Catalytic dehydration of glycerol over vanadium phosphate oxides in the presence of molecular oxygen

Article abstract of DOI:10.1016/j.jcat.2009.09.024

We report the dehydration of glycerol over vanadium phosphate oxide (VPO) catalysts. Catalytic reactions were conducted in a gas-phase fixed-bed reactor at temperatures from 250 to 350 °C with O₂/glycerol ratios of 0-13.6. Hemihydrate VOHPO

Full text of DOI:10.1016/j.jcat.2009.09.024

Journal of Catalysis 268 (2009) 260–267
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Catalytic dehydration of glycerol over vanadium phosphate oxides in the presence
of molecular oxygen
b
a,
Feng Wang a, , Jean-Luc Dubois , Wataru Ueda
*
*
a Catalysis Research Center, Hokkaido University, North 21, West 10, Sapporo 001-0021, Japan
b Arkema France, 420 Rue d’Estienne d’Orves, 92705 Colombes, France
a r t i c l e i n f o
a b s t r a c t
Article history:
We report the dehydration of glycerol over vanadium phosphate oxide (VPO) catalysts. Catalytic reac-
tions were conducted in a gas-phase fixed-bed reactor at temperatures from 250 to 350 °C with O2/glyc-
erol ratios of 0–13.6. Hemihydrate VOHPO4ꢀ0.5H2O oxide emerged as the best catalyst. At 300 °C, glycerol
conversion was 100% with 66% selectivity to acrolein and 14% selectivity to acetaldehyde. The carbon bal-
ance was 93%. The advantage of adding oxygen is that the catalyst can be maintained in an oxidized state.
It was shown that carbon deposition was inhibited and that side products were greatly reduced. More-
over, the structure of the catalyst did not change during the reaction as confirmed by X-ray diffraction
(XRD), infrared (IR), and thermogravimetric–differential thermal analysis (TG–DTA). We also investigated
the reaction network and proposed a possible reaction pathway.
Received 7 August 2009
Revised 25 September 2009
Accepted 25 September 2009
Available online 24 October 2009
Keywords:
Dehydration
Glycerol
Acrolein
Ó 2009 Elsevier Inc. All rights reserved.
Acetaldehyde
Hemihydrate vanadium phosphate oxide
Molecular oxygen
1. Introduction
ported to produce 1,2-propanediol, ethylene glycol, and propane
over Pt, Pd, Ru, Ni, and Co catalysts [7–10]. There have been several
Glycerol is the main by-product of transesterification of vegeta-
ble oils (triglycerides) and methanol for producing biodiesel. In this
process, 1000 kg of biodiesel usually generates about 100 kg of
glycerol. Increase in biodiesel production therefore results in the
accumulation of glycerol. Currently, about 350,000 tons of glycerol
is produced per annum in the USA and about 600,000 tons is pro-
duced per annum in Europe. The levels will increase because of the
implementation of EU directive 2003/30/EC, which requires
replacement of 5.75% of petroleum fuels with biofuel across all
member states by 2010. Therefore, it is imperative to find value-
added ways of utilizing glycerol that will be beneficial for both
the environment and biodiesel manufacturing. In recent years,
the use of glycerol has become an important research topic in
catalysis. Several reviews have recently been published giving a
comprehensive overview of the topic [1–3].
studies on the decomposition of glycerol to syngas by pyrolysis
reaction or to hydrogen by reforming reaction [11–13]. In some
cases, glycerol is considered as ‘‘organic water” and employed as
a solvent [14,15]. Selective oxidation of glycerol to acids has been
conducted over precious metal catalysts in liquid-phase with oxy-
gen or hydrogen peroxide as an oxidant to produce highly func-
tionalized carboxylic acids, such as tartronic acid, glyceric acid,
hydroxypyruvic acid, and glyoxylic acid [16–18].
Acrolein is an important product from glycerol dehydration. In
1933, Scheering–Kahlbaum reported glycerol dehydration over
pumice-supporting lithium phosphate or copper phosphate cata-
lysts at temperature from 300 to 600 °C [19]. Several groups, such
as Corma et al. [20], Xu et al. [21–23], and Tsukuda et al. [24] have
recently studied glycerol dehydration over acid catalysts, such as
sulfates, phosphates, zeolites, and metal oxides, under conditions
of gas-phase, liquid-phase, supercritical water, or hot-compressed
water conditions [25–29]. These studies had various problems
such as harsh reaction conditions, severe coke formation, and large
amounts of by-products. These problems must be solved in order
to achieve continuous and economical industrial production. Ark-
ema Company described a way in patents to inhibit catalyst deac-
tivation and reduce side products through addition of oxygen
during the dehydration reaction [30,31]. Preferred catalysts de-
scribed in these patents are acidic materials, such as WO3, ZrO2,
ZrO2/SO4, and Al2O3.
Glycerol has been converted to mono, di, and triacetylesters by
acetylation reaction or to glycerol carbonate by carboxylation reac-
tion with CO2 or urea [4,5]. There are some reports on the synthesis
of 1,3-propanediol either by direct hydrogenation or by processes
of fermentation and on the synthesis of epichlorohydrine and
polyethers [6]. Catalytic hydrogenation of glycerol has been re-
* Corresponding authors. Fax: +81 11 706 9164 (F. Wang).
E-mail addresses: wangfeng@cat.hokudai.ac.jp (F. Wang), ueda@cat.hokudai.
ac.jp (W. Ueda).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.09.024

F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
261
The objective of this study was to develop a process with high
2.2. Characterization of catalysts
selectivity for acrolein and without severe coke formation. We fo-
cused on glycerol dehydration in the presence of oxygen over VPO
catalysts. The advantages of adding oxygen in the dehydration
reaction are as follows. First, an oxidative atmosphere for dehydra-
tion reaction is effective for inhibiting or reducing carbon deposi-
tion and thus for increasing longevity of the catalyst. Strongly
adsorbed hydrocarbon species on the catalyst surface can be
quickly removed by oxidation before the formation of intensive
carbon coke, and catalytic active sites are recovered. Second, glyc-
erol is a strong reductant at a high reaction temperature. The pres-
ence of oxygen enables maintenance of the catalytic redox cycle of
metal oxides. For example, the reduction of metal centers with
higher valences, such as V5+ to V4+ or V3+, usually leads to catalyst
deactivation. Third, it would be interesting if conversion of glycerol
to acrylic acid is realized in one-step. Tandem-type reactors, one
reactor being loaded with a dehydration catalyst to produce acro-
lein and the other reactor being loaded with an oxidation catalyst
to produce acrylic acid, are applicable in industrial process. How-
ever, a one-step process would greatly reduce the investment cost
and simplify the engineering process. Furthermore, a better heat
balance over the catalyst bed could be achieved by a one-step pro-
cess because the dehydration reaction is endothermic and the oxi-
dation is exothermic.
VPO oxides, such as pyrophosphate (VO)2P2O7, have been stud-
ied in the selective oxidation of butane, isobutene, propane, or eth-
ane to the corresponding acids [32–37]. As far as we know, VPO
oxides have not been studied for the utilization of glycerol. We re-
port here glycerol dehydration over VPO oxides in the presence of
molecular oxygen. We found that vanadyl hydrogen phosphate
hemihydrate VOHPO4ꢀ0.5H2O exhibited the best performance with
66% selectivity for acrolein at 100% conversion of glycerol and
without coke deposition. Its catalytic performance is comparable
with that in the literature works. The carbon balance was 93%,
which was the highest among the three catalysts and was even
better than the previously reported results. Furthermore it is the
first time that VOHPO4ꢀ0.5H2O is reported as an active and selec-
tive catalyst. The stability of the catalyst was examined and the
possible reaction pathways were investigated.
XRD measurements were performed with a Rigaku RINT Ultima+
diffractometer with Cu K radiation (K 1.54056 Å) and X-ray
a
a
power of 40 kV/20 mA. Field emission scanning electron micros-
copy (FE-SEM) was performed on a JSM-7400F (JEOL). The samples
for SEM were dusted on an adhesive conductive carbon paper at-
tached to a brass mount. Specific surface areas were measured by
nitrogen adsorption at 77 K using the Brunauer–Emmett–Teller
(BET) method over Autosorb 6AG (Quantachrome Instruments). IR
spectra were recorded on a Perkin–Elmer FT-IR spectrometer by
accumulating 32 scans at a spectrum resolution of 2 cmꢁ1. The sam-
ple pellet of 1.5 mm in diameter was prepared by pressing a mix-
ture of the sample and KBr. TG–DTA measurements were
performed in a TG-8120 (Rigaku) thermogravimetric analyzer.
Dry air provided by
a pressured tank with a flow rate of
30 mL minꢁ1 was used as the carrier. The baseline was subtracted
from a blank run. Catalyst and reference alumina samples were
loaded into twin holders. The experiments were carried out from
room temperature to 700 °C with a heating rate of 10 °C minꢁ1
.
2.3. Catalytic tests
Glycerol dehydration tests were conducted under normal pres-
sure in a vertical fix-bed homemade Pyrex reactor with internal
diameter of 5 mm. The catalyst was mixed with quartz (50–70
mesh) and loaded in the middle of the reactor with quartz wool
packed in both ends. The catalyst was pretreated at 300 °C for
1 h in a flow of nitrogen (18 mL minꢁ1). The reaction temperature
was monitored by a thermocouple inserted into the middle of the
catalyst. Typically, 0.2 g of catalyst was used and diluted with 3.0 g
of quartz. Aqueous glycerol (20% wt/wt) was fed at the speed of
0.50 g hꢁ1 by a syringe pump. The composition of fed gas was
N2:O2:H2O:glycerol, equal to 57.2:12.7:28.7:1.4 in molar ratio.
The sampling interval was 10 h. The product was collected in a cold
trap (ꢁ78 °C, mixture of acetone and dry ice). The collected product
was dissolved in 5 mL of aqueous solution of internal standard 2-
butanol and was analyzed by GC–MS (Shimadzu 15A) with a cap-
illary column (GL Sciences, TC-FFAP 60 m ꢂ 0.25 mm ꢂ 0.5
lm)
and a flame ionization detector (FID). The chromatograph column
was run at 110–250 °C with a ramping rate of 5 °C minꢁ1 and at
250 °C for 10 min. Volatile compounds running out of the cold trap
were analyzed by three online GCs with two thermal conductivity
detectors (TCD) and one FID detector. The GC analyses allowed
quantifications of CO, CO2, acetaldehyde, acrolein, and acetic acid.
All the results of GC analyses for liquid-phase and gas-phase
were combined to calculate the conversion, product selectivity,
and mass balance. Calculation of glycerol conversion (Xgly, mol%)
was based on the concentrations of glycerol at the inlet and the
outlet of the reactor:
2. Experimental
2.1. Preparation of catalysts
All reagents were of analytical-reagent grade purchased from
Wako Pure Chemical Company and were used without further
purification. Distilled water prepared by using a Yamato Autostill
WG25 (Tokyo, Japan) was utilized throughout this work.
The preparation of dihydrate VOPO4ꢀ2H2O has been reported
elsewhere [38]. In brief, a mixture of V2O5 (24 g), H3PO4 (85%,
133 mL), and H2O (577 mL) was refluxed at 115 °C for 16 h. The
reaction produced a bright-yellow solid, which was collected by fil-
tration, washed with 100 mL acetone, and dried for 10 h under
ambient conditions. Both XRD and IR characterizations confirmed
the solid to be VOPO4ꢀ2H2O. VOHPO4ꢀ0.5H2O was prepared by
reduction of VOPO4ꢀ2H2O in organic alcohol. A suspension of VO-
PO4ꢀ2H2O (5 g) powder in 2-butanol (50 mL) was stirred under re-
flux for 23 h (oil bath temperature: 101 °C). The resulting light-
blue solid was collected by filtration, washed with 100 mL of ace-
tone, and dried for 10 h under ambient conditions. The solid was
characterized by XRD and IR and its phase was confirmed to be
hemihydrate VOHPO4ꢀ0.5H2O. The precursor of VOHPO4ꢀ0.5H2O
Mgly;in ꢁ Mgly;out
Xgly
¼
ꢂ 100;
ð1Þ
Mgly;in
where Mgly,in and Mgly,out are the quantities of fed and unreacted
glycerol, respectively. The conversion of oxygen (Xoxy, mol%) was
determined using online GC by measuring the integration area ra-
tios of oxygen to nitrogen for the inlet and outlet gases, assuming
that the amount of nitrogen was constant. Product selectivity (Si,
mol%) was calculated from the molar concentrations of product
and glycerol according to Eq. (2):
Mi
zi
3
Si ¼
ꢂ
ꢂ 100;
ð2Þ
Mgly;in ꢁ Mgly;out
(1.0 g) was treated at 550 °C for 4 h under nitrogen (50 mL minꢁ1
)
at the heating rate of 1 °C minꢁ1. The resulting gray solid was con-
where z represents the number of carbon atoms, for example, z = 3
for glycerol and Mi represents the detected molar quantity of
firmed to be pyrophosphate (VO)2P2O7 by XRD and IR.

262
F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
1800
001
1600
200
P O
(VO)2
2
7
1400
001
024
032
1200
1000
800
101
VOHPO.0.5H2O
4
121
021
220
102
031
201
600
200
400
101
002
201
200
0
102
VOPO.2H2O
4
10
20
30
40
50
60
2-Theta/theta
Fig. 2. XRD patterns of VOPO4ꢀ2H2O, VOHPO4ꢀ0.5H2O, and (VO)2P2O7.
VOPO .2H2O
4
Fig. 1. The equations and products for calculating carbon balance.
VOHPO .0.5H2O
4
product i. Carbon balance (mol%) was calculated by summing up the
unreacted glycerol and the total quantities of detected and cali-
brated products, as shown in Fig. 1.
(VO)2P2O7
3. Results and discussion
3.1. Characterization of the catalysts
2000
1800
1600
1400
1200
1000
800
600
400
Three VPO materials, dihydrate (VOPO4ꢀ2H2O), hemihydrate
(VOHPO4ꢀ0.5H2O), and pyrophosphate ((VO)2P2O7), were prepared.
Their XRD patterns are shown in Fig. 2. The reflection peaks at
2h = 12.1°, 18.7°, 24.1°, and 29.0° are indexed to be (0 0 1),
(1 0 1), (0 0 2), and (2 0 0) plane of lamellar VOPO4ꢀ2H2O phase,
respectively. After reduction in 2-butanol, a strong reflection peak
appears at 2h = 15.8°, which corresponds to the (0 0 1) plane of
VOHPO4ꢀ0.5H2O phase. During the reduction process, 2-butanol is
oxidized and VOPO4ꢀ2H2O phase is reduced, resulting in the inter-
calation of hydrogen into crystalline layers. Under thermal treat-
ment in nitrogen atmosphere, the VOHPO4ꢀ0.5H2O phase
undergoes a topotactic dehydration to (VO)2P2O7 phase, as con-
firmed by the characteristic peaks at 2h = 23.3°, 28.7°, 30.3°, and
43.6°, which are in good agreement with those in the literature
[32,39].
Wavenumber [cm-1]
Fig. 3. FTꢁIR spectra of VOPO4ꢀ2H2O, VOHPO4ꢀ0.5H2O, and (VO)2P2O7.
(d (PO3)), and 456 cmꢁ1 (d (OPO)) belong to the phase of (VO)2P2O7
[42].
Representative SEM images are shown in Fig. 4. The three sam-
ples have platelet morphologies with uniform particle size distri-
bution. VOPO4ꢀ2H2O mainly consists of crystals with sizes of 2–
4
lm and thicknesses of 0.1–0.2
lm. The reduction of VOPO4ꢀ2H2O
produces VOHPO4ꢀ0.5H2O orienting along the (0 0 1) plane, which
has larger platelets with sizes of 10–20 m and thicknesses of
0.1–0.3 m. The surface of the (VO)2P2O7 platelet is covered with
l
l
pores owing to the dehydration reaction. The specific surface areas
of VOPO4ꢀ2H2O, VOHPO4ꢀ0.5H2O, and (VO)2P2O7 are 7 m2 gꢁ1
,
In FT-IR spectra (Fig. 3), intensive changes are observed in the
range of 400–2000 cmꢁ1. Bending modes of water molecules can
be seen around 1630 cmꢁ1 in three samples. In general, bands be-
tween 600 and 1300 cmꢁ1 correspond to the stretching of PꢁO and
V–O, while bands below 600 cmꢁ1 are assigned to the bending of
12 m2 gꢁ1, and 8 m2 gꢁ1, respectively, which agree well with the
reported results [39,43].
3.2. Screening catalysts
OꢁVꢁO and OꢁPꢁO [40]. The IR spectrum shows 1084 cmꢁ1
(mas
(PꢁO)), 994 cmꢁ1
(m
(V@O)), 944 cmꢁ1
(m
(VꢁOH)), 672 cmꢁ1 (d
Catalytic results over the three VPO catalysts in the presence of
molecular oxygen are shown in Table 1. The major products of the
glycerol dehydration were acrolein, acetaldehyde, acetic acid, ac-
rylic acid, propionic acid, and esters. Products grouped under the
label of ‘‘other CHO” include methyl vinyl ketone, butanone, cyclo-
pentanone, and methacrolein. We did not observe the formation of
CO, probably because of the oxygen-rich reaction condition. In
most cases, carbon balance was less than 100%. Several factors
may contribute to the difference, such as some unknown minor
peaks in chromatograms, some decomposed products during anal-
ysis in the GC injector, and a small amount of carbon deposits. The
(V–OH) or d (P–OH)), and 576 cmꢁ1 (das P–O–P) bands, which are
characteristics of VOPO4ꢀ2H2O [41]. IR bands of 1200 cmꢁ1
(mas
(PO3)), 1132 cmꢁ1 (dip (P–OH)), 1104 cmꢁ1 mas (PO3)), 1054 cmꢁ1
(
(
mas (PO3)), 980 cmꢁ1 (V@O)), 931 cmꢁ1 (P–OH)), 684 cmꢁ1
(m (m
(
x
(coordinated H2O)), 642 cmꢁ1 (doop (P–OH)), 548 cmꢁ1 (d
(OPO)), 534 cmꢁ1 (d (OPO)), 480 cmꢁ1 (d (OPO)), and 417 cmꢁ1 (d
(OPO)) are characteristics of VOHPO4ꢀ0.5H2O [40]. IR bands occur-
ring at 1244 cmꢁ1 mas (PO3)), 1220 cmꢁ1 (dip (P–OH)), 1142 cmꢁ1
(
(
(
mas (PO3)), 1084 cmꢁ1
ms (P–O–P)), 634 cmꢁ1
(
mas (PO3)), 970 cmꢁ1
(m
(V@O)), 744 cmꢁ1
(
mas (PO3)), 576 cmꢁ1 (d (OPO)), 514 cmꢁ1

F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
263
Fig. 4. SEM images of VOPO4ꢀ2H2O (upper left), VOHPO4ꢀ0.5H2O (upper right), and (VO)2P2O7.
The reaction produced substantial amounts of CO2, CHO, and
unknown products (total selectivity of ca. 25%) over the VO-
PO4ꢀ2H2O catalyst. Pyrophosphate (VO)2P2O7 catalyst, which is
well known in butane oxidation, realized 100% conversion of glyc-
erol but produced 41% selectivity to acrolein with more than 25%
selectivity to by-products. The carbon balance of 72% was much
smaller than that of the reaction over the VOHPO4ꢀ0.5H2O. We
speculate that (VO)2P2O7 catalyst strongly adsorbs glycerol, acro-
lein, and other reaction intermediates. The strong adsorption sites,
usually acid sites, are generated by thermal treatment of VOH-
PO4ꢀ0.5H2O at a high temperature. Moreover, the 12% selectivity
to hydroxyacetone over the (VO)2P2O7 catalyst is more than those
over the other two catalysts, suggesting that the catalyst is active
for dehydration reaction at a terminal hydroxyl group of glycerol.
Although the dehydration of glycerol was conducted in oxygen
atmosphere, we did not observe high selectivity to acrylic acid.
There have been some reports of oxidation of acrolein over Mo,
V, or W oxides [44,45]. Arkema described in a patent that a small
amount of acrylic acid was formed in a single step from glycerol
over MoVWO-type catalysts [46]. In the present study, however,
although excess oxygen to glycerol (oxygen:glycerol = 9.1:1) was
added, acrolein was not oxidized to acrylic acid. The selectivities
for acrylic acid were less than 5% over the three catalysts. This re-
sult indicates that the VPO catalysts are capable of catalyzing the
dehydration of glycerol to acrolein but are unable to catalyze the
oxidation of acrolein to acrylic acid. The presence of oxygen inhib-
ited the formation of hydrogenated products such as propane, bu-
tane, and diols reported in the dehydration of glycerol without
adding oxygen [20]. The effect of oxygen concentration will be dis-
cussed in the following section.
Table 1
Catalytic results over the three VPO catalysts.
Catalyst
VOPO4ꢀ2H2O
81.0
14
VOHPO4ꢀ0.5H2O
100
14
(VO)2P2O7
Glycerol conversion (mol%)
Oxygen conversion (mol%)
Carbon balance (mol%)
100
7
72
85
93
Products distribution (mol%)
Acrolein
Acetaldehyde
Hydroxyacetone
Acetone
55
12
0
0
0
0
1
0
6
66
14
4
0.2
0.4
5
3
2
0.5
4
41
4
12
0
0.1
1
1
0
1
14
Propanal
Acetic acid
Acrylic acid
Esters
CO2
Other CHO
16
Reaction conditions: 0.2 g of catalyst was used and diluted with 3.0 g of quartz. The
aqueous glycerol (20% wt/wt) was fed at the speed of 0.5 g hꢁ1 by a syringe pump.
The composition of fed gas was N2:O2:H2O:glycerol, equal to 57.2:12.7:28.7:1.4 in
mole ratio. The sampling interval was 10 h.
conversion of oxygen (Coxy) is 6–14%. This value is larger than the
calculated value by summing up the total quantity of oxygen
sources from products, such as acetic acid, acrylic acid, CO, and
CO2. There are two possible reasons for the missing oxygen: (1)
water as a co-feeding substance was not included in the calculation
and (2) some unknown by-products probably contributing to oxy-
gen consumption were not determined.
VOHPO4ꢀ0.5H2O is often used as a precursor of (VO)2P2O7 cata-
lyst. However, catalysis by VOHPO4ꢀ0.5H2O has been overlooked. In
this study, we found that VOHPO4ꢀ0.5H2O is the best catalyst, offer-
ing 100% conversion of glycerol, 66% selectivity to acrolein, 14%
selectivity to acetaldehyde, and 5% selectivity to acetic acid. Its cat-
alytic performance is comparable with that in the literature works.
The carbon balance was 93%, which is the highest among the three
catalysts and is even better than the literature results. We em-
ployed the VOHPO4ꢀ0.5H2O catalyst in detailed studies.
3.3. Effect of oxygen concentration
The concentration of oxygen in the feeding stream had a
remarkable effect on the catalytic results of glycerol dehydration
as shown in Table 2. The VOHPO4ꢀ0.5H2O catalyst was inactive in
the absence of oxygen with 43% conversion of glycerol and carbon
balance of 68%. When oxygen was co-fed with glycerol in a ratio of

264
F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
Table 2
did not appear on the used catalyst of the reaction with oxygen.
Thus, the presence of oxygen can greatly decrease the adsorbed
carbon species. Such species are converted into heavy carbon spe-
cies, such as carbon coke, in a long-run reaction. Carbon coke is rel-
atively stable and requires a higher temperature to be removed
than in situ oxidation removal as soon as it is formed. Moreover,
oxygen can maintain the redox cycle of metal sites. During the
reaction, the catalyst surface contains many species with C–O,
C@O, C–C, C@C, or C–H bonds. Conversion of these species usually
needs oxygen species, and thus the catalyst surface easily gets
starved for oxygen. Once oxygen is removed from the catalyst sur-
face, an acid site will be created, which in turn serves as an active
site for removal of next oxygen. Therefore, timely supply of oxygen
is necessary to keep the catalyst active. In this study, we added ex-
cess oxygen to avoid reduction of the catalyst, a method similar to
butane oxidation, where excess oxygen is employed to avoid over-
reduction of the catalyst [36].
Effect of oxygen concentration on catalytic results over the VOHPO4ꢀ0.5H2O catalyst.
O2/N2 (mL/mL)
0/18
0.5/18
2/18
4/18
6/18
Ratio of O2/glycerol (mol/mol)
Glycerol conversion (mol%)
Oxygen conversion (mol%)
Carbon balance (mol%)
0
43
0
1.1
90
79
4.5
95
40
9.1
100
14
13.6
100
15
68
66
77
91
81
Products distribution (mol%)
Acrolein
Acetaldehyde
Hydroxyacetone
Acetone
35
6
33
0
0
0
0
0
1
24
4
15
0
0
4
2
2
15
51
15
3
0
0
4
2
5
5
66
15
4
0
0
5
3
1
4
43
15
0
0
0
25
8
7
Propanal
Acetic acid
Acrylic acid
CO2
Other CHO
1
Reaction conditions: see note in Table 1.
The selectivity to hydroxyacetone decreased with increasing
oxygen concentration. The selectivity to hydroxyacetone was 33%
without addition of oxygen, rapidly decreased to 15% with addition
of oxygen (oxygen/glycerol = 1.1), and became 0% with further
addition of oxygen (oxygen/glycerol = 13.6). Hydroxyacetone is
the product of 1,2-dehydration with loss of the hydroxyl group
from the terminal carbon [48]. Calculations show that 1,2-dehy-
dration resulting in the loss of the central or terminal hydroxyl
1.1 (oxygen:glycerol), the conversion of glycerol significantly in-
creased to 90% with better carbon balance. Further increase in oxy-
gen concentration increased the conversion of glycerol, decreased
the unknown products, and improved the carbon balance. One
hundred percent conversion of glycerol was achieved at the oxy-
gen/glycerol ratio of 9.1. Increase in oxygen/glycerol ratio up to
13.6 did not change the conversion of glycerol. A blank test (with-
out a catalyst) showed that there was no autocatalysis or catalysis
initiated by reactor walls. Therefore, the observed conversion of
glycerol truly took place on the catalyst.
Catalytic active sites easily get blocked and poisoned by
strongly adsorbed substances, such as heavy carbon species, finally
deactivating the catalyst. We washed a used catalyst from a reac-
tion without adding oxygen with dimethyl sulfoxide solvent. The
catalyst was filtered out and the filtrate was analyzed by GC–MS.
We found a large amount of mass fragments with m/z ratios in
the range of 200–500. These fragments indicate that heavy carbon
species had formed on the catalyst surface [47]. On the other hand,
we observed only a small amount of those fragments when treat-
ing the catalyst from a reaction conducted with oxygen/glycerol ra-
tio of 9.1 in a similar way. TG–DTA results of used catalysts are
shown in Fig. 5. The catalyst of the reaction without adding oxygen
has a weight loss at 600 °C in the TG curve and an exothermal peak
in the DTA curve correspondingly, ascribed to the combustion of
carbon species. However, similar weight loss and exothermal peak
group has nearly equal energy barriers (terminal: 73.2 kcal molꢁ1
;
central: 70.9 kcal molꢁ1) [48]. In this study, the change in hydrox-
yacetone selectivity may have depended on the nature of the active
sites, which can be modified in situ by molecular oxygen. It is also
possible that hydroxyacetone is oxidized to other products with
increasing oxygen concentration, such as methylglyoxal, which is
difficult to be determined in the present analysis.
The selectivities to acrolein and acetaldehyde increased with
the increase in oxygen concentration until the oxygen/glycerol ra-
tio reached 9.1, suggesting that oxygen promotes 1,2-dehydration
with the loss of the hydroxyl group from the central carbon, and
1,3-dehydration route, through which acetaldehyde is formed
[48]. Further increase in the oxygen/glycerol ratio from 9.1 to
13.6 decreased the selectivity of acrolein from 66% to 43%, slightly
affected the selectivity of acetaldehyde, and enhanced the selectiv-
ity of acetic acid from 5% to 25% and that of CO2 from 0.5% to 7%.
Acetic acid and CO2 are the products of acrolein oxidation since
only the selectivity to acrolein sharply decreased. In the presence
of oxygen, the reaction realizes good carbon balance and produces
fewer by-products. Corma et al. [20] and Chai et al. [21–23] re-
ported catalytic conversion of aqueous glycerol solution without
oxygen over acid-type catalysts. In their studies, large amounts
of coke formation (3–30%) were observed. Even under similar reac-
tion conditions without adding oxygen, the present study showed
that VOHPO4ꢀ0.5H2O is a better catalyst for converting glycerol to
acrolein in high selectivity while maintaining high activity and
generating less coke formation and fewer by-products.
2
0
10
-2
-4
5
-6
3.4. Effect of reaction temperature
-8
-10
-12
-14
-16
-18
0
The conversion of glycerol and the conversion of oxygen
remarkably increased with increase in reaction temperature from
250 °C to 350 °C (Table 3). The selectivities to acrolein and hydrox-
yacetone increased simultaneously. At 350 °C, C–C bond breakage
took place as indicated by the generation of large amounts of C1
and C2 products. The selectivity to hydroxyacetone decreased from
29% at 250 °C to 0% at 350 °C, suggesting that 1,2-dehydration at a
lower temperature mainly occurs at the central carbon site. The
selectivity to acrylic acid was 8% at 350 °C, indicating that a high
temperature is favorable for oxidizing acrolein to acrylic acid.
Reaction temperature of 300 °C is suitable for achieving high
TG (with oxygen)
TG (without oxygen)
DTA (with oxygen)
DTA (without oxygen)
-5
100
200
300
400
500
600
700
Temperature / oC
Fig. 5. Comparison of TGꢁDTA profiles of the used catalyst from a reaction without
the addition of oxygen and the used catalyst from a reaction with an O2/glycerol
ratio of 9.1.

F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
265
Table 3
Effect of reaction temperature on catalytic results over the VOHPO4ꢀ0.5H2O catalyst.
Used catalyst
Temperature (°C)
250
275
300
325
350
Glycerol conversion (mol%)
Oxygen conversion (mol%)
Carbon balance (mol%)
68
0
70
78
10
74
100
14
91
100
30
80
100
27
74
Fresh catalyst
Products distribution (mol%)
Acrolein
Acetaldehyde
Hydroxyacetone
Acetone
12
2
29
0
25
5
14
7
66
15
4
48
18
0
36
20
0
0
0
0
Propanal
0
0
0
0
0
Acetic acid
Acrylic acid
CO
CO2
Other CHO
9
0
0
0
3
1
0
2
5
3
0
1
18
5
6
5
0
15
8
14
8
2000
1800
1600
1400
1200
1000
800
600
400
Wavenumber / cm-1
7
14
4
0
Fig. 6. IR spectra of used and fresh VOHPO4ꢀ0.5H2O catalyst. The used catalyst was
collected after removing the quartz diluents by sieving, and used without any
treatment.
Reaction conditions: see note in Table 1.
selectivity to acrolein and carbon balance larger than 90%. Com-
pared with the temperatures of 300–700 °C for conducting reac-
tions in previous works, the present reaction temperature is mild.
35
0
-4
30
25
20
15
10
5
3.5. Effect of space velocity rate
DTA curves
Change in the space velocity rate (WHSV) is realized by adjust-
ing the weight of the catalyst and keeping the feeding gas constant
(Table 4). The space velocity rate has little effect on product distri-
butions. The main products were acrolein and acetaldehyde. With
increase in WHSV from 2.6 to 20.5 hꢁ1, the conversion of oxygen
decreased. The conversion of glycerol was 100% even when the
reaction is performed at a high space velocity rate (20.5 hꢁ1), indi-
cating that the VOHPO4ꢀ0.5H2O catalyst is highly reactive for glyc-
erol dehydration.
440
434
-8
400 420 440 460 480 500
-12
-16
0
Solid line: fresh catalyst
Dot line: used catalyst
-5
50 100 150 200 250 300 350 400 450 500 550 600
Temperature / oC
3.6. Stability of the VOHPO4ꢀ0.5H2O catalyst
We used several techniques to study the stability of the VOH-
PO4ꢀ0.5H2O catalyst. We previously thought that the VOH-
PO4ꢀ0.5H2O was unstable and perhaps changed to other active
phases, catalyzing the reaction. However, the used catalyst showed
an IR spectrum similar to that of a fresh one as shown in Fig. 6. A
Fig. 7. TGꢁDTA curves of the used and fresh VOHPO4ꢀ0.5H2O catalyst.
ature of the used catalyst is 434 °C, which is 6 °C lower than that
(440 °C) of the fresh catalyst and (2) there was greater weight loss
for the used catalyst at temperatures between 400 °C and 500 °C.
The former difference is explained by the disturbance of crystalline
disorder by water absorption, and the latter is caused by slow
structure dehydration. Globally, the used catalyst has a weight loss
of 10.2%, and the fresh one has a weight loss of 9.5%. XRD charac-
terizations (Fig. 8) showed that the two catalysts have nearly iden-
tical structures except that the d spacing of the (0 0 1) facet of the
used catalyst is slightly shifted to a lower angle, implying widening
of the two adjacent [0 0 l] layers. Such an effect is probably caused
by partial hydration of the V@O or P@O bond between the two
adjacent layers. Based on IR and TGA results, it is concluded that
the VOHPO4ꢀ0.5H2O catalyst is stable during the reaction. Small
changes take place because of water adsorption since the reaction
is conducted under humid reaction conditions (partial water pres-
sure: 0.0287 MPa).
small increase in IR transmission intensity appears at 1620 cmꢁ1
,
which is assigned to the vibration of water, suggesting that the
used catalyst contains more water. TG–DTA analysis (Fig. 7) re-
vealed that the two catalysts have similar weight loss and thermal
transformation except that (1) the central decomposition temper-
Table 4
Effect of space velocity rate (WHSV) on catalytic results over the VOHPO4ꢀ0.5H2O
catalyst.
WHSV (hꢁ1
)
20.5
10.2
5.1
2.6
Glycerol conversion (mol%)
Oxygen conversion (mol%)
Carbon balance (mol%)
100
11
86
100
14
91
100
28
86
100
32
85
Products distribution (mol%)
Acrolein
Acetaldehyde
Hydroxyacetone
Acetone
65
18
0
66
15
4
62
14
0
62
17
0
3.7. Reaction pathway
0
0
0
0
Propanal
0
0
0
0
Acetic acid
Acrylic acid
CO2
5
4
2
2
5
3
1
4
10
6
3
5
4
7
2
The complexities of the dehydration of glycerol in the presence
of oxygen are largely because of the three hydroxyl groups and
highly functionalized derivatives. To classify the reaction path-
ways, we employed several major products as substrates to inves-
tigate their products under the same reaction conditions: reaction
Other CHO
1
Reaction conditions: see note in Table 1.

266
F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
2000
1000
0
ably generated from the oxidation of acetaldehyde. In the test, we
2000
1500
1000
500
0
did not observe the product of acrolein. This result suggests that
the dehydration of glycerol occurs in two distinct ways, one lead-
ing to hydroxyacetone and the other leading to acrolein.
d001 =0.5656 nm
Acetone conversion: The conversion of acetone was 10%, indicat-
ing that acetone was less active than acetaldehyde and hydroxyac-
etone, largely owing to its robust molecular structure. The major
product was acetic acid with 80% selectivity. Oxidation of acetone
needed breakage of either the C–C bond or the C–H bond. We did
not observe an oxidized product via C–H bond activation at the ter-
minal carbon site of acetone, and we detected acetic acid and acet-
aldehyde, formed by C–C bond activation. As shown by Corma
et al., acetone could be converted into butene and acetic acid via
a bimolecular mechanism [20]. We did not obtain a butane product
in this study.
d001 =0.5610 nm
Used catalyst
Fresh catalyst
14.5 15.0 15.5 16.0 16.5 17.0
10
20
30
40
50
60
Acrolein conversion: The conversion of acrolein was only 4% with
acetaldehyde, acrylic acid, acetic acid, and propanal as main prod-
ucts, indicating that acrolein is much less reactive than hydroxyac-
etone, acetaldehyde, and acetone. Although the reaction was
conducted in the presence of excess oxygen, it was still difficult
for acrolein to be oxidized. The test confirmed that the VOH-
PO4ꢀ0.5H2O catalyst has limited oxidation ability. Since 20% con-
version of acetaldehyde (C2) was achieved under the same
reaction conditions, we suggest that a steric effect or geometric
adsorption may play a key role, that is, a short molecule (C2) is eas-
ier to be oxidized than a longer molecule (C3).
Acrylic acid and acetic acid conversion: We are surprised that
both acrylic acid and acetic acid are inactive. It is generally
acknowledged that acid compounds are easily oxidized by C–C
activation to smaller molecule compounds. However, we observed
that these acids, such as acrylic acid and acetic acid, are stable in
the oxidation atmosphere over the VOHPO4ꢀ0.5H2O catalyst.
Based on the above results, we propose a simple reaction net-
work starting from glycerol (Fig. 10). The dehydration of glycerol
becomes more complex in the presence of oxygen, which possibly
involves dehydration, oxidation, hydrogenation, and dehydrogena-
tion reactions. When protonation occurs at the central hydroxyl
group of glycerol, a water molecule and a proton are eliminated
from the protonated glycerol, and then 3-hydroxypropanal is pro-
duced by tautomerism. In the actual reaction, 3-hydroxypropanal
2 Theta/theta
Fig. 8. XRD patterns of the used and fresh VOHPO4ꢀ0.5H2O catalyst.
temperature of 300 °C, 0.1 MPa, and 0.2 g of catalyst diluted by
3.0 g of quartz. Catalytic results are given in Fig. 9.
Acetaldehydeconversion: Twenty percent conversion of acetalde-
hyde was obtained with 100% selectivity to acetic acid. It was pro-
posed by Corma et al. that acetaldehyde also further reacted to
form oligomers [20]. However, we did not observe the formation
of oligomers, owing to the oxidative atmosphere.
Hydroxyacetone conversion: The conversion of hydroxyacetone
was 55%, which is far less than that of glycerol, suggesting that
hydroxyacetone is less active than glycerol. The products were ace-
tone (35%), acetic acid (28%), and acetaldehyde (21%). The forma-
tion of acetone revealed that some step may involve the
hydrogenation of 1,2-propanediol, an unstable intermediate that
was not detected in this study [48]. Surface hydroxyl and adsorbed
hydrocarbon species possibly provide hydrogen species for hydro-
genation. The formation of acetaldehyde may be via 2-oxoprop-
anal, an active intermediate formed by the dehydrogenation or
oxidation of hydroxyacetone [49]. Another missing carbon may
be converted to CO or adsorbed carbon species. Acetic acid is prob-
Fig. 10. Proposed general glycerol conversion scheme. The undetected products are
Fig. 9. Catalytic results using different substance over the VOHPO4ꢀ0.5H2O catalyst.
shown by dashed rectangles.

F. Wang et al. / Journal of Catalysis 268 (2009) 260–267
267
[11] Y. Fernandez, A. Arenillas, M.A. Diez, J.J. Pis, J.A. Menendez, J. Anal. Appl. Pyrol.
84 (2009) 145.
is readily dehydrated into acrolein. In contrast, when protonation
proceeds at a terminal hydroxyl group of glycerol, hydroxyacetone
is produced through dehydration and deprotonation accompanied
by tautomerism.
[12] S. Adhikari, S.D. Fernando, A. Haryanto, Chem. Eng. Technol. 32 (2009) 541.
[13] M. Bowker, P.R. Davies, L.S. Al-Mazroai, Catal. Lett. 128 (2009) 253.
[14] A. Karam, N. Villandier, M. Delample, C.K. Koerkamp, J.P. Douliez, R. Granet, P.
Krausz, J. Barrault, F. Jerome, Chem. Eur. J. 14 (2008) 10196.
[15] A. Wolfson, C. Dlugy, Y. Shotland, Environ. Chem. Lett. 5 (2007) 67.
[16] S. Srivastava, V. Srivastava, V. Gupta, L. Chaudhary, Oxid. Commun. 30 (2007)
553.
4. Conclusions
[17] W.C. Ketchie, M. Murayama, R.J. Davis, Top. Catal. 44 (2007) 307.
[18] N. Dimitratos, A. Villa, C.L. Bianchi, L. Prati, M. Makkee, Appl. Catal. A 311
(2006) 185.
[19] E. Schwenk, M. Gehrke, F. Aichner, US Patent 1 916 743, Schering-Kahlbaum
AG, 1933.
[20] A. Corma, G.W. Huber, L. Sauvanauda, P. O’Connor, J. Catal. 257 (2008) 163.
[21] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Green Chem. 10 (2008) 1087.
[22] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Appl. Catal. A 353 (2009) 213.
[23] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, J. Catal. 250 (2007) 342.
[24] E. Tsukuda, S. Sato, R. Takahashi, T. Sodesawa, Catal. Commun. 8 (2007) 1349.
[25] H. Atia, U. Armbruster, A. Martin, J. Catal. 258 (2008) 71.
[26] W. Yan, G.J. Suppes, Ind. Eng. Chem. Res. 48 (2009) 3279.
[27] Q.B. Liu, Z. Zhang, Y. Du, J. Li, X.G. Yang, Catal. Lett. 127 (2009) 419.
[28] L. Ott, M. Bicker, H. Vogel, Green Chem. 8 (2006) 214.
[29] M. Watanabe, T. Lida, Y. Aizawa, T.M. Aida, H. Inomata, Bioresour. Technol. 98
(2007) 1285.
[30] J.L. Dubois, C. Duquenne, W. Holderich, J. Kervennal, FR2882053 (A1), Arkema
(FR), 2006.
[31] J.L. Dubois, C. Duquenne, W. Holderich, FR2882052, Arkema (FR), 2006.
[32] G.J. Hutchings, J. Mater. Chem. 19 (2009) 1222.
[33] G.S. Patience, R.E. Bockrath, J.D. Sullivan, H.S. Horowitz, Ind. Eng. Chem. Res. 46
(2007) 4374.
[34] G. Centi, J.L. Nieto, D. Pinelli, F. Trifiro, Ind. Eng. Chem. Res. 28 (1989) 400.
[35] V.V. Guliants, S.A. Holmes, J.B. Benziger, P. Heaney, D. Yates, I.E. Wachs, J. Mol.
Catal. A 172 (2001) 265.
The performances of three VPO catalysts, VOPO4ꢀ2H2O, VOH-
PO4ꢀ0.5H2O, and (VO)2P2O7, in the dehydration of glycerol in the
presence of excess molecular oxygen were studied. The hemihy-
drate VOHPO4ꢀ0.5H2O, which has never been known as an active
catalyst, showed the best catalytic performance: 100% conversion
of glycerol with 66% selectivity to acrolein and 14% selectivity to
acetaldehyde. Both quantities and kinds of by-products were fewer
than those in the literature works. The best carbon balance was
achieved at 93%, which is much better than the literature results.
GC–MS and TGA analyses indicated that there is no coke formation
on the used catalyst when the reaction is conducted with an oxy-
gen/glycerol ratio of 9.1. XRD, IR, and TGA characterizations re-
vealed that the catalyst is stable during reaction. The reaction
pathways were uncovered and this information will be helpful
for further studies and as references. It has been shown that the
VOHPO4ꢀ0.5H2O catalyst has limited ability for catalyzing the oxi-
dation of acrolein to acrylic acid. In order to achieve the one-step
conversion of glycerol to acrylic acid, a catalyst with greater oxida-
tion ability and mild acidity is needed. Our future research will be
focused on improving the oxidation ability of VOHPO4ꢀ0.5H2O by
doping with other elements to achieve efficient conversion of glyc-
erol to acrylic acid in one-step.
[36] N. Ballarini, F. Cavani, C. Cortelli, S. Ligi, F. Pierelli, F. Trifiro, C. Fumagalli, G.
Mazzoni, T. Monti, Top. Catal. 38 (2006) 147.
[37] Q. Jiang, J. Zhao, X.X. Li, W.J. Ji, Z.B. Zhang, C.T. Au, Appl. Catal. A 341 (2008) 70.
[38] N. Yamamoto, N. Hiyoshi, T. Okuhara, Chem. Mater. 14 (2002) 3882.
[39] G. Koyano, T. Okuhara, M. Misono, J. Am. Chem. Soc. 120 (1998) 767.
[40] N. Mizuno, H. Hatayama, M. Misono, Chem. Mater. 9 (1997) 2697.
[41] C. R’Kha, M.T. Vandenborre, J. Livage, R. Prost, E. Huard, J. Solid State Chem. 63
(1986) 202.
References
[42] B.H. Sakakini, Y.H. Taufiq-Yap, K.C. Waugh, J. Catal. 189 (2000) 253.
[43] G.J. Hutchings, J. Mater. Chem. 14 (2004) 3385.
[44] S. Endres, P. Kampe, J. Kunert, A. Drochner, H. Vogel, Appl. Catal. A 325 (2007)
237.
[45] C. Schmitt, L. Giebeler, R. Schierholz, S. Endres, C. Fasel, H. Vogel, H. Fuess, Z.
Phys. Chem. 221 (2007) 1525.
[46] J.L. Dubois, C. Duquenne, W. Holderich, FR2884817, Arkema (FR), 2006.
[47] J. Barrault, J.M. Clacens, Y. Pouilloux, Top. Catal. 27 (2004) 137.
[48] M.R. Nimlos, S.J. Blanksby, X.H. Qian, M.E. Himmel, D.K. Johnson, J. Phys. Chem.
A 110 (2006) 6145.
[1] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411.
[2] Y. Zheng, X. Chen, Y. Shen, Chem. Rev. 108 (2008) 5253.
[3] C.-H. Zhou, J.N. Beltramini, Y.-X. Fan, G.Q. Lu, Chem. Soc. Rev. 37 (2008) 527.
[4] M. Aresta, A. Dibenedetto, F. Nocito, C. Pastore, J. Mol. Catal. A 257 (2006) 149.
[5] J.W. Yoo, Z. Mouloungui, Stud. Surf. Sci. Catal. 146 (2003) 757.
[6] G. Lewandowski, M. Bartkowiak, E. Milchert, Oxid. Commun. 31 (2008) 108.
[7] K. Murata, I. Takahara, M. Inaba, React. Kinet. Catal. Lett. 93 (2008) 59.
[8] E.P. Maris, R.J. Davis, J. Catal. 249 (2007) 328.
[9] T. Miyazawa, S. Koso, K. Kunimori, K. Tomishige, Appl. Catal. A 329 (2007) 30.
[10] S. Wang, H.C. Liu, Catal. Lett. 117 (2007) 62.
[49] M. Ai, K. Ohdan, Bull. Chem. Soc. Jpn. 72 (1999) 2143.