased
in the order Mt‐P > Mt‐Ac > Mt‐S > Mt‐Cl (Table 3). The free
acid arises from H+ ions adsorbed on the external Mt surface.
There is therefore no linear relationship between the amount
of free acid and the total acidity.
between
neighboring
WOx
groups
to
form
+
+
[WO4/WO6]‐containing clusters, two‐dimensional [WO6] poly‐
tungstates, and isolated three‐dimensional WO3 crystals with
increasing W loading from 4% to 16%. WOx species with dif‐
ferent coordination structures provided the WOx‐loaded cata‐
lysts with Brönsted and Lewis acid sites.
3.1.6. Generation of acidity on Mt
3.1.6.1. Acid treatment
3.1.6.2. Supported tungsten oxides
Raw Ca2+‐Mt is a layered aluminum silicate with a 2:1 lay‐
ered structure predominantly consisting of Al–O octahedral
sheets sandwiched by two Si–O tetrahedral sheets. The hy‐
drated Ca2+ cations are adsorbed in the interlayer spaces of Mt
to balance the negatively charged clay surfaces [29]. Acid
treatment can be used to increase the acidity of the raw Ca2+‐Mt
(Scheme 1). The type of acid used, acid concentration, and reac‐
tion conditions (temperature and time) affect the acid sites and
the acidity of the acid‐activated Mt [45,50,58].
The WOx loading on Mt‐P determines the type of Brönsted
and Lewis acid sites. The DR UV‐vis spectra show that surface
monotungstate species (coordinated structure: [WO4/WO6]
clusters), polytungstates (coordinated structure: [WO6]), and
isolated crystalline WO3 particles were deposited on the sur‐
faces of the WOx‐loaded catalysts (Scheme 1). The polytung‐
state WOx species were well dispersed on the catalyst surfaces
and provided Brönsted acid sites because the extended net‐
work of polytungstate WOx species had an excess negative
charge and could therefore bind protons [69,70]. The mono‐
In acid treatment, first, H+ ions are intercalated into the in‐
terlayer spaces of Mt through cation exchange with Ca2+ ions.
For acid‐activated Mt, the acidity is mainly provided by H+ ions
(Brönsted acid sites) hosted by the interlayer spaces of Mt.
When Mt is attacked by an acid, octahedrally coordinated
metal ions such as Fe3+ and mainly Al3+ are preferentially re‐
leased from the Mt lamellar structure, and the lamellar struc‐
ture of Mt partially decomposes [50]. Some of the unsaturated
Al3+ ions (the framework Al sites) are therefore exposed on the
external Mt surface. Mt‐Cl and Mt‐P have the greater numbers
of Lewis acid sites than Mt‐S and Mt‐Ac because of leaching of
octahedral Al3+ ions from the Mt lamellar structures. These
observations are confirmed by the IR spectra. The Lewis acidity
of acid‐activated Mt generally arises from exposed Al3+ ions
[59]. Negatively charged Si–O groups are formed by rupture of
bonds between Al–O and Si. The negatively charged Si–O
groups attract protons, creating Mt Brönsted acid sites.
Acid treatment
Supporting WOx
Mt
Acid-activated Mt
WOx-supported Mt
Bronsted acid sites
Si
O
H+(H2O)
Proton
OH2+
Al
Ca2+(H2O)
OH
Polytungstate [WO6]
O

W
Exposed Al3+
O
O
O
O
Monotungstate [WO4]
O
Lewis acid sites
W
W
Crystalline WO3
Scheme 1. Generation of acid sites by acid treatment and WOx loading
on Mt.
H+ ions can also be adsorbed by ionized Al–O− and Si–O−

1094
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
Table 3
Catalytic performances of acid‐treated and WOx‐supported Mt catalysts.
Aqueous glycerol
(wt%)
Free acid
(mmol/g)
X a
(%)
Y b
(%)
Selectivity (%)
Catalyst
Acrolein
69.1
72.8
63.3
65.6
62.3
67.3
71.8
67.0
81.8
78.3
64.9
Acetol
12.3
11.1
15.6
13.6
13.4
13.5
10.4
11.4
6.3
Propenol
6.1
Acetaldehyde
Others
8.0
5.4
7.0
4.6
5.4
3.9
4.4
3.4
5.1
4.3
5.7
Mt‐Ac
Mt‐P
20
20
20
20
20
20
20
20
15
10
5
1.69
1.73
1.58
1.66
—
—
—
—
68.2
87.5
94.6
65.7
65.6
69.4
93.7
75.7
89.6
76.4
72.8
47.1
63.7
59.9
43.1
40.9
46.7
67.3
50.7
73.3
59.8
47.2
4.5
4.8
4.7
6.0
6.9
6.1
4.8
6.0
2.9
5.5
8.9
5.9
Mt‐Cl
Mt‐S
9.4
10.2
12.0
9.2
8.6
12.2
3.9
4%W/Mt‐P
8%W/Mt‐P
12%W/Mt‐P
16%W/Mt‐P
12%W/Mt‐P
12%W/Mt‐P
12%W/Mt‐P
—
—
—
6.3
10.8
5.7
9.7
Reaction conditions: catalyst 0.4 g, aqueous glycerol 0.1 mL/min, N2 carrier gas 10 mL/min, 320 °C, 3 h.
a Conversion of glycerol.
b Yield of acrolein.
tungstate WOx species and crystalline WO3 particles provided
Lewis acid sites because of the lack of delocalized negative
charges for the formation of Brönsted acid sites [71].
temperature for desorption of NH3 adsorbed on acid sites. Acid
sites are defined as weak, medium, and strong at desorption
temperatures of 50−200, 200−400, and 400−800 °C, respec‐
tively [58,72]. Both Mt‐P and 12%W/Mt‐P showed bimodal
acid strengths, reflected by the presence of two desorption
stages, at low‐ and high‐temperature regions of about 100−400
and 400−800 °C. Tong et al. [58] suggested that the higher de‐
sorption temperature (>400 °C) can be ascribed to desorption
of NH3 adsorbed on framework Al sites (Lewis acid sites). The
results suggest that Mt‐P and 12%W/Mt‐P have acid sites with
a broad distribution of acid strengths from weakly to strongly
acid.
In general, the coordinated WOx structures depend on the
WOx loading (WOx surface coverage), the preparation process,
and the support used [7]. For a support, there is a dispersion
threshold for forming monolayer‐state WOx species on the
support surface. When the WOx loading is far lower than the
dispersion threshold, WOx is usually present as monotungstate
species. In contrast, when the WOx loading is close to or over
the dispersion threshold, WOx forms polytungstate species and
finally crystalline WO3 [27].
The W surface densities of the catalysts are listed in Table 1.
In this work, based on the XRD and DR UV‐vis spectra, the dis‐
persion threshold was about 3.2 W/nm2 because crystalline
WO3 was detected on 12%W/Mt‐P (3.2 W/nm2). 16%W/Mt‐P
(4.0 W/nm2) provided more Lewis acid sites than 12%W/Mt‐P
because more mono‐WO3 crystals were formed on the surface
of 16%W/Mt‐P. For 12%W/Mt‐P and 16%W/Mt‐P, the loaded
WOx species afforded both Brönsted acid sites and Lewis acid
sites. For 4%W/Mt‐P (1.0 W/nm2) and 8%W/Mt‐P (2.1
W/nm2), which have low surface coverages by WOx, no
mono‐WO3 crystalline particles were detected. The WOx species
on these catalysts therefore provided mainly Lewis acid sites.
This is similar to the results reported by Wachs et al. [35]. They
loaded WOx on Al2O3, TiO2, and ZrO2 supports. Surface mono‐
tungstate species (Lewis acid sites) were predominant at low
W surface coverages (<2 W/nm2). When the W surface cover‐
age was intermediate, i.e., 3−4 W/nm2, surface polytungstate
species became predominant and both Brönsted and Lewis acid
sites were present. At higher W surface coverages (>4 W/nm2),
crystalline WO3 particles predominated and the number of
Lewis acid sites increased.
The acid strength changed after loading WOx on Mt‐P.
Low‐temperature desorption occurred at 240 °C for
12%W/Mt‐P, whereas the corresponding desorption for Mt‐P
was centered near 213 °C. Loading WOx on Mt‐P shifted the
desorption temperature from 213 to 240 °C. These results in‐
dicate that the strengths of the overlapped acid sites at 50−200
°C (weak) and 200−400 °C (medium) on 12%W/Mt‐P in‐
creased. The second desorption occurred at 653 °C for
12%W/Mt‐P, whereas the corresponding desorption for Mt‐P
653 oC
700 oC
240 oC
(2)
213 oC
(1)
3.1.7. Acid strength
200
400
600
800
Temperature (oC )
NH3‐TPD was used to compare the acid strengths of the
W‐impregnated catalyst and non‐impregnated acid‐treated
Mt‐P (Fig. 6). A greater acid strength corresponds to a higher
Fig. 6. NH3‐TPD profiles for Mt‐P (1) and 12%W/Mt‐P (2).

Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
1095
was centered near 700 °C; this indicates that loading WOx on
Mt‐P reduced the strength of the strong acid sites (400−800 °C)
in the case of 12%W/Mt‐P.
75.7%, with an acrolein yield of 50.7%. The results suggest that
the catalytic activity reaches a maximum when the W loading
reaches a dispersion threshold. Chai et al. [7] similarly reported
that the selectivity for acrolein increased with increasing W
surface density up to 4 W/nm2, and then declined with further
increases in the W surface density up to 16 W/nm2 for a
WOx‐impregnated Al2O3 or ZrO2 catalyst. We also found that the
acetol selectivities of the samples with low W loadings, i.e.,
4%W/Mt‐P (1.0 W/nm2) and 8%W/Mt‐P (2.1 W/nm2), and
that with the highest W loading, i.e., 16%W/Mt‐P (4.0 W/nm2),
were higher than that of 12%W/Mt‐P (3.2 W/nm2). This can be
attributed to the presence of more Lewis acid sites on the sur‐
faces of 4%W/Mt‐P and 8%W/Mt‐P because of the presence of
monotungstate species, and on 16%W/Mt‐P because of crystal‐
line WOx species [3,7]. In the present work, a 12% W loading
was therefore used for the preparation of a catalyst for acrolein
production.
It was also found that 4%W/Mt‐P and 8%W/Mt‐P gave
lower glycerol conversions and lower acrolein selectivities than
the parent Mt‐P. This can be attributed to the following factors:
(1) low W loadings on 4%W/Mt‐P and 8%W/Mt‐P, resulting in
more Lewis acid sites; and (2) changes in the textural proper‐
ties such as decreased surface areas and pore volumes and a
broader distribution of pore sizes than in the case of Mt‐P (Ta‐
ble 1).
3.2. Catalytic activity
3.2.1. Effect of type of acid used in activation
Mt samples activated by 20 wt% H2SO4, HCl, H3PO4, or
CH3COOH were used as catalysts for the gas phase dehydration
of glycerol to acrolein (Table 3). The main reaction product was
identified as acrolein. Acetol, acetaldehyde, propenol, and cyclic
ethers were also detected. The glycerol conversions showed
that the catalytic activities of the acid‐activated Mt decreased in
the order Mt‐Cl > Mt‐P > Mt‐Ac > Mt‐S. The highest glycerol
conversion rate, i.e., 94.6%, with an acrolein selectivity of
63.3%, was achieved using Mt‐Cl, as a result of its high acidity.
The selectivity for acetol achieved with Mt‐Cl was also higher
than those obtained with Mt‐S, Mt‐P, and Mt‐Ac. This is because
Mt‐Cl has the largest number of Lewis acid sites. Mt‐P gave the
highest selectivity for acrolein, i.e., 72.8%, with a maximum
acrolein yield of 63.7%. There was no linear relationship be‐
tween the amount of free acid and glycerol conversion or acro‐
lein selectivity. These findings indicate that the catalytic activity
depended on the acidity and acid sites, and free acids did not
play a primary role in glycerol conversion or affect the acrolein
yield [18].
3.2.3. Effect of glycerol feedstock concentration
3.2.2. Effect of W surface density
Feedstocks with various glycerol concentrations (5, 10, 15,
and 20 wt%) were used in glycerol dehydration over the
12%W/Mt‐P catalyst. It was observed that increasing the glyc‐
erol concentration of the aqueous solution from 5 to 20 wt%
led to an increase in the glycerol conversion from 72.8% to
93.7%; the acrolein selectivity increased from 64.9% to 81.8%,
then decreased to 71.8%, and the acrolein yield increased from
47.2% to 73.3%, and then decreased to 67.3% (Table 3). Acro‐
lein is produced from secondary carbocations and is thermo‐
dynamically controlled; acetol, the major by‐product, is formed
from primary carbocations and is kinetically controlled [2,75].
According to Le Chatelier’s principle, water inhibits glycerol
dehydration because it is a product of glycerol dehydration.
More water shifts the equilibrium and partially suppresses
glycerol dehydration [18]. Aqueous glycerol of low concentra‐
tion (≤10 wt%), i.e., containing more water, gives lower glycer‐
ol conversions and acrolein yields. However, aqueous glycerol
of high concentration can promote hydrogen transfer [76],
leading to more cracking and coking. This results in some pore
channels becoming inaccessible because of deposition of car‐
bonaceous materials and coverage of some active catalytic ac‐
tive sites with cokes [19,77]. The glycerol conversion and acro‐
lein selectivity obtained using a 20 wt% glycerol solution were
therefore lower than those obtained using a 15 wt% glycerol
solution. Based on the glycerol conversion rate and acrolein
selectivity, 15 wt% aqueous glycerol was therefore used to
evaluate the catalyst reusability.
The W surface density (W/nm2) on the catalyst affects the
catalytic performance in the gas‐phase dehydration of glycerol
by changing the coordination structures of WOx species (i.e., the
amount and acidity of acid sites) in the catalyst [7,35]. The
glycerol conversion depends on the total acidity, and the selec‐
tivity for acrolein is related to the presence of Brönsted acid
sites [3]. The total acidities of the WOx‐loaded Mt catalysts are
listed in Table 1. The amount of acid in the Mt‐P sample was
2.60 mmol NH3/g, which is lower than the reported value (4.30
mmol NH3/g) for the common clay K‐10 Mt [73]. Table 1 shows
that when the W surface density increased from 1.0 to 3.2
W/nm2, the total acidity increased from 2.83 to 3.71 mmol
NH3/g. When the W surface density was further increased to
4.0 W/nm2, the total acidity decreased to 3.35 mmol NH3/g.
The 12%W/Mt‐P sample had the highest acidity. The trends in
the total acidities of the Mt‐P catalysts were similar to those for
WOx‐loaded Pd/Al2O3 catalysts prepared using the same pre‐
cursor, i.e., (NH4)10W12O41, for impregnation [74]. The differ‐
ences among the total acidities led to large differences among
the glycerol conversions.
The glycerol conversions and acrolein selectivities over the
WOx‐loaded Mt catalysts increased with increasing W surface
density from 1.0 W/nm2 in 4%W/Mt‐P to 3.2 W/nm2 in
12%W/Mt‐P because of the increase in the acidity and the
number of Brönsted acid sites; the results depend on appropri‐
ate W loadings (Section 3.1.6.2). 12%W/Mt‐P gave a maximum
glycerol conversion of 93.7%, with an acrolein yield of 67.3%.
However, for 16%W/Mt‐P, in which the W surface density in‐
creased to 4.0 W/nm2, the glycerol conversion decreased to
3.3. Deactivation and reuse of catalyst

1096
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
Catalyst deactivation was studied based on reaction for
and 12%W/Mt‐P catalysts. This mass loss was caused by the
removal of organic species on the catalysts. The amounts of
organic species deposited on the Mt‐P and 12%W/Mt‐P cata‐
lysts are listed in Table 2. The mass losses in the temperature
region 150−350 °C were about 1.1% and 1.2% for the used
Mt‐P and 12%W/Mt‐P catalysts, respectively, whereas in the
temperature regions 350−500 and 500−630 °C, the mass losses
were 3.8% and 5.0%, respectively, for the used Mt‐P, i.e., higher
than those of 2.9% and 3.5%, respectively, for the used
12%W/Mt‐P. The total mass loss for the used Mt‐P catalyst
(9.9%) was higher than that for the used 12%W/Mt‐P catalyst
(7.6%) in the temperature region 150−630 °C. Dalil et al. [77]
studied carbon deposition during gas‐phase glycerol dehydra‐
tion over a WO3/TiO2 catalyst. TG analysis at 500 °C showed
that the mass loss of the coked WO3/TiO2 catalyst was 4.6%
(the mass fraction of carbon); this is higher than that of the
used 12%W/Mt‐P catalyst (4.1% at 500 °C) and lower than that
of the used Mt‐P (4.9% at 500 °C) in the present work. The re‐
sults suggest that deactivation of the Mt‐P catalyst is easier
than 12%Mt‐P catalyst deactivation because the strong acid
sites of Mt‐P are stronger than those of 12%Mt‐P, as shown by
the NH3‐TPD data. The catalyst stability was therefore related
to the strength of the strong acid sites.
1−10 h over fresh and regenerated 12%W/Mt‐P catalysts to
determine the stability of the catalyst in glycerol dehydration
(Fig. 7). For the fresh 12%W/Mt‐P catalyst (Fig. 7(a)), glycerol
conversion increased from 43.8% to 93.0% with increasing
time‐on‐stream in the first 4 h, and then decreased to 78.1% at
4−6 h. The glycerol conversion remained stable
(77.8%−78.1%) at 6–8 h, and then decreased to 66.5% by 10 h.
The acrolein selectivity increased from 55.2% to 83.0% in 1 to
2 h and then remained unchanged at 6−8 h. At 8−10 h, the
acrolein selectivity decreased slowly to 77.3%. A comparative
test was performed using fresh Mt‐P to determine the effect of
WOx species on the catalyst stability. Deactivation of the fresh
Mt‐P catalyst was more serious than that of the fresh
12%W/Mt‐P catalyst. After 10 h, the fresh Mt‐P catalyst
achieved only 55.4% glycerol conversion and 51.0% selectivity
for acrolein. The difference between the acid strengths, deter‐
mined from the NH3‐TPD data, led to different catalytic per‐
formances. Sites that were too strongly acidic accelerated Mt‐P
deactivation [3]. These results are confirmed by the TG‐DTG
curves for Mt‐P and 12%W/Mt‐P before and after the reactions
(Fig. 8).
The 12%W/Mt‐P catalyst was regenerated by treating the
used catalyst in a muffle furnace at 500 °C for 4 h. After 10 h,
the regenerated 12%W/Mt‐P catalyst gave 53.3% glycerol
conversion and 50.8% selectivity for acrolein (Fig. 7(b)). These
results show a decrease in the catalytic performance during the
reaction period from 2 to 10 h compared with that of the fresh
12%W/Mt‐P catalyst.
Catalyst deactivation can be ascribed to coke deposition on
the catalyst surface, as shown by TG‐DTG analysis (Fig. 8). Two
major mass‐loss stages were detected in the TG‐DTG profiles of
the fresh Mt‐P and 12%W/Mt‐P catalysts. The mass losses at
78 and 605 °C for W/Mt‐P and at 63 and 582 °C for
12%W/Mt‐P correspond to the release of physically adsorbed
water molecules and dehydration of the Mt structure, respec‐
tively [62]. This indicates that no organic matter was present
on the fresh Mt‐P and 12%W/Mt‐P catalysts. After their use in
the reactions, rapid mass loss in the temperature region
150−630 °C was observed in the TG‐DTG curves of the Mt‐P
3.4. Reaction mechanism
Reaction mechanisms for gas‐phase dehydration of glycerol
over different types of acidic and basic catalysts have been
proposed by several research groups [3,19]. The reaction
pathways shown in Scheme 2 are postulated based on the reac‐
tion mechanism proposed by Tsukuda et al. [21] and the cata‐
lytic performances of the acid‐activated and WOx‐loaded cata‐
lysts in the present work. A proton from a Brönsted acid site
attacks the secondary hydroxyl group of glycerol to form the
unstable intermediate 1,3‐dihydroxypropene, which tautomer‐
izes to 3‐hydroxypropanal. 3‐Hydroxypropanal is unstable and
is dehydrated to acrolein via a keto‐enol rearrangement. The
unstable 3‐hydroxypropanal can also decompose to formalde‐
hyde and acetaldehyde. The formaldehyde reacts continuously
with glycerol to form other by‐products such as 1,3‐dioxan‐5‐ol
100
90
80
70
60
50
40
100
90
80
70
60
50
40
90
80
70
60
50
40
70
60
50
40
(a)
(b)
0
2
4
6
8
10
0
2
4
6
8
10
Time on stream (h)
Time on stream (h)
Fig. 7. Deactivation and reuse of 12%W/Mt‐P. (a) Fresh catalyst; (b) regenerated catalyst. Reaction conditions: glycerol concentration 15 wt%, aque‐
ous glycerol 0.1 mL/min, N2 carrier gas 10 mL/min, 320 °C.

Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
1097
100
98
96
94
92
90
100
98
96
94
92
90
100
0.2
Fresh Mt-P
Used Mt-P
TG
0.0
DTG
98
96
94
92
90
100
98
96
94
92
90
0.0
-0.2
-0.4
-0.6
-0.8
605 oC
69 oC
-0.2
-0.4
DTG
TG
78 oC
536oC
400
200
400
600
800
200
600
Temperature ( oC )
Temperature ( oC )
0.2
Fresh 12W%/Mt-P
Used 12W%/Mt-P
TG
0.0
DTG
0.0
582 oC
DTG
67 oC
-0.2
-0.4
-0.6
596 oC
314 oC
-0.2
-0.4
TG
63 oC
559 oC
200
400
600
800
200
400
600
Temperature ( oC )
Temperature ( oC )
Fig. 8. TG‐DTG curves of Mt‐P and 12%W/Mt‐P before and after reactions.
cific surface area from 97 m2/g for raw Ca2+‐Mt to 174 m2/g for
Mt‐P.
Bronsted acid site
-H2O
O
OH
Hydroxypropanal
HO
OH
1,3-Dihydroxypropene
WOx‐loaded Mt‐P catalysts were prepared by impregnation
of Mt‐P with 4−16 wt% W. The W loading affected the type of
WOx species on the catalyst surface. The addition of 12% W to
Mt‐P gave suitable acid sites and appropriate acid strengths for
achieving higher catalytic activity and better resistance to de‐
activation compared with the parent Mt‐P and other W/Mt‐P
catalysts. The highest acrolein yield, i.e., 73.3%, and a glycerol
conversion of 89.6% were achieved at 320 °C using an aqueous
15 wt% glycerol solution as the feedstock. When more than
12% W was loaded on the Mt‐P, a monoclinic crystalline WO3
phase was formed.
The catalytic performance of the catalysts was affected by
the type of acid used in the activation, the surface coverage by
W, and the glycerol concentration in the feedstock. Mt treated
with H3PO4 (Mt‐P) was highly selective for acrolein in
gas‐phase dehydration of glycerol. A W/Mt‐P catalyst with a W
surface coverage of 3.2 W/nm2 and a feedstock concentration
of 15 wt% glycerol was the best choice for glycerol dehydration
to acrolein.
-H2O
O
O
OH
HO
OH
OH
O
Glycerol
Hydroxyacetone
Acrolein
Formaldehyde
Acetaldehyde
H2
Other by-products
such as H2
OH
OH
-H2O
Lewis acid site
HO
Propenol
2,3-Dihydroxypropene
Scheme 2. Reaction pathways in catalytic dehydration of glycerol over
acid catalysts.
and hydrogen [78]; therefore formaldehyde is not detected in
the products. The formed hydrogen reacts with acrolein to
form propenol [79]. Protonation of a terminal hydroxyl group
of glycerol at a Lewis acid site results in formation of the unsta‐
ble intermediate 2,3‐dihydroxypropene, which tautomerizes to
acetol.
4. Conclusions
The catalyst suffered from deactivation and the regenerated
catalyst showed decreased activity. The strength of the acid
sites was responsible for the catalyst stability. High‐strength
strong acid sites resulted in easy deactivation of Mt‐P. Further
investigations are needed to better understand glycerol dehy‐
dration and improve the stability and dispersion of WOx species
in WOx‐loaded Mt catalysts.
Acid treatment of Mt with 20 wt% H2SO4, HCl, H3PO4, or
CH3COOH at 80 °C for 4 h did not substantially alter the layered
structure of Mt. Such acid activation removed some of the im‐
purities in the raw Mt clay and converted Ca2+‐Mt to H+‐Mt by
ion exchange. It also led to leaching of some octahedral Al3+
ions from the Mt lamellar structure and an increase in the spe‐

1098
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
178−186.
References
[33] P. C. Joshi, M. F. Aldersley, J. W. Delano, J. P. Ferris, J. Am. Chem.
Soc., 2009, 131, 13369−13374.
[1] A. B. F. Moreira, A. M. Bruno, M. M. V. M. Souza, R. L. Manfro, Fuel
Process. Technol., 2016, 144, 170–180.
[2] T. H. Kang, J. H. Choi, Y. Bang, J. Yoo, J. H. Song, W. Joe, J. S. Choi, I. K.
Song, J. Mol. Catal. A, 2015, 396, 282–289.
[3] B. Katryniok, S. Paul, V. Bellière‐Baca, P. Rey, F. Dumeignil, Green
Chem., 2010, 12, 2079−2098.
[4] A. Seretis, P. Tsiakaras, Fuel Process. Technol., 2016, 142, 135–146.
[5] Z. Q. Wang, Z. Zhang, W. J. Yu, L. D. Li, M. H. Zhang, Z. B. Zhang, Fuel
Process. Technol., 2016, 142, 228–234.
[34] C. H. Zhou, Appl. Clay Sci., 2011, 53, 97−105.
[35] I. E. Wachs, T. Kim, E. I. Ross, Catal. Today, 2006, 116, 162−168.
[36] E. I. Ross‐Medgaarden, I. E. Wachs, J. Phys. Chem. C, 2007, 111,
15089−15099.
[37] T. Kim, A. Burrows, C. J. Kiely, I. E. Wachs, J. Catal., 2007, 246,
370−381.
[38] S. K. Bharadwaj, P. K. Boruah, P. K. Gogoi, Catal. Commun., 2014,
57, 124−128.
[39] M. Kruk, M. Jaroniec, Chem. Mater., 2001, 13, 3169−3183.
[40] P. Kumar, R. V. Jasra, T. S. G. Bhat, Ind. Eng. Chem. Res., 1995, 34,
1440−1448.
[41] T. H. Wang, T. Y. Liu, D. C. Wu, M. H. Li, J. R. Chen, S. P. Teng, J.
Hazard. Mater., 2010, 173, 335−342.
[42] J. P. Kumar, P. V. R. K. Ramacharyulu, G. K. Prasad, B. Singh, Appl.
Clay Sci., 2015, 116, 263−272.
[43] W. H. Yu, Q. Q. Ren, D. S. Tong, C. H. Zhou, H. Wang, Appl. Clay Sci.,
2014, 97−98, 222−234.
[44] L. Le Forestier, F. Muller, F. Villieras, M. Pelletier, Appl. Clay Sci.,
2010, 48, 18−25.
[45] L. Zatta, L. P. Ramos, F. Wypych, Appl. Clay Sci., 2013, 80−81,
236−244.
[46] L. M. Wu, D. S. Tong, L. Z. Zhao, W. H. Yu, C. H. Zhou, H. Wang, Appl.
Clay Sci., 2014, 95, 74−82.
[47] Z. Mojović, P. Banković, A. Milutinović‐Nikolić, B. Nedić, D. Jo‐
vanović, Appl. Clay Sci., 2010, 48, 179−184.
[6] E. Kraleva, R. Palcheva, L. Dimitrov, U. Armbruster, A. Bruckner, A.
Spojakina, J. Mater. Sci., 2011, 46, 7160−7168.
[7] S. H. Chai, B. Yan, L. Z. Tao, Y. Liang, B. Q. Xu, Catal. Today, 2014,
234, 215−222.
[8] C. H. Zhou, H. Zhao, D. S. Tong, L. M. Wu, W. H. Yu, Catal. Rev. Sci.
Eng., 2013, 55, 369−453.
[9] G. Paulo da Silva, M. Mack, J. Contiero, Biotechnol. Adv., 2009, 27,
30−39.
[10] A. Talebian‐Kiakalaieh, N. A. S. Amin, H. Hezaveh, Renew. Sust.
Energy Rev., 2014, 40, 28−59.
[11] L. Q. Shen, H. B. Yin, A. L. Wang, X. F. Lu, C. H. Zhang, F. Chen, Y. T.
Wang, H. J. Chen, J. Ind. Eng. Chem., 2014, 20, 759–766.
[12] L. Q. Shen, H. B. Yin, A. L. Wang, Y. H. Feng, Y. T. Shen, Z. A. Wu, T. S.
Jiang, Chem. Eng. J., 2012, 180, 277–283.
[13] A. Martin, U. Armbruster, H. Atia, Eur. J. Lipid Sci. Technol., 2012,
114, 10−23.
[14] M. Massa, A. Andersson, E. Finocchio, G. Busca, F. Lenrick, L. R.
Wallenberg, J. Catal., 2013, 297, 93−109.
[48] D. S. Tong, X. Xia, X. P. Luo, L. M. Wu, C. X. Lin, W. H. Yu, C. H. Zhou,
Z. K. Zhong, Appl. Clay Sci., 2012, 74, 147−153.
[49] N. Yildiz, Z. Aktas, A. Calimli, Particulate Sci. Technol., 2004, 22,
21−33.
[15] P. Lauriol‐Garbry, G. Postole, S. Loridant, A. Auroux, V.
Belliere‐Baca, P. Rey, J. M. M. Millet, Appl. Catal. B, 2011, 106,
94−102.
[50] Y. H. Zhao, Y. J. Wang, Q. Q. Hao, Z. T. Liu, Z. W. Liu, Fuel Process.
Technol., 2015, 136, 87−95.
[16] B. Katryniok, S. Paul, F. Dumeignil, ACS Catal., 2013, 3,
1819−1834.
[51] D. Olszewska, Appl. Clay Sci., 2011, 53, 353−358.
[52] K. V. Bineesh, M. Kim, G. H. Lee, M. Selvaraj, D. W. Park, Appl. Clay
Sci., 2013, 74, 127−134.
[53] T. T. Li, J. F. Zhang, X. M. Xie, X. M. Yin, X. An, Fuel, 2015, 143,
55−62.
[54] H. Liu, K. Tao, H. B. Yu, C. Zhou, Z. Ma, D. S. Mao, S. H. Zhou,
Comptes Rendus Chimie, 2015, 18, 644−653.
[55] I. M. Szilágyi, J. Madarász, G. Pokol, P. Király, G. Tárkányi, S. Saukko,
J. Mizsei, A. L. Tóth, A. Szabó, K. Varga‐Josepovits, Chem. Mater.,
2008, 20, 4116–4125.
[56] F. Corà, A. Patel, N. M. Harrison, R. Dovesi, C. R. A. Catlow, J. Am.
Chem. Soc., 1996, 118 12174–12182.
[57] M. A. Vicente‐Rodriguez, M. Suarez, M. A. Banares‐Munoz, J. D.
Lopez‐Gonzalez, Spectrochim. Acta Part A, 1996, 52, 1685−1694.
[58] D. S. Tong, Y. M. Zheng W. H. Yu, L. M. Wu, C. H. Chun, Appl. Clay
Sci., 2014, 100, 123−128.
[17] H. P. Decolatti, B. O. Dalla Costa, C. A. Querini, Microporous
Mesoporous Mater., 2015, 204, 180−189.
[18] H. Zhao, C. H. Zhou, L. M. Wu, J. Y. Lou, N. Li, H. M. Yang, D. S. Tong,
W. H. Yu, Appl. Clay Sci., 2013, 74, 154−162.
[19] G. D. Yadav, R. V. Sharma, S. O. Katole, Ind. Eng. Chem. Res., 2013,
52, 10133−10144.
[20] R. Liu, T. F. Wang, Y. Jin, Catal Today, 2014, 233, 127–132.
[21] E. Tsukuda, S. Sato, T. Takahashi, T. Sodesawa, Catal. Commun.,
2007, 8, 1349−1353.
[22] L. Q. Shen, Y. H. Feng, H. B. Yin, A. L. Wang, L. B. Yu, T. S. Jiang, Y. T.
Shen, Z. A. Wu, J. Ind. Eng. Chem., 2011, 17, 484–492.
[23] H. Atia, U. Armbruster, A. Martin, J. Catal., 2008, 258, 71−82.
[24] B. Katryniok, S. Paul, M. Capron, F. Dumeignil, ChemSusChem.,
2009, 2, 719−730.
[25] S. H. Chai, H. P. Wang, Y. Liang, B. Q. Xu, Green Chem., 2008, 10,
1087–1093.
[59] M. K. Yadav, C. D. Chudasama, R. V. Jasra, J. Mol. Catal., 2004, 216,
51–59.
[26] M. Massa, A. Andersson, E. Finocchio, G. Busca, J. Catal., 2013, 307,
170−184.
[60] J. Díaz‐Reyes, R. J. Delgado‐Macuil, V. Dorantes‐García, A.
Pérez‐Benítez, J. A. Balderas‐López, J. A. Ariza‐Ortega, Mater. Sci.
Eng. B, 2010, 174, 182−186.
[61] G. S. Rao, N. P. Rajan, M. H. Sekhar, S. Ammaji, K. V. R. Chary, J. Mol.
Catal. A, 2014, 395, 486−493.
[62] H. M. Liu, P. Yuan, Z. H. Qin, D. Liu, D. Y. Tan, J. X. Zhu, H. P. He, Appl.
Clay Sci., 2013, 80, 398−406.
[63] M. J. G. Fait, H. J. Lunk, M. Feist, M. Schneider, J. N. Dann, T. A. Frisk,
Thermochim. Acta, 2008, 469, 12−22.
[27] K. S. Song, H. B. Zhang, Y. H. Zhang, Y. Tang, K. J. Tang, J. Catal.,
2013, 299, 119–128.
[28] L. M. Wu, D. S. Tong, C. S. Li, S. F. Ji, C. X. Lin, H. M. Yang, Z. K. Zhong,
C. Y. Xu, W. H. Yu, C. H. Zhou, Appl. Clay Sci., 2016, 119, 116−125.
[29] C. H. Zhou, J. Keeling, Appl. Clay Sci., 2013, 74, 3−9.
[30] B. Thomas, V. G. Ramu, S. Gopinath, J. George, M. Kurian, G. Lau‐
rent, G. L. Drisko, S. Sugunan, Appl. Clay Sci., 2011, 53, 227–235.
[31] C. H. Zhou, X. Xia, C. X. Lin, D. S. Tong, J. Beltramini, Chem. Soc. Rev.,
2011, 40, 5588−5617.
[64] A. M. Baghdasaryan, O. M. Niazyan, H. L. Khachatryan, S. L.
[32] T. Santhi, S. Manonmani, T. Smitha, J. Hazard. Mater., 2010, 179,

Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
1099
Graphical Abstract
Chin. J. Catal., 2017, 38: 1087–1100 doi: 10.1016/S1872‐2067(17)62813‐4
Acid‐activated and WOx‐loaded montmorillonite catalysts and
their catalytic behaviors in glycerol dehydration
Weihua Yu, Pengpeng Wang, Chunhui Zhou *, Hanbin Zhao,
Dongshen Tong, Hao Zhang, Huimin Yang, Shengfu Ji, Hao Wang
Zhejiang University of Technology, China; University of Southern
Queensland, Australia; Zhejiang Institute of Geology and Mineral
Resource, China; China National Bamboo Research Center, China;
Beijing University of Chemical Technology, China
Acrolein
Glycerol
Proton
Si
O
Polytungstate [WO6]
Exposed Al3+
Al
+
H (H2O)
Monotungstate [WO4]
Catalysts consisting of WOx supported on H3PO4‐activated mont‐
morillonite showed high activities in glycerol conversion to acro‐
lein. The WOx state was related to the acid type, strength, and
amount on the support surface.
OH2+
Crystalline WO3
WOx / H3PO4-activated montmorillonite catalyst
Kharatyan, Int. J. Refract. Metal Hard Mater., 2014, 43, 216−221.
[65] J. Madarász, I. M. Szilágyi, F. Hange, G. Pokol, J. Anal. Appl. Pyrolysis,
2004, 72, 197−201.
Appl. Clay Sci., 2013, 80, 407−412.
[73] G. B. B. Varadwaj, S. Rana, K. Parida, Chem. Eng. J., 2013, 215–216,
849–858.
[66] M. Ardestani, H. Arabi, H. R. Rezaie, H. Razavizadeh, Int. J. Refract.
Metal Hard Mater., 2009, 27, 796−800.
[74] S. H. Zhu, X. Q. Gao, Y. L. Zhu, Y. W. Li, J. Mol. Catal. A, 2015, 398,
391–398.
[67] D. G. Barton, M. Shtein, R. D. Wilson, S. L. Soled, E. Iglesia, J. Phys.
Chem. B, 1999, 103, 630−640.
[75] C. J. Yue, M. M. Gan, L. P. Gu, Y. F. Zhuang, J. Taiwan Inst. Chem. Eng.,
2014, 45, 1443–1448.
[68] W. H. Yu, N. Li, D. S. Tong, C. H. Zhou, C. X. Lin, C. Y. Xu, Appl. Clay
Sci., 2013, 80–81, 443–452.
[69] D. G. Barton, S. L. Soled, G. D. Meitzner, G. A. Fuentes, E. Iglesia, J.
Catal., 1999, 181, 57–72.
[70] C. D. Baertsch, S. L. Soled, E. Iglesia, J. Phys. Chem. B, 2001, 105,
1320–1330.
[71] S. García‐Fernández, I. Gandarias, J. Requies, M. B. Güemez, S.
Bennici, A. Auroux, P. L. Arias, J. Catal., 2015, 323, 65−75.
[72] D. Liu, P. Yuan, H. M. Liu, D. Y. Tan, H. P. He, J. X. Zhu, T. H. Chen,
[76] C. S. Carriço, F. T. Cruz, M. B. Santos, H. O. Pastore, H. M. C.
Andrade, A. J. S. Mascarenhas, Microporous Mesoporous Mater.,
2013, 181, 74−82.
[77] M. Dalil, D. Carnevali, J. L. Dubois, G. S. Patience, Chem. Eng. J.,
2015, 270, 557–563.
[78] J. Deleplanque, J. L. Dubois, J. F. Devaux, W. Ueda, Catal. Today,
2010, 157, 351−358.
[79] S. H. Chai, H. P. Wang, Y. Liang, B. Q. Xu, Green Chem., 2007, 9,
1130−1136.
催化甘油脱水反应的酸活化蒙脱石负载WOx催化剂的研究
俞卫华a,b, 王朋朋b, 周春晖b,c,d,*, 赵汉彬b, 童东绅b, 张 浩b, 杨慧敏e, 季生福f, 王 浩c
a浙江工业大学之江学院, 浙江杭州310024, 中国
b浙江工业大学化学工程学院, 绿色化学合成技术国家重点实验室培育基地, 浙江杭州310032，中国
c南昆士兰大学未来材料研究所, 图文巴4350, 澳大利亚
d浙江省地质矿产研究所, 浙江省非金属矿物工程研究中心, 浙江杭州310007，中国
e国家林业局竹子研究开发中心, 浙江省竹子高效加工重点实验室, 浙江杭州310012，中国
f北京化工大学, 化工资源有效利用国家重点实验室, 北京100029，中国
摘要: 甘油是一种可由生物资源生产、可持续的、可降解的平台化学品, 是生物柴油、肥皂化工等工业生产过程中的主要
副产物. 催化甘油脱水反应生产丙烯醛, 有望能替代丙烯等石油裂解产物合成丙烯醛的传统工业路线. 丙烯醛是一种重要
的化工中间体, 被用于合成蛋氨酸、丙烯酸、3-甲基吡啶和1,3-丙二醇, 并被广泛地应用于农药、医药、高分子材料等领域.
随着全球可持续能源发展, 生物柴油生产迅速发展, 将产生大量的副产物甘油. 利用甘油为原料, 通过合适的催化剂的催
化脱水反应生成丙烯醛, 是近十多年来国内外工业催化的研究热点之一.
用于催化甘油脱水合成丙烯醛的酸催化剂有杂多酸、金属氧化物、沸石与酸性粘土矿物等. 钨磷杂多酸(H3PW12O40)
负载的催化剂虽然具有较强的酸性, 有利于催化甘油脱水, 但容易导致结焦, 而且热稳定差, 容易失活. 钨磷杂多酸负载于
SiO2, TiO2, Al2O3, SiO2-Al2O3, K-10蒙脱石上表现出不同的催化活性, 表明催化剂和载体的表面酸性和孔结构影响催化性
能. 近来研究发现, 负载于ZrO2, Al2O3的钨氧化物(WOx)催化剂热稳定性好、酸性高, 在甘油脱水反应生成丙烯醛中表现出
良好的催化性能. 但有关钨氧化物(WOx)结构、催化活性受载体组成、酸性影响的本质和规律一直不清楚. 本文采用20 wt%
的硫酸、盐酸、磷酸和乙酸对蒙脱石进行酸改性, 并在磷酸改性的蒙脱石上负载W含量为4–16 wt%的WOx作为催化剂, 用

1100
Weihua Yu et al. / Chinese Journal of Catalysis 38 (2017) 1087–1100
于甘油气相脱水反应. X-射线衍射(XRD)、热重-差热法(TG-DTG)、氨程序升温脱附(NH3-TPD)、红外光谱(FT-IR)和紫外
漫反射可见光谱(DR UV-vis)等表征, 探讨了酸改性和负载WOx的蒙脱石对催化剂催化性能的影响.
蒙脱石经过20wt%的硫酸、盐酸、磷酸和乙酸的活化, 酸性增加. 四种酸改性的蒙脱石对甘油气相脱水反应均有催化
活性, 这是因为在蒙脱石酸活化过程中, H+经过阳离子交换反应进入蒙脱石层间, 同时蒙脱石八面体中的部分Al3+被浸出,
使层板上出现不饱和Al3+, 为催化剂提供了L酸位, 蒙脱石硅氧四面体上的Si−OH以及[AlO4]上吸附的H3O+提供了B酸位.
XRD分析表明, 负载WOx的蒙脱石表面存在WO2.72, WO2.9和WO3三种不同类型的WOx, 当钨负载量从12 wt%增至16
wt%, 孤立的单斜晶系WO3晶粒增多. NH3-TPD和DR UV-vis结果表明, WOx负载在蒙脱石表面以[WO5/WO6](B酸位)、
[WO4]和单斜晶系WO3相(L酸位)形式存在. 蒙脱石上负载WOx能够调节催化剂的酸强度、酸量和酸位. 随着钨负载量从4
wt%增至12 wt%, 丙烯醛收率从40.9%增加到67.3%; 进一步增加钨负载量到16 wt%, 丙烯醛收率降为50.7%. 结果发现, 随
着钨负载量的增加, 催化活性组分含量增加, [WO5/WO6](B酸位)增加, 使催化活性增加; 当W负载量达到16 wt%时, WOx分
散性降低, 且在催化剂表面形成孤立的单斜晶系WO3相(L酸位), 不利于提高丙烯醛选择性. 当反应温度为320 oC, 甘油水
溶液浓度为15 wt%时, 磷酸活化蒙脱石负载12 wt%W的催化剂上甘油转化率为89.6%, 丙烯醛收率达到73.3%.
关键词: 甘油; 丙烯醛; 脱水; WOx; 酸活化粘土; 催化剂
收稿日期: 2017-01-13. 接受日期: 2017-03-10. 出版日期: 2017-06-05.
*通讯联系人. 电话/传真: (0571)88320062; 电子信箱: clay@zjut.edu.cn
基金来源: 国家自然科学基金(21373185, 41672033, 21506188, 21404090); 浙江省非金属矿工程研究中心开放基金项目
(ZD2015K07); 绿色化学合成技术国家重点实验室培育基地开放基金项目(GCTKF2014006); 浙江省竹子高效加工重点实验室开
放基金项目(2016); 北京化工大学化工资源有效利用国家重点实验室开放基金项目(CRE-2016-C-303).
本文的英文电子版由Elsevier出版社在ScienceDirect上出版(http://www.sciencedirect.com/science/journal/18722067).