Coke promoters improve acrolein selectivity in the gas-phase dehydration of glycerol to acrolein

Article abstract of DOI:10.1016/j.apcata.2016.04.022

Commercial processes partially oxidize propylene to acrolein but glycerol can replace the petroleum derived feeds as a bio-feedstock for this large volume specialty chemical. Metal oxides, particularly acid catalysts, dehydrate glycerol at elevated temperature. Dehydration of glycerol to acrolein is usually performed in traditional fixed-bed reactors. We tested phosphotungstic acid (HPW) loaded on titania catalysts (a weight fraction of 10, 20 and 30%) at 280 °C in a fluidized-bed reactor. With time-on-stream, the acrolein selectivity increased and the coke selectivity decreased. As much as 85% of the glycerol formed coke in the first hour and less than 20% acrolein. We developed a new method to spray the liquid feed directly on the catalyst bed and the reaction occurred in the gas phase. We also compared the performance of pure titania supports with respect to the acrolein selectivity. Acrolein selectivity was 13% higher on TiO₂ with an average pore diameter of 17.3 nm versus the TiO₂ support with a pore of 5.6 nm. Three times more coke formed on the small pore diameter TiO₂. We treated the catalyst with hydrogen rich coke promoters - tetralin and decalin - to passivate non-selective catalytic sites and to evidence hydrogen transfer reactions. This treatment increased acrolein selectivity from 10% to 30% in the first 30 min of reaction. As expected, they hydrogenated the intermediates to undesired compounds including acetone and propanaldehyde (propanal).

Full text of DOI:10.1016/j.apcata.2016.04.022

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Applied Catalysis A: General 522 (2016) 80–89
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Coke promoters improve acrolein selectivity in the gas-phase
dehydration of glycerol to acrolein
Marjan Dalila,∗, Mahesh Edakea, Camille Sudeaua, Jean-Luc Duboisb, Gregory S. Patiencea
a Department of Chemical Engineering, Polytechnique Montréal, C.P. 6079, Succ. CV, Montréal, H3C 3A7 Québec, Canada
b ARKEMA, 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:
Commercial processes partially oxidize propylene to acrolein but glycerol can replace the petroleum
derived feeds as a bio-feedstock for this large volume specialty chemical. Metal oxides, particularly
acid catalysts, dehydrate glycerol at elevated temperature. Dehydration of glycerol to acrolein is usually
performed in traditional fixed-bed reactors. We tested phosphotungstic acid (HPW) loaded on titania cat-
alysts (a weight fraction of 10, 20 and 30%) at 280 ◦C in a fluidized-bed reactor. With time-on-stream, the
acrolein selectivity increased and the coke selectivity decreased. As much as 85% of the glycerol formed
coke in the first hour and less than 20% acrolein. We developed a new method to spray the liquid feed
directly on the catalyst bed and the reaction occurred in the gas phase.
Received 15 February 2016
Received in revised form 20 April 2016
Accepted 21 April 2016
Available online 22 April 2016
Keywords:
Glycerol dehydration
Acrolein
We also compared the performance of pure titania supports with respect to the acrolein selectivity.
Acrolein selectivity was 13% higher on TiO2 with an average pore diameter of 17.3 nm versus the TiO2
support with a pore of 5.6 nm. Three times more coke formed on the small pore diameter TiO2. We treated
the catalyst with hydrogen rich coke promoters – tetralin and decalin – to passivate non-selective catalytic
sites and to evidence hydrogen transfer reactions. This treatment increased acrolein selectivity from 10%
to 30% in the first 30 min of reaction. As expected, they hydrogenated the intermediates to undesired
compounds including acetone and propanaldehyde (propanal).
Coke
HPW-TiO2
Fluidized-bed
Coke-promoter
© 2016 Elsevier B.V. All rights reserved.
1. Introduction
chemical industry and is a feedstock for acrylic acid and its esters,
glutaraldehyde, methionine, polyurethanes and polyester resins
Renewable feedstocks could replace petroleum to produce
energy and chemicals. Researchers and industries endeavour to
develop sustainable processes based on bio-resources. Adding
biodiesel to petro-diesel is a first step to substitute petroleum fuel
with bio-based fuels. Microalgae [1] and waste cooking oil [2] are
possible large volume sources for biodiesel synthesis.
Glycerol is the main co-product of biodiesel synthesis (a mass
fraction of 10%) and with the growing biodiesel industry the mar-
ket has been overwhelmed resulting in low prices. Due to its
multifunctional structure, glycerol is widely applied in various
industries such as, food, cosmetics, pharmaceuticals and plastics.
However, crude glycerol contains NGOM (non-glycerin organic
matter), water, methanol and traces of fatty acids and must be puri-
fied. Distillation is an extra step in refining and thereby increases
the cost of the process. Acrolein is versatile intermediate for the
[3–5]. The principal commercial process to synthesize acrolein
relies on propylene as a feedstock and multi-tubular fixed bed
reactors. Mixed-metal oxide containing Mo and Bi catalyses this
reaction at 250–400 ◦C [6,7].
Due to the availability of glycerol from biodiesel synthesis, dehy-
dration of glycerol to acrolein is a potential economic alternative
to propylene oxidation. The early studies of glycerol dehydration
were patented in 1930 by Kahlbum [8]. The yield of acrolein was
75% over lithium phosphate catalyst.
Synthesis of acrolein from glycerol has been studied in both
gas and liquid phase reactions. Generally, gas-phase dehydration
is preferred due to the environmental and operational issues of
liquid-phase reactions.
Several classes of catalysts such as zeolites, mixed-metal oxides
and heteropoly acids dehydrate glycerol to acrolein. Brønsted acids
are active and selective in acrolein synthesis, but deactivation
of the catalyst is the major drawback in this case. Dubois et al.
[9] suggested in situ regeneration of the catalyst by co-feeding
of molecular oxygen. They observed that molecular oxygen not
∗
Corresponding author.
E-mail address: marjan.dalil@polymtl.ca (M. Dalil).
http://dx.doi.org/10.1016/j.apcata.2016.04.022
0926-860X/© 2016 Elsevier B.V. All rights reserved.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
M. Dalil et al. / Applied Catalysis A: General 522 (2016) 80–89
81
only decreased the coking rate, but also improved the acrolein
selectivity. Additionally, the catalyst maintained its activity and
reduced the by-products such as acetol and phenol. Few studies
have focused on the effect of coke formation on catalyst perfor-
mance. Erfle et al. [10] analysed used vanadium based catalysts
by FTIR spectroscopy and concluded that coke forms on Brønsted
sites. This conclusion was confirmed by Suprun et al. [11] who eval-
uated carbon deposits on phosphate catalysts. They also observed
that higher reaction temperatures and small pore diameter of the
catalyst lead to further carbon formation. Pethan Rajan et al. [12]
characterized spent VPO by TPD-NH3 and FTIR-pyridine. The results
showed that the acidity and Brønsted acidic sites of the catalyst
decreased after 40 h time-on-stream. Consequently, the conversion
and selectivity of the catalyst decreased. The selectivity of acrolein
over coked WO3/TiO2 catalysts is higher than on fresh catalyst and
it remains active even after 6 h reaction time [13]. As a first step,
we evaluated the performance of the coked catalyst and observed
that partial regeneration of the catalyst improved the selectivity of
acrolein and less coke formed in the early stages of reaction.
Although gas-phase dehydration of glycerol to acrolein has been
studied since 1930 [8], most of the focus in this area is dedicated
to fixed-bed reactors. Fixed-bed reactors are associated with some
major drawbacks such as poor heat and mass transfer and hot spot
formation in case of exothermic reactions. Considering the rapid
deactivation of acid catalysts, catalyst regeneration is necessary in
dehydration of glycerol. A fluidized bed reactor is the most suit-
able choice for this case as it provides a homogeneous heat and
mass transfer and avoids local accumulation of the coke in catalytic
bed.
A TA-Q50 thermogravimetric analyzer measured the weight loss
as a function of temperature. A Platinel II thermocouple placed
2 mm above the sample pan monitored the temperature. For each
run, we loaded 20 mg of sample to a 10 ␮m aluminium crucible. The
resolution and accuracy of the balance were 0.1 ␮g and > 0.1%.
A Philipps Xpert diffractometer scanned samples at room tem-
perature (Cu anode, K = 0.15406 nm recorded the patterns at 50 kV
voltage and 40 mA current) and produced the diffraction pattern
(XRD) from which we deduced the HPW/TiO2 crystal phases. The
diffraction angle (2 theta range) varied between 20◦ and 90◦ at a
0.020◦ step size.
A
JEOL JSM-7600TFE Field Emission Scanning Electron
Microscopy recorded the microscopic images of the prepared
catalysts. The samples were mounted on an aluminium sample
holder containing a double graphite adhesive layer. We used
both LEI (lower secondary electron image) and LABE (low-angle
backscattered electron) detectors.
The samples prepared for FESEM analysis were also analysed at
the same time for EDX imaging. An OXFORD (X-Max) instrument
recorded the EDX images for elemental analysis of the samples.
For Fourier transform infrared spectroscopy (FTIR) a Spotlight
400 – PerkinElmer FTIR spectrometer was used. Prior to the anal-
ysis, the catalyst samples were treated with pyridine as the probe
molecule. We recorded the background spectra of samples before
analysing any samples. Then we treated the samples with pyridine
at room temperature and desorbed the excess pyridine at 100 ◦C
for 1 h. After cooling to room temperature, the instrument recorded
the spectra. The IR range of the analysis was 600–4000 cm−1 with
resolution of 16 cm−1
.
Here, we tested three loadings of tungsten oxide over large
pore and narrow pore titania supports. We focused on the effect
of pore diameter on acrolein, by-products and coke selectivity.
As preserving some coke on the surface of WO3/TiO2 improved
acrolein selectivity in the first 15 min of reaction [13], we devel-
oped a novel method to treat our catalysts with hydrogen rich coke
promoters such as tetralin and decalin to passivate non-selective
sites of the catalyst and increase acrolein selectivity and to evidence
the hydrogen transfer reactions taking place in the early phases of
the reaction.
A heat flow calorimeter (Setaram C80) connected to a volumet-
ric apparatus measured the acidity and basicity of the catalyst at
150 ◦C. The instrument was equipped with a Barocel capacitance
manometer to monitor pressure. The probe molecules (ammo-
nia for acidity and sulphur dioxide for basicity) were purified by
successive freeze–pump–thaw cycles. First, 100 mg sample was
pre-treated overnight at 250 ◦C under vacuum. To record the dif-
ferential heats of adsorption, small doses of the adsorbate were
introduced onto the catalyst and the pressure reached 66 Pa. The
samples outgassed for 30 min at the same temperature, and the pro-
cedure was repeated at 150 ◦C and 27 Pa. The difference between
the amounts adsorbed in the first and second steps represents the
irreversibly adsorbed quantity (Virr) of a respective gas from which
we estimate the number of acidic/basic sites.
2. Methodology
2.1. Catalyst preparation
2.3. Experimental
We prepared catalysts with a mass fraction of 10, 20 and 30%
phosphotungstic acid (H3PW12O40 or HPW) on large pore and nar-
row pore titania supports provided by ARKEMA (Hombikat 11010
– HKT-1 and Hombikat 11060 – HKT-2). A LA-950 (Horiba) laser
diffractometer measured the particle size distribution (PSD) of the
pure titania. The mean particle size of the supports was 95 ␮m and
106 ␮m for the HKT-1 and the HKT-2 samples, respectively. The
HPW (Nippon Inorganic Color & Chemical Co., Ltd.) dissolved in
10 ml of distilled water in a rotavapor flask (BUCHI R-210). We sub-
sequently loaded 10 g of titania to the flask. The rotavapor mixed
the slurry for 2 h at room temperature. We applied a 60 mbar vac-
uum and heated the flask to 60 ◦C. The samples then dried for 10 h
at 120 ◦C and calcined for 3 h at 500 ◦C [14].
2.3.1. Catalytic reaction set-up
HPW/TiO2 catalysts dehydrated glycerol in a quartz fluidized-
bed reactor. The reactor height was 52 mm and its inner diameter
was 8 mm. A 20 ␮m ceramic frit distributed the gas uniformly
across the reactor. Fluidized-beds are ideal reactors to regenerate
catalysts: the high solids mixing rate minimizes thermal and con-
centration gradients. In fixed bed reactors, coke may build up on
catalysts preferentially at the entrance of the reactor and in the cen-
tral region, which complicates interpreting reaction kinetics [15]. A
three-zone furnace heated the reactor and a thermocouple 100 mm
above the distributor monitored the bed temperature (Fig. 1).
A mixture of argon and oxygen fluidized the catalyst. To deter-
mine the minimum fluidization velocity (Umf), we recorded the
pressure drop at ascending and descending gas flow rates (Fig. 2a).
The Umf of all samples was 1.4 cm s−1. The total gas velocity was
held three times higher than this value for all the experiments.
The molecular oxygen (a mixture of 21 mol% O2/Ar) co-fed
with glycerol to maintain the catalyst life time and increase
acrolein selectivity while decreasing aromatic and hydrogenated
by-products [9,13].
2.2. Characterization techniques
An Autosorb-1 porosimeter (Quantachrome) measured the N2
adsorption/desorption isotherms at −196 ◦C. We assigned the sur-
face area based on the multi point BET (Brunauer, Emmett and
Teller) equation and the pore size and pore volume of the samples
on the BJH (Barrett–Joyner–Halenda) equation.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
82
M. Dalil et al. / Applied Catalysis A: General 522 (2016) 80–89
GC (offline)
Vent
The conversion of glycerol (Xgly.) as well as products selectivities
(Sp) were calculated as follows;
in
gly.
out
gly.
n
− n
TIC
Electro-Conductivity
Furnace
MFC
Xgly. (mol%) =
× 100
(1)
(2)
in
n
gly.
np
zp
HPLC
Sp (mol%) =
×
× 100
in
gly.
out
gly.
zgly.
(offline)
n
− n
Quench
in
gly.
out
gly.
where n
and n
are the molar flow rates of glycerol at reactor
MFC
entrance and exit. In Eq. (2), np is the molar stream of each product.
zp and zgly. represent the number of carbon atoms of the products
and glycerol.
Catalyst
3. Results and discussion
Liquid
3.1. Catalyst characterization
injector
Pump
Feed
MFC
3.1.1. N2 adsorption/desorption
The surface area of pure HKT-1 TiO2 was 71 m2 g−1, whereas
the average pore diameter and total pore volume were 17 nm and
0.22 cm3 g−1. The HKT-2 TiO2 surface area was 148 m2 g−1 with
the same pore volume. The average pore diameter, however, was
three times smaller than HKT-1 titania (5.6 nm). The tungsten load-
ing and calcination affected the physiochemical properties of TiO2:
both surface area and pore volume dropped as the tungsten loading
increased: tungsten covers the surface and partially fills the pores.
These properties for calcined samples were lower still. The surface
area of TiO2 dropped by 70% for HKT-1 and by 40% for HKT-2, when
loaded with 30% HPW and calcined (Table 1).
Sampling valve
Fig. 1. Diagram of experimental set-up for dehydration of glycerol. Glycerol was
injected above the grid as shown in our previous publication [13].
An HPLC pump injected the glycerol aqueous solution (28 wt%
glycerol in water, flow rate: 0.5 ml min−1
) directly into the
fluidized-bed of catalyst. It should be noted that the reaction con-
ditions were selected to evidence the transient phenomena during
the initial phase of catalytic reaction. The liquid feed entered the
reactor through a 1.6 mm nozzle without preheating. A flow of
argon assisted the liquid feed to enhance the quality of liquid
spray. The volumetric ratio of argon/glycerol solution was constant
(200) for all the experiments. To ensure that the droplets were uni-
form, we tested the injector with a series of gas/liquid ratios. Large
droplets agglomerate the catalyst and block the injector. The flow
of Ar in the injector should be high enough to spray the liquid feed
into the catalyst bed but too high flow rate attrits the catalyst.
The effluent gases from the reactor passed through two
quenches to trap all the condensable compounds. A Hanna elec-
trical conductivity meter monitored the product quench (first
quench); the conductivity is correlated to the concentration of ion-
izable solutes. The concentration of the liquid products trapped in
the quench increased with reaction time. Using the conductivity
meter we were able to monitor and follow the production rate by
reaction time (Fig. 2b).
Knowing that Keggin unit covers 1.2 nm2, we calculated the the-
oretical coverage of the pure TiO2 by HPW. On HKT-1 titania with a
surface area of 71 m2 g−1, for a monolayer coverage, 22 wt% loading
is needed. On the other hand, for HKT-2 titania with a surface area
of 148 m2 g−1, the loading of 37 wt% is required to create a mono-
layer. This explains the higher drop in the surface area of HKT-1
rather than HKT-2. We can also conclude that with HKT-2 one can
have a higher tungsten loading before a drop in surface area rather
with HKT-1; because HPW adsorbs on the surface until a monolayer
coverage is reached.
Pore volume also decreased by 80% and 40% for HKT-1 and
HKT-2, respectively but the pore diameter remained constant.
Calcination impacts pore volume of the catalysts which can be
correlated to the changes in phases.
3.1.2. X-ray diffraction
The X-ray diffraction patterns of fresh TiO2 (HKT-1, HKT-2) and
30% HPW/TiO2 suggested that the synthesized catalysts are crys-
talline (Fig. 3). All samples exhibited main relative peaks (101, 004,
A Bruker GC (equipped with Hyesep Q, Molsieve 5A and FFAP
columns) analysed the gas and liquid products (Fig. 1).
Fig. 2. (a) Minimum fluidization graph, (b) concentration of product mixture increased as reaction continued.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
M. Dalil et al. / Applied Catalysis A: General 522 (2016) 80–89
83
Table 1
Textural properties of various HPW/TiO2 catalysts.
Sample
SA (m2 g−1
)
ꢀp (cm3 g−1
)
dp (nm)
Not-calcined
Calcined
Not-calcined
Calcined
Not-calcined
Calcined
Pure TiO2 (HKT-1)
10% HPW/HKT-1
20% HPW/HKT-1
30% HPW/HKT-1
Pure TiO2 (HKT-2)
10% HPW/HKT-2
20% HPW/HKT-2
30% HPW/HKT-2
71
68
38
–
64
33
22
–
137
119
88
0.22
0.19
0.12
0.08
0.20
0.20
0.18
0.14
–
17
17
17
17
6
5
5
5
–
17
17
17
–
5
5
5
0.17
0.10
0.05
–
0.18
0.15
0.12
25
148
144
121
92
Fig. 3. XRD patterns for pure titania and 30% loading of HPW on (a) HKT-1 and (b) HKT-2 supports. *Crystalline phase of WO3.
200, 211, 204) of the anatase form of TiO2. We attributed the peaks
at 23.4◦, 37.6◦, 54.5◦ and 67.2◦ to crystalline tungsten oxide peaks.
Although calcination could decompose the heteropoly acid
(HPW), in the presence of water vapor HPW could be reconstructed
from dispersed metal oxides. However, the characteristic peak of
Keggin type phosphotungstic acid was not observed in the XRD
pattern of calcined catalysts.
monolayer. This explains the appearance of WO3 peaks for HKT-1
in the XRD pattern. However, the peak intensities of both supports
decreased when tungsten was loaded on titania [16,17].
3.1.3. Morphology-FESEM
The high magnification images of the uncalcined and calcined
HPW/TiO2 (FE-SEM) show that they are spheroidal (Fig. 4) and
their diameter varied between 80 ␮m and 120 ␮m. The surface was
rough. Calcining the catalyst improves the catalytic activity and
may distribute the HPW over the TiO2 since fewer micro-domains
of HPW are present on calcined samples (white dots on spheres in
Fig. 4). Catalyst assemblies were closely packed and marginal cracks
appeared on the grain surfaces of the calcined samples. The cracks
might be formed due to the high capillary pressure generated dur-
ing the impregnation process that manifest themselves when the
In Fig. 3a, the peak at 33.7◦ represents the crystalline phase of
WO3. On the other hand, 30%HPW/ HKT-2 catalyst did not show
characteristic peaks of W oxide (Fig. 3b), which indicates that HPW
clusters were well dispersed on the support with higher surface
area (Table 1) and a monolayer coverage would be reached at higher
tungsten loading. As mentioned before, for a monolayer coverage
over HKT-1 titania, 22 wt% loading of HPW is required. For HKT-2
titania, on the other hand, a loading of 37 wt% is needed to create a
Fig. 4. FE-SEM images of (a) uncalcined and (b) calcined 30% HPW on HKT-1 supports.