Direct conversion of glycerol to acrylic acid via integrated dehydration-oxidation bed system

Article abstract of DOI:10.1016/j.apcata.2011.10.045

Acrylic acid can be successfully produced in a single reactor via subsequent oxidation of the glycerol-dehydrated products. Selective dehydration of glycerol to acrolein was studied at 275-400 °C over HZSM-5, HBeta, HMordenite and HY. The V-Mo oxides (15-70 mol%V) on silicic acid support (20-100 wt% mixed oxides loading) were then included as a second bed for subsequent oxidation of the dehydrated products. Over the acid zeolites, acrolein and acetol are mainly generated, together with acetaldehyde, propionaldehyde, pyruvaldehyde and other oxygenates as secondary products. A complete conversion of glycerol with high selectivity to acrolein (up to 81 mol%) can be obtained when medium pore zeolites (HZSM-5) and low glycerol concentration (10-30 wt%) was used at 300 °C. A separated-sequential bed system provides high selectivity for acrylic acid with small amount of acetic acid and acetaldehyde (～15 mol%). The catalyst with high V content promotes total oxidation of the dehydrated products to CO while that with highly dispersed V-Mo-O phases affords 98% selectivity to acrylic acid with 48% acrolein conversion.

Full text of DOI:10.1016/j.apcata.2011.10.045

Applied Catalysis A: General 413–414 (2012) 109–116
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Direct conversion of glycerol to acrylic acid via integrated dehydration–oxidation
bed system
Ayut Witsuthammakul, Tawan Sooknoi∗
Department of Chemistry, Faculty of Science, King Mongkut’s Institute of Technology Ladkrabang, Chalongkrung Road, Bangkok 10520, Thailand
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 August 2011
Received in revised form 25 October 2011
Accepted 30 October 2011
Available online 6 November 2011
Acrylic acid can be successfully produced in a single reactor via subsequent oxidation of the glycerol-
dehydrated products. Selective dehydration of glycerol to acrolein was studied at 275–400 ◦C over
HZSM-5, HBeta, HMordenite and HY. The V–Mo oxides (15–70 mol%V) on silicic acid support (20–100 wt%
mixed oxides loading) were then included as a second bed for subsequent oxidation of the dehydrated
products. Over the acid zeolites, acrolein and acetol are mainly generated, together with acetaldehyde,
propionaldehyde, pyruvaldehyde and other oxygenates as secondary products. A complete conversion
of glycerol with high selectivity to acrolein (up to 81 mol%) can be obtained when medium pore zeo-
lites (HZSM-5) and low glycerol concentration (10–30 wt%) was used at 300 ◦C. A separated-sequential
bed system provides high selectivity for acrylic acid with small amount of acetic acid and acetaldehyde
(∼15 mol%). The catalyst with high V content promotes total oxidation of the dehydrated products to CO
while that with highly dispersed V–Mo–O phases affords 98% selectivity to acrylic acid with 48% acrolein
conversion.
Keywords:
Glycerol
Acrolein
Acrylic acid
HZSM-5
Vanadium–molybdenum oxides
Dehydration
Oxidation
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
C3 petrochemicals, namely acrylic acid [10–12], acrylonitrile [13],
allyl alcohol [14], etc. The dehydration can be generally promoted
Nowadays, attempts to replace the petroleum resource by
biomass have been increasingly focused for the production of both
liquid fuel and petrochemical feedstock. In particular, the use of
alkyl ester as biodiesel via trans-esterification between vegetables
oil and alcohol [1,2], becomes increasingly attractive for the coun-
tries with agriculture-based economics. However, a large amount
of glycerol, the by-product, is continually produced, while its cur-
rent demand remained unchanged. Despite efforts are made to
utilize glycerol as alternative fuel, its combustion emits toxic gases
[3]. Alternatively, glycerol is a sustainable feedstock that can be
used for manufacture of some C3 derivatives. Several processes, so
called “deoxygenation”, were applied to eliminate its oxygen con-
tent. These include hydrogenolysis [4,5] and dehydration [6,7]. For
example, conversion of glycerol to propylene glycol [8] has been
widely investigated. However, most of the process consumes large
amount of hydrogen and high pressure is generally applied.
over acid catalysts at 275–350 ◦C. The common acid oxides such
as ␩-Al2O3, SiO2/Al2O3 [15] and TiO2 [16] have been applied but
they did not provide a satisfied selectivity to acrolein. Improved
selectivity can be obtained from the reactions over Nb2O5 (∼50%)
[17], WO3/ZrO2 (75%) [18], phosphoric-doped metal oxides (i.e.
Al2O3–PO4, TiO2–PO4 [19]), heteropoly acids (i.e. HSiW, HPW,
HPMo on silica supports [16]) and zeolites (i.e. ZSM-5, Y, Beta
[20–22], and SAPOs [19]). However, low selectivity is typically
obtained at high conversion and the catalysts usually possess low
stability. This is presumably because acrolein is highly reactive
and undergoes secondary reactions to form other products during
the dehydration process. Accordingly, a subsequent conversion of
acrolein to a desirable and more stable product, immediately after
its formation, would not only enhance the designed product selec-
tivity, but also eliminate handling problems of acrolein storage and
transportation.
In an economic point of view, deoxygenation of glycerol via
dehydration become more attractive since more than 60% oxygen
content can be simply eliminated without hydrogen consump-
tion under relatively mild reaction condition [9]. Moreover, the
dehydrated product, acrolein, can be readily converted to other
Major industrial use of acrolein is for production of acrylic acid
that is widely used for in adhesive, paint, plastic and rubber syn-
thesis. Typical oxidation catalysts involve vanadium–molybdenum
mixed oxides on alumina or silica supports due to their high selec-
tivity to acrylic acid [23]. It was also report that Mo:V ratio and
oxygen partial pressure play important role on the catalyst perfor-
mance [23]. In general, selective oxidation competes against total
combustion and the TPR experiment showed that CO2 could be
minimized at temperature below 400 ◦C [24]. While acrylic acid
∗
Corresponding author. Tel.: +66 81 929 8288; fax: +66 2 326 4415.
E-mail address: kstawan@kmitl.ac.th (T. Sooknoi).
0926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2011.10.045

110
A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
could be selectively generated with low conversion at temperature
between 300 and 450 ◦C, depending on Mo:V ratio [25].
Accordingly, it is possible to match up the reaction conditions of
both dehydration and oxidation for a two-bed system that allows
single stage conversion of glycerol to acrylic acid with high selec-
tivity. To accomplish the idea, a complete conversion of glycerol
and high acrolein selectivity are required in the dehydration step
to avoid oxidation of glycerol, and maximize acrylic acid yield in the
second bed. In addition, no oxidation should take place in the first
bed (dehydration), while water shall not interfere the oxidation
process. Thus, the two-bed system would allow feeding glycerol
with high water content and the product effluent would be acrylic
acid solution that can be directly used for production of acrylic
emulsion [26]. In this paper, we demonstrate a practical two-bed
system consisting of zeolite and vanadium–molybdenum mixed
oxides as catalysts for direct conversion of glycerol to acrylic acid.
The dehydration of glycerol over zeolites was primarily investi-
gated at various concentrations. The effects of temperature, oxygen
partial pressure, mixed oxide composition and loading on the sub-
sequent oxidation were also highlighted.
Fig. 1. The glycerol conversion over HZSM-5, HBeta, HMordenite and HY. Reaction
conditions; contact time: 177 g h mol−1, temperature: 300 ◦C, feed: 1.7 mmol h−1 of
glycerol at 10 wt%, carrier gas: 30 ml min−1 of helium.
product was quantitatively analyzed by gas chromatography with
a flame ionization detector (GC–FID). A capillary EC-Wax (Alltech;
0.53 mm × 30 m) was used as separating column. While, the gas
products was analyzed by the online gas chromatography equipped
with thermal conductivity detector (GC–TCD) and a molecular
sieve 13X (1/8 in. × 8 ft) column. Conversion, yield, selectivity and
turnover frequency can be calculated as following equations;
2. Experimental procedure
HZSM-5 (13) and NH4+ZSM-5 (140) were obtained from
Zeochem® and Zeolyst® respectively. While NH4+Beta (14), HMor-
denite (15) and HY (100) were supplied by Tohso Co., Ltd. All
zeolite samples were calcined in air zero at 500 ◦C for 5 h, then
pelletized to the size of 600–850 ␮m. For the oxidation catalyst,
the vanadium–molybdenum oxides on silicic acid support was pre-
pared using incipient wetness impregnation technique. Firstly, the
ammonium metavanadate (NH4VO3; Asia Pacific Specialty Chem-
icals Ltd.) and ammonium heptamolybdate ((NH4)6Mo7O24·4H2O;
Ajax Fine Chem) precursors were dissolved in distillated water at
60 ◦C. The precursor solution was slowly dropped onto the silicic
acid support (chromatography grade; BDH Chemicals Ltd.) until
wet. The samples were dried at 60 ◦C in an oven for 15 min, and
then the loading was repeated until metal content was reached.
The final loaded samples were kept to dry at 60 ◦C over night. After
that, the catalysts were calcined at 300 ◦C for 5 h under a flow of
dry air (60 ml min−1) and pelletized to the size of 600–850 ␮m. The
sample were designated as [metal oxide loading]VMo[vanadium
mol% in mixed metal oxide]; i.e. 45VMo(21).
Surface area of all catalysts was determined using nitro-
gen adsorption analyzer (Quantachrome Autosorb-1) at −196 ◦C
and 0.05–0.30P/P0. The BET equation was used for surface
area calculation. Elemental composition and crystal structure of
vanadium–molybdenum oxides on silicic acid were determined by
X-ray fluorescence spectrometer (XRF; Siemens SRS3400) and X-
ray diffractometer (XRD; Siemens D8 Advance) respectively. The
morphology and deep-surface composition of the catalysts were
analyzed by field emission spectrometer equipped with energy-
dispersive X-ray spectrometer (SEM–EDX; Hitachi S-4800).
The reaction was conducted in a tubular down flow reac-
tor (6 mm i.d. Pyrex®) located in an electrical furnace. Zeolite
was packed as a single bed catalyst for dehydration testing. The
vanadium–molybdenum oxide on silicic acid was packed under
the zeolite bed for subsequent oxidation testing. Helium and air
zero mixed with nitrogen were regulated using mass flow con-
trollers. The catalysts were activated at 300 ◦C (1 K min−1) for 5 h
in stream of air zero (30 ml min−1). Then, the temperature was
adjusted to the reaction temperature (275–400 ◦C) and hold for
30 min before the reaction. Glycerol solution (10–50 wt%) was
then fed at 1.7 mmol min−1 using a syringe pump. The reaction
was carried out for 7 h on stream. The product effluence was
trapped by a cooling condenser (<5 ◦C) and collected hourly. Liquid
%Conversion
(mole of glycerol fed − mole of glycerol remaining) × 100
=
mole of glycerol fed
mole of product × 100
%Yield =
mole of glycerol fed
product yield × 100
%Selectivity =
glycerol conversion
Normalized reaction rate
mole of acrolein (from 1st bed) converted per hour
=
mole of V + Mo in the 2nd bed
3. Result
3.1. Dehydration of glycerol to acrolein
The dehydration of glycerol was primarily investigated to
understand the catalytic parameters that effect acrolein production
in the first bed. The appropriate catalyst and condition are chosen
for further study.
3.1.1. Effect of zeolite frameworks
For the dehydration of glycerol over different zeolites, high con-
version can be initially observed over HZSM-5 (13) and HZSM-5
(140) while other catalysts show a severe drop in activity after cer-
tain time on stream (Fig. 1). As the dehydrated product, acrolein, is
highly reactive, it can be readily activated and undergo secondary
reactions such as condensation and oligomerization forming coke
deposit on the catalyst surface. The thermal degradation study of
the used catalysts (Table 1) confirms that hard coke is found for
all catalysts. Thus, it is believed that the HZSM-5 (13) and HZSM-5
(140) were also continually deactivated but the excess acid sites
kept the conversion 100%, despite part of the catalyst bed would
gradually loss their active sites. In addition to high activity, HZSM-
5 also gives higher selectivity to acrolein, as compared to HBeta,
HMordenite and HY (Table 2). While, the large pore zeolites yield
large amount of acetaldehyde and other high molecular weight

A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
111
Table 1
Thermo gravimetric analysis of various zeolites after 7 h on stream.
Framework type
Reaction temperature (◦C)
Glycerol concentration (wt%)
Contact time (g h mol−1
)
Coke (wt% of used catalyst)
HZSM-5 (13)
HBeta (14)
300
300
300
300
325
400
300
300
300
10
10
10
10
10
10
10
30
50
177
177
177
177
177
177
59
11.8
23.7
10.3
5.9
6.2
7.1
12.8
12.8
13.5
HMordenite (15)
HZSM-5 (140)
HZSM-5 (140)
HZSM-5 (140)
HZSM-5 (13)
HZSM-5 (13)
HZSM-5 (13)
59
59
10 K min−1 in air from 50 ◦C to 900 ◦C.
Table 2
The products selectivity and conversion of glycerol dehydration over HZSM-5, HBeta, HMordenite and HY.
Catalyst (Si/Al) Aciditya (␮mol/g) Conversion (mol%) Selectivity (mol%)
Acetaldehyde Propionaldehyde Acrolein Pyruvaldehyde Acetol Acetic acid Propanoic acid Others
HZSM-5 (13)
HBeta (14)
HMor (15)
HZSM-5 (140)
HY (100)
27.3
29.9
25.8
2.34
1.73
100.0
90.66
10.57
100.0
5.79
5.71
13.95
7.44
6.23
25.79
2.19
2.72
1.52
2.67
3.26
81.10
63.97
67.25
78.24
41.09
0.64
2.29
3.43
1.40
5.17
10.11
10.50 2.38
10.27 2.55
10.28
8.09 1.10
0
0
0.25
4.04
4.68
2.59
14.67
0.14
2.85
0
0
0.84
a
Data obtained from NH3-TPD (150–700 ◦C, 10 K min−1) of the sample calcined at 500 ◦C. The reaction condition is similar to Fig. 1. The results average between 5th and
7th hour on stream.
Table 3
products. They are cyclic unsaturated oxygenate compounds such
as phenol, dihydrofuran, cyclopentenone, methyl cyclopentenone
The acetol conversion over HZSM-5 (140).
and cyclohexenone. These products are referred to as “other prod-
ucts” that can be evolved to coke deposit causing the catalyst
deactivation, particularly for those with large pore size.
Products
Yield (mol%)
Selectivity (mol%)
Acetaldehyde
Propionaldehyde
Acetone
1.63
1.10
1.93
6.28
4.21
7.44
Others
21.10
81.17
3.1.2. Effect of contact time
Reaction conditions; contact time: 142 g h mol−1
,
temperature: 350 ◦C, feed:
For dehydration, the contact time that provides the complete
conversion is required because unreacted glycerol can interfere
with the oxidation of the dehydrate products in the second bed. As
shown in Fig. 2a, a complete conversion can be obtained at con-
tact time >58 g h mol−1. Acrolein and acetol are major products
generated in parallel from glycerol dehydration [16,19,21] while
acetaldehyde is presumably derived from the decomposition of the
feed [16]. It is noted that pyruvaldehyde and propionaldehyde are
somewhat equally found (∼2–3%) as minor products as shown in
Fig. 2b. When an excess of catalyst is used (>58 g h mol−1), the ace-
tol yield drops gradually while acrolein and acetaldehyde yields
remain unchanged. Together with an observed increase in other
products, this suggests that acetol can further react over acid sites
to form secondary products. In a support manner, a separate exper-
iment shows that acetol conversion on HZSM-5 gives mainly other
2.1 mmol h−1 of acetol at 10 wt%, carrier gas: 30 ml min−1 of helium; the results
average between 5th and 7th hour on stream; only 26 mol% acetol is converted.
products with a small amount of acetone, propionaldehyde and
acetaldehyde (Table 3). In addition, 5.2 wt% coke is found for the
used catalyst.
3.1.3. Effect of reaction temperature
Fig. 3 shows that the glycerol conversion increases with the
reaction temperature and a complete conversion can be obtained
at ≥300 ◦C. However, the selectivity toward dehydrated prod-
ucts (acrolein and acetol) readily decreases when the temperature
is increased. In the opposite manner, acetaldehyde selectivity
increases. This suggests that, at ≥300 ◦C, decomposition of the feed
Fig. 2. Glycerol conversion and products yields over HZSM-5 (140) as function of contact time. Reaction conditions; temperature: 350 ◦C, feed: 1.7 mmol h−1 of glycerol at
10 wt%, carrier gas: 30 ml min−1 of helium; (a) major products and (b) minor products; the results average between 5th and 7th hour on stream.

112
A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
when the feed concentration is higher than 30 wt%. In addition,
the secondary products including acetaldehyde, cyclic unsaturated
oxygenate compounds and coke (Table 1) are also increased. This
shows that water content plays marked roles in product selectivity,
presumably via interaction with glycerol and the active acid sites.
As intra-molecular hydrogen bonding between glycerol become
more attractive in concentrated solution, the dehydration selec-
tivity may well be modified. Hence, the feed concentration of
10–30 wt% was used for further study, in order to maximize the
acrolein yield.
3.2. Subsequent oxidation of glycerol-dehydrated products
3.2.1. Separated versus mixed bed system
Fig. 3. Glycerol conversion and product selectivities over HZSM-5 (140) as func-
tion of reaction temperature. Reaction conditions; contact time: 177 g h mol−1, feed:
1.7 mmol h−1 of glycerol at 10 wt%, carrier gas: 30 ml min−1 of helium; the results
average between 5th and 7th hour on stream; propionaldehyde and pyruvaldehyde
were found for 3% and 2% respectively.
Separate sequential beds of HZSM-5 (13) and 45VMo(21) and
a single bed containing physical mixture of above catalysts were
studied comparatively. It is important to note that no oxidation
takes place in the first bed when oxygen is used as carrier gas,
and the observed oxidized products in both separated and mixed
bed system are solely contributed from the oxidation activity of
the additional mixed oxide catalysts. It can be seen in Fig. 5 that
acrylic acid is successfully produced while acetol disappears from
the product stream. It is likely that acetol can be readily oxidized
presumably to acetic acid [27–29]. This is deduced from a similar
yield of acetic acid observed in the separated bed system (Fig. 5a),
as compared to the acetol produced from the first bed (Fig. 4). The
separate bed system gives ∼40 mol% yield of acrylic acid (Fig. 5a),
while the mixed one gives considerable amounts of acetic acid
and acetaldehyde, together with only ∼30 mol% yield of acrylic
acid (Fig. 5b). One may expected that acrylic acid should be solely
produced by acrolein oxidation. Hence, yields of C3 oxygenates
(acrolein + acrylic acid) would represent the dehydration selectivity
toward desirable products, while %acrylic acid over the C3 oxy-
genates would indicates an oxidation activity for the system. It can
be seen that lower C3 oxygenates yield is obtained from the mixed
bed system, as compared with that from the separated bed. While,
the oxidation activity is somewhat higher. It is likely that, in the
mixed catalysts system, direct oxidation of glycerol to undesirable
C2 oxygenates [30–32] can be promoted when glycerol is in con-
tact with the oxidation catalyst. While, in the separate bed, glycerol
is completely converted in the first bed (HSZM-5) and only dehy-
drate products are oxidized in the second one (45VMo(21)). The
result also shows that no change in products distribution over 7 h
on stream for both systems. In line with this result, thermogravi-
metric analysis of the used 45VMo(21) reveals no coke deposit
suggesting good stability of the vanadium–molybdenum oxides
catalyst.
Fig. 4. Glycerol conversion and product selectivities over HZSM-5 (13) as function
of feed concentration. Reaction condition; contact time: 59 g h mol−1, temperature:
300 ◦C, carrier gas: 30 ml min−1 of helium; the results average between 5th and 7th
hour on stream.
becomes increasingly competitive with the dehydration. Moreover,
as a relatively higher coke deposits are obtained at higher temper-
ature (Table 1), it is likely that secondary reactions of the products
can be promoted over the acid catalyst when the reaction temper-
ature is increased.
3.1.4. Effect of glycerol concentration
As an acrylic emulsion is desirable for this work, aqueous glyc-
erol (10–50 wt%) was tested as feed for dehydration over HZSM-5
(Si/Al ∼13). It can be seen from Fig. 4 that the selectivity to acrolein
decreases, together with an increase of acetol selectivity, only
Fig. 5. Comparative yields of (a) separate bed and (b) mix bed for subsequent oxidation of glycerol dehydrated products. Reaction condition; temperature: 300 ◦C, feed:
1.7 mmol h−1 of glycerol at 10 wt%, carrier gas: 30 ml min−1 of 20 vol% O2 in N2, separate system catalyst: 1st bed: contact time: 88 g h mol−1, catalyst: HZSM-5 (13), 2nd bed:
contact time: 177 g h mol−1, catalyst: 45VMo(21), mix system: contact time: 265 g h mol−1, catalyst 1:2 (by wt) of HZSM-5 (13): 45VMo(21). For all cases, the contact time is
on glycerol basis.

A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
113
Fig. 6. The yield from subsequent oxidation of glycerol dehydrated products as function of second bed contact time. Reaction condition; temperature: 300 ◦C, 1st bed: contact
time: 88 g h mol−1, catalyst: HZSM-5 (13), 2nd bed: catalyst: 45VMo(21), feed: 1.7 mmol h−1 of glycerol at 10 wt%, carrier gas: 30 ml min−1 of 10 vol% O2 in N2; (a) is major
products and (b) is minor products; the results average between 5th and 7th hour on stream. For all cases, the contact time is on glycerol basis.
3.2.2. Effect of contact time
It can be seen from Fig. 6a that yield of acrylic acid is proportion-
ally increased with a decrease in acrolein as the contact time of the
second bed is increased. An exactly inversed slope of their changes
suggests that high selectivity for acrolein oxidation to acrylic acid
can be obtained throughout the range of contact time studied. For
minor products (Fig. 6b), yield of acetol sharply drops while yield
of decomposition products including acetaldehyde and acetic acid
markedly increase when the contact time is increased. In consistent
with the result from acetol conversion (Table 4), it is clear that ace-
tol was readily oxidized/decomposed to acetic acid, acetaldehyde
and pyruvaldehyde. Moreover, equivalent mole of CO is produced
Fig. 7. The yield for subsequent oxidation of glycerol dehydrated products as func-
as respected to mole of the acetol converted (not shown). At contact
tion of vanadium content in metal. Reaction condition; temperature: 300 ◦C, 1st
time >120 g h mol−1, the decrease in acetaldehyde is observed with
bed: contact time: 88 g h mol−1, catalyst: HZSM-5 (13), 2nd bed: contact time:
an increase in acetic acid yield. While propionaldehyde gradually
decreases presumably by oxidation to acrolein as the contact time is
increased. The presence of small aldehyde at fairly low contact time
indicates partial oxidation activity of the catalysts. It is also worth
noting that the minor products appear to derive mainly from acetol
and acetaldehyde produced from the first bed and relatively high
selectivity of acrolein oxidation can be obtained.
177 g h mol−1, catalyst: 45VMo(y), feed: 1.7 mmol h−1 of glycerol at 10 wt%, carrier
gas: 30 ml min−1 of 10 vol% O2 in N2; the results average between 5th and 7th hour
on stream. For all cases, the contact time is on glycerol basis.
3.2.4. Effect of mixed oxides loading
As the mixed oxide catalyst with low vanadium content
(21 mol%) provides reasonably activity and high acrylic acid selec-
tivity without carbon monoxide formation, effect of mixed oxides
loading on silicic support were tested at higher W/F (295 g h mol−1).
It can be seen in Fig. 8 that the acrolein conversion increases when
the mixed oxides loading is increased from 20 to 50 wt%. This is sim-
ply due to an increase in the active phase of the catalysts. However,
3.2.3. Effect of the mixed oxides composition
The composition of metal oxide strongly affects the products
distribution as shown in Fig. 7. It is clear that oxidation activity
is enhanced as vanadium content is increased. Acrolein conver-
sion is readily promoted with a proportional increase in acrylic
acid yield over the catalysts with 15–40% vanadium. However,
the decomposition–oxidation product i.e. carbon monoxide con-
tinuously increases despite the conversion of acrolein remain
unchanged when the vanadium content is higher 40%. Together
with a decrease in acrylic acid yield, it is suggested that a sub-
sequent oxidation of acrylic acid become more favorable at high
vanadium content. A similar result was reported by Kuznetsova
et al. [33], and this was presumably due to a competitive adsorption
of acrylic acid over the vanadium active sites.
Table 4
The acetol conversion over 45VMo(21).
Products
Yield (mol%)
Fig. 8. The yield for subsequent oxidation of glycerol dehydrated products as
function of vanadium–molybdenum oxides loading on silicic acid and normal-
ized reaction rate. Reaction condition; temperature: 300 ◦C, 1st bed: contact time:
88 g h mol−1, catalyst: HZSM-5 (13), 2nd bed: contact time: 295 g h mol−1, cata-
lyst: xVMo(21), feed: 1.7 mmol h−1 of glycerol at 10 wt%, carrier gas: 30 ml min−1
of 10 vol% O2 in N2; the results average between 5th and 7th hour on stream. For all
cases, the contact time is on glycerol basis.
Acetaldehyde
Pyruvaldehyde
Acetic acid
5.79
2.04
92.17
Reaction condition; contact time: 710 g h mol−1
,
temperature: 300 ◦C, feed:
0.4 mmol h−1 of acetol at 2 wt%; the results average between 5th and 7th hour on
stream; 100 mol% acetol is converted.

114
A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
presence of water in the feed can somewhat reduce the strong
adsorption of the intermediates and products that preferentially
retain on the catalyst surface. This rather suppresses the forma-
tion of the “surface oxygenate pools”, leading to better stability
and selectivity to acrolein when a diluted feed is employed. How-
ever, such “surface oxygenate pools” can be readily cracked to cyclic
unsaturated oxygenate compounds and acetaldehyde [34–36], par-
ticularly at high temperature (Fig. 3). This is also the case for
reaction with high feed concentration (Fig. 4) and that using acetol
as feed (Table 3) where high extent of the pools would be accu-
mulated. Meanwhile, such pools can evolve to coke deposit that
is the cause of catalyst deactivation in all cases (Fig. 1). In line
with the small oxygenates produced, relatively more coke deposit
is observed in the used catalyst (Table 1) at high reaction temper-
ature. In addition, the evolving coke species would facilitate some
hydrogen transfer, which is evidenced by trace of propionaldehyde
and pyruvaldehyde in the dehydration product stream.
Fig. 9. The yield for subsequent oxidation of glycerol dehydrated products as func-
tion of oxygen mole fraction in carrier gas and the selectivity to acrylic acid.
Reaction condition; temperature: 300 ◦C, 1st bed: contact time: 88 g h mol−1, cata-
lyst: HZSM-5 (13), 2nd bed: contact time: 177 g h mol−1, catalyst: 45VMo(21), feed:
1.7 mmol h−1 of glycerol at 10 wt%, Carrier gas: 30 ml min−1 of 5–20 vol% O2 in N2;
the results average between 5th and 7th hour on stream. For all cases, the contact
time is on glycerol basis.
For the HZSM-5
- a medium pore zeolite, condensa-
tion/oligomerization of intermediate products to form “surface
oxygenate pools” is somewhat restricted, as compared to that in
HY, HBeta and HMordenite. Moreover, the two-dimensional pore
system may well facilitate mass transfer of the products to diffuse
out of the pore. Accordingly, the coke deposit is relative low, as
compared that on other acid catalysts (Table 1). Although coke
is continuously accumulated, the selectivity to acrolein remains
relatively high for HZSM-5 (Table 2). In contrast, the secondary
reactions can be readily promoted in the large cavities of HY, HBeta
and Hmordenite. Larger extend of surface oxygenate pools, and
hence coke, can be produced blocking access of the feed to the active
sites. This leads to a severe deactivation of these catalysts after cer-
tain time on stream (Fig. 1). Nevertheless, the three-dimensional
pore system of HBeta affords a better mass transfer, as compared
to those in the one-dimensional pore of HMordenite and in a large
pocket of HY cage structure. Thus, the HBeta can be active for
a longer period of time as opposed to other large pore zeolites,
despite large amount of coke is deposited (Table 1). In line with
this view, high selectivity to cyclic unsaturated oxygenates and
acetaldehyde is obtained, particularly for HY (Table 2).
Dehydrated products are successfully oxidized over
vanadium–molybdenum oxide bed. Acrylic acid can be selec-
tively obtained only in the case of the separated-bed system since
glycerol competitively interacts and is oxidized over the mixed
oxide catalyst, leading to high yield of small oxygenates in the
mixed bed system (Fig. 5b). This is also the case for the separated
bed when glycerol is incompletely converted over the first bed.
In addition to acrolein, minor products derived from first bed
can be readily decomposed–oxidized over the second bed. Acetol
undergo oxidative dehydrogenation primarily to pyruvaldehyde
and subsequently decomposed to acetaldehyde and acetic acid
[27,28] and carbon monoxide. The acetaldehyde both produced
from the first bed and acetol decomposition is readily oxidized
to acetic acid while propionaldehyde can also undergo oxidative
dehydrogenation/oxidation to acrolein. It appears that all minor
products were eventually oxidized to acetic acid and the selectivity
of acrylic acid is determined by selectivity of acrolein in the first
bed. The reaction pathways for the two-separated bed system can
be proposed as shown in Fig. 10.
the acrolein conversion decreases gradually as the mixed oxides
loading reaches ∼50 wt%. This is presumably due to a poor disper-
sion of the mixed oxide phase as indicated by a traditional volcanic
plot of the turnover frequency (Fig. 8). The catalyst with 45 wt%
mixed oxides loading appears to be the most efficient in the study
range.
3.2.5. Effect of oxygen concentration
Oxygen concentration plays important role in oxidation activity
as shown in Fig. 9. Yield of acrylic acid increases when the oxygen
concentration is increased while a slightly decrease in selectivity of
the decomposed products including acetaldehyde and acetic acid
is also observed. The selectivity to acrylic acid also increases up
to 98 mol% (based on acrolein converted) at 0.20 O2 mole fraction
in carrier gas (Fig. 9) and no carbon dioxide is detected despite gas
phase O2 is ∼4 times in excess of the glycerol fed. This suggests that
oxidation of the dehydrated products of glycerol proceed catalyt-
ically with no homogeneous combustion taken place. However, a
slight decrease in the total yield of C2 oxygenates can be attributed
to the decomposition–oxidation to carbon monoxide. It is noted
that, over the catalyst bed, the O2 in carrier gas is diluted by large
amount of steam from water contained in the glycerol feed. The val-
ues shown in Fig. 9 are the component of the carrier gas (O2 + N2),
not the partial pressure of O2 in the gas phase.
4. Discussion
As glycerol possesses two different hydroxyl groups, the dehy-
dration typically takes place either at secondary or primary
hydroxyl group resulting in two parallel products, acrolein and ace-
tol, respectively [16]. It is clear from Fig. 2 that glycerol is mainly
dehydrated at the secondary hydroxyl. As a result, an intermedi-
ate called 3-hydroxypropionaldehyde (3-HPA) would be formed
and subsequently dehydrated to acrolein as major product. This is
because protonation of a secondary hydroxyl (a relatively electron-
rich moiety) would be more favorable over the Brønsted acid sites in
the zeolite. However, the protonation is less selective at higher glyc-
erol concentration, and an increase in acetol selectivity is observed
(Fig. 4). This is because, at high concentration, the hydrogen bond-
ing between glycerol molecules becomes relatively strong and the
difference between primary and secondary hydroxyl group is elim-
inated. Both can be readily protonated, hence; a low selectivity to
acrolein is obtained.
With an optimum condition from the first bed, the acrolein
conversion can be improved by an increase in loading of the
well dispersed mixed oxides on the silicic acid support. How-
ever, the active surface is readily reduced with the loading. This
is due to agglomeration of the mixed oxides phase as seen
by the increased crystallinity of the mixed oxides (XRD, not
shown). Accordingly, an optimum mixed oxides loading is found
at ∼48 wt%. Not only the loading of the active phase, the compo-
sition of mixed oxides strongly also influences both activity and
As 3-HPA and acetol are very reactive [18,19], they can also
undergo secondary reactions, i.e. condensation/oligomerization,
to form higher molecular weight oxygenates on the surface. The

A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
115
Fig. 10. The overall proposed reaction pathways for the glycerol dehydration and subsequent oxidation to acrylic acid.
selectivity since the metal ions in the mixed oxides possess dif-
5. Conclusion
ferent role in Mars-van Krevelen oxidation–reduction cycles [37],
The increase in vanadium content readily increases the V–Mo–O
hexagonal [38,39], orthorhombic [40] and triclinic phases [39,41]
which are more active for acrolein oxidation, as compared to ␣-
MoO3 alone [42,43]. However, the increase in vanadium content
>21 mol% results in formation of excess V2O5 phase [44] that pro-
motes the decomposition–oxidation of oxygenates to undesirable
carbon monoxide [33]. In the mechanistic point of view, acrolein is
oxidized over the surface oxygen of the mixed oxides [23–25] and
the recovery of this active site becomes a rate-determining step.
Accordingly, an increase in the oxygen partial pressure readily pro-
motes oxidation activity and selectivity. With low concentration of
oxygen in the gas phase, the active oxygen may not only be ade-
quately replenished, but an oxygen vacancy may be also created.
This site is inactive for oxidation but can lead to strong adsorption
of acrolein and decomposition to small oxygenates. Over a well-
dispersed V–Mo–O species with relatively low V content (<21%),
such side reaction can be inhibited despite of high oxygen partial
pressure. The observed acetaldehyde, acetic acid, and CO are the
results from the decomposition/oxidation of the minor products
from the first bed. In this study, up to ∼98 mol% selectivity to acrylic
acid (based on acrolein conversion) can be achieved at 0.20 mole
fraction of O2 in carrier gas, over 45% loading of VMo(21) catalyst in
the second bed.
Dehydration/subsequent oxidation to produce selectively
acrylic acid from glycerol can be achieved using a separate
HZSM-5/vanadium–molybdenum oxides bed system. A complete
conversion in the first bed is required to avoid oxidation of glycerol
in the second bed. Selectivity toward acrylic acid is determined
by acrolein selectivity from the first bed and that is drastically
decreased at temperature >300 ◦C and when the glycerol concen-
tration is higher 30 wt%. This is due to decomposition/evolution
of the “surface pools” to small oxygenates (i.e. acetaldehyde, ace-
tol, acetic acid) and coke deposit that readily leads to catalyst
deactivation. Alternatively, more suitable process i.e. fluidized and
moving bed reactor could be applied and the catalyst regenera-
tion should be further investigated. Increasing O2 partial pressure
also enhance the acrylic acid selectivity. The increase in mixed
oxides loading on silicic acid support readily improves the oxida-
tion activity, as long as a high dispersion of the active phases is
retained. The molybdenum rich catalyst possesses low activity due
to the formation of the less active ␣-MoO3. On the other hand, the
vanadium rich catalyst shows a high activity but low selectivity
to acrylic acid due to the formation of excess V2O5 that promotes
the further oxidation to CO. This study obviously shows the poten-
tial of subsequent oxidation of glycerol-dehydrated products to the
industrial practice. The glycerol can be converted to higher value

116
A. Witsuthammakul, T. Sooknoi / Applied Catalysis A: General 413–414 (2012) 109–116
chemicals i.e. acrylic acid, acetic acid and CO in a single reactor.
However, improved acrolein conversion should be further devel-
oped.
[17] S.-H. Chai, H.-P. Wang, Y. Liang, B.-Q. Xu, J. Catal. 250 (2007) 342–349.
[18] A. Ulgen, W. Hoelderich, Catal. Lett. 131 (2009) 122–128.
[19] W. Suprun, M. Lutecki, T. Haber, H. Papp, J. Mol. Catal. A: Chem. 309 (2009)
71–78.
[20] Y.T. Kim, K.-D. Jung, E.D. Park, Microporous Mesoporous Mater. 131 (2010)
28–36.
[21] A. Corma, G.W. Huber, L. Sauvanaud, P. O’Connor, J. Catal. 257 (2008) 163–171.
[22] Y.T. Kim, K.-D. Jung, E.D. Park, Appl. Catal. A: Gen. 393 (2011) 275–287.
[23] J. Tichy, Appl. Catal. A: Gen. 157 (1997) 363–385.
Acknowledgement
The authors are grateful to PTT Public Co. Ltd., for part of the
financial support.
[24] J. Kunert, A. Drochner, J. Ott, H. Vogel, H. Fue␤, Appl. Catal. A: Gen. 269 (2004)
53–61.
[25] A.H. Adams, F. Haa␤, T. Buhrmester, J. Kunert, J. Ott, H. Vogel, H. Fuess, J. Mol.
Catal. A: Chem. 216 (2004) 67–74.
References
[26] D. Benda, J. Snuparek, V. Cermak, Eur. Polym. J. 37 (2001) 1247–1253.
[27] M. Ai, K. Ohdan, Bull. Chem. Soc. Jpn. 72 (1999) 2143–2148.
[28] M. Ai, A. Motohashi, S. Abe, Appl. Catal. A: Gen. 246 (2003) 97–102.
[29] X. Li, E. Iglesia, Appl. Catal. A: Gen. 334 (2008) 339–347.
[30] F. Wang, W. Ueda, Catal. Today 144 (2009) 358–361.
[31] E. Santacesaria, A. Sorrentino, R. Tesser, M. Di Serio, A. Ruggiero, J. Mol. Catal.
A: Chem. 204–205 (2003) 617–627.
[32] F. Wang, J.-L. Dubois, W. Ueda, J. Catal. 268 (2009) 260–267.
[33] T.G. Kuznetsova, T.V. Andrushkevich, T.P. Gorshkova, React. Kinet. Catal. Lett.
30 (1986) 149–156.
[34] X. Zhu, L.L. Lobban, R.G. Mallinson, D.E. Resasco, J. Catal. 271 (2010) 88–98.
[35] T.Q. Hoang, X. Zhu, T. Sooknoi, D.E. Resasco, R.G. Mallinson, J. Catal. 271 (2010)
201–208.
[36] H.S. Cerqueira, G. Caeiro, L. Costa, F.R. Ribeiro, J. Mol. Catal. A: Chem. 292 (2008)
1–13.
[1] Y. Liu, L. Wang, Chem. Eng. Process. 48 (2009) 1152–1156.
[2] A. Demirbas, Energy Convers. Manage. 50 (2009) 923–927.
[3] M.D. Bohon, B.A. Metzger, W.P. Linak, C.J. King, W.L. Roberts, Proc. Combust.
Inst. 33 (2011) 2717–2724.
[4] L. Gong, Y. Lu, Y. Ding, R. Lin, J. Li, W. Dong, T. Wang, W. Chen, Appl. Catal. A:
Gen. 390 (2010) 119–126.
[5] X. Guo, Y. Li, R. Shi, Q. Liu, E. Zhan, W. Shen, Appl. Catal. A: Gen. 371 (2009)
108–113.
[6] L. Ning, Y. Ding, W. Chen, L. Gong, R. Lin, Y. Lu, Q. Xin, Chin. J. Catal. 29 (2008)
212–214.
[7] M. Akiyama, S. Sato, R. Takahashi, K. Inui, M. Yokota, Appl. Catal. A: Gen. 371
(2009) 60–66.
[8] Z. Yuan, J. Wang, L. Wang, W. Xie, P. Chen, Z. Hou, X. Zheng, Bioresour. Technol.
101 (2010) 7088–7092.
[37] C. Bartholomew, R. Farrauto, Fundamentals of Industrial Catalytic Processes,
second ed., John Wiley & Sons, Inc., New Jersey, 2005, pp. 579–582.
[38] Inorganic crystal structure database (ICSD) collection code: 073877.
[39] M. Roussela, M. Boucharda, E. Bordes-Richarda, K. Karim, S. Al-Sayari, Catal.
Today 99 (2005) 77–87.
[40] P.M. Woi, R. Irmawati, Y.H. Taufiq-Yap, Malays. J. Anal. Sci. 11 (2007) 160–165.
[41] T. Katou, D. Vitry, W. Ueda, Chem. Lett. 32 (2003) 1028–1029.
[42] T.V. Andrushkevich, L.M. Plyasova, G.G. Kuznetsova, V.M. Bondareva, T.P. Gor-
shkova, I.P. Olenkova, N.I. Lebedeva, React. Kinet. Catal. Lett. 12 (1979) 463–467.
[43] S.M. Mathew, A.V. Biradar, S.B. Umbarkar, M.K. Dongare, Catal. Commun. 7
(2006) 394–398.
[9] H. Atia, U. Armbruster, A. Martin, J. Catal. 258 (2008) 71–82.
[10] S. Endres, P. Kampe, J. Kunert, A. Drochner, H. Vogel, Appl. Catal. A: Gen. 325
(2007) 237–243.
[11] J. Tichy, J. Machek, Catal. Lett. 15 (1992) 401–404.
[12] V.M. Bondareva, T.V. Andrushkevieh, L.M. Plyasova, R.I. Maksimovskaya, N.N.
Chumachenko, React. Kinet. Catal. Lett. 63 (1998) 201–206.
[13] H. Oka, K. Miyake, Y. Haranoa, T. Imoto, J. Appl. Chem. Biotechnol. 25 (1975)
663–670.
[14] T. Vanderspurt, Ann. N. Y. Acad. Sci. 333 (1980) 155–164.
[15] Y.T. Kim, K.-D. Jung, E.D. Park, Appl. Catal. B: Environ. 107 (2011) 177–187.
[16] E. Tsukuda, S. Sato, R. Takahashi, T. Sodesawa, Catal. Commun. 8 (2007)
1349–1353.
[44] Crystallography open database (COD) entry: 1011291.