Gas phase glycerol dehydration over H-ZSM-5 zeolite modified by alkaline treatment with Na2CO3

Article abstract of DOI:10.1016/j.jcat.2018.07.036

Modified H-ZSM-5 zeolite catalysts were studied for the dehydration of glycerol to acrolein in gas phase. The modifications were made by alkaline treatment using Na₂CO₃ solutions in different concentrations, followed by consecutive i

Full text of DOI:10.1016/j.jcat.2018.07.036

Journal of Catalysis 366 (2018) 16–27
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Gas phase glycerol dehydration over H-ZSM-5 zeolite modified by
alkaline treatment with Na2CO3
⇑
C.D. Lago, H.P. Decolatti, Lucas G. Tonutti, B.O. Dalla Costa, C.A. Querini
Instituto de Investigaciones en Catálisis y Petroquímica (INCAPE) – FIQ – UNL – CONICET, RN168, Km 0, (3000) Santa Fe, Argentina
a r t i c l e i n f o
a b s t r a c t
Article history:
Modified H-ZSM-5 zeolite catalysts were studied for the dehydration of glycerol to acrolein in gas phase.
The modifications were made by alkaline treatment using Na2CO3 solutions in different concentrations,
followed by consecutive ion exchanges with NH4NO3 solution in order to recover the acidic form of the
materials. The mild conditions selected (alkali concentrations, time and temperature) allowed to preserve
the crystallinity and framework characteristic in the zeolite, selectively removing silicon while maintain-
ing practically all the aluminum content. This process has important advantages over the alkaline treat-
ment with NaOH (at an equivalent pH), such as high selectivity to remove silicon instead of aluminum,
and in addition the yield of synthesized material respect to the initial mass is markedly higher (80% vs
50%). This treatment with Na2CO3 solutions leads to multiple positive effects as: increments in the meso-
porosity and external surface area; and changes in the nature and accessibility to the acid active sites. The
catalytic activity and selectivity were greatly improved. Further, the deactivation by coke deposition was
attenuated.
Received 11 April 2018
Revised 24 July 2018
Accepted 28 July 2018
Keywords:
H-ZSM-5
Alkaline treatment
Sodium carbonate
Glycerol dehydration
Ó 2018 Elsevier Inc. All rights reserved.
1. Introduction
conversion of glycerol into acrolein offers a cost-effective and sus-
tainable alternative route to replace the present production tech-
The particular rise in demand of clean fuels and chemicals
obtained from renewable sources has led to an intensive research
in order to develop industrial applications of biomass of all types
[1,2]. In this sense, one of the most important processes is biodiesel
production by transesterification of vegetable oils or animal fats as
raw materials. Generally, 10 wt% of glycerol is generated as co-
product during biodiesel production, and due to the increase of
the biodiesel manufacturing in the last decade, a glycerol oversup-
ply in the global market has occurred, converting this polyol into a
very promising low-cost feedstock for the synthesis of value-added
specialty and fine chemicals [3]. The price of glycerol has decreased
from US$ 0.43 kgꢀ1 in 2003 to US$ 0.18 kgꢀ1 in 2010 for pure glyc-
erol, and only US$ 0.02 kgꢀ1 for crude glycerol [4]. This molecule
with three hydroxyl groups can undergo dehydration, oxidation,
reforming, hydrogenolysis, etherification and esterification
reactions to produce a lot of chemical commodities [5,6]. Dehydra-
tion of glycerol produces two important bulk chemicals:
1-hydroxyacetone (acetol) and propenal (acrolein). Acrolein
production is currently carried out by partial oxidation of propy-
lene, using mixed metal oxides as catalysts [7]. Therefore, catalytic
nology. Acrolein is used for the production of acrylic acid, acid
esters, super absorbers polymers, detergents, fragrances, dyes
and agrochemicals like methionine, among others products [8,9].
The catalytic conversion of glycerol into acrolein by dehydration
can be made in either the gaseous or liquid phase. However, in
the liquid phase reaction, the glycerol conversion was limited to
15–20%, to avoid excessive loss of selectivity [10]. Manipulation
of conversion and selectivity to acrolein is easier in the gas phase.
Several solid acid catalysts such as sulfates, phosphates, and zeo-
lites have been used either in gaseous or liquid phase reactions
[11,12]. Other solid acid catalysts as metal oxides [13–15] and
heteropolyacids (HPAs) supported on metal oxides [13,16–18]
has also been studied in gas phase reaction. The catalytic activity
is a function of the textural and acidic properties of the solid.
The pore size is a key factor in order to obtain good activity and
selectivity. Catalysts with small pores showed lower activity and
stability than catalysts with mesopores ranged between 6 and
10 nm [16]. The total amount of acids sites, as well as its strength
and nature have a major influence in the performance of the cata-
lysts. In terms of acidity, it has been reasonable established that
Brønsted acid sites catalyzes acrolein production [19,20], whereas
Lewis acid sites increases the selectivity to hydroxyacetone
[21,22]. However, some studies in the literature have proven that
⇑
Corresponding author.
E-mail address: querini@fiq.unl.edu.ar (C.A. Querini).
https://doi.org/10.1016/j.jcat.2018.07.036
0021-9517/Ó 2018 Elsevier Inc. All rights reserved.

C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
17
Brønsted acid sites can also catalyze hydroxyacetone production,
what occurs when glycerol interacts via the hydroxyl groups
attached to primary carbons. Foo et al. [23] demonstrated, using
FTIR spectroscopy, a cooperative role of Brønsted and Lewis acid
sites in the glycerol dehydration to hydroxyacetone over niobia
catalysts. These authors proposed several steps for this reaction.
In a first step glycerol is adsorbed on a Lewis acid site, and in a sec-
ond step it is transformed into 2-propene-1,2-diol, which further
tautomerizes to hydroxyacetone The authors concluded that the
second step involved the participation of an adjacent Brønsted acid
site to generate the 2-propene-1,2-diol.
although the changes at molecular or atomic scale were discarded
by XRD results. An alternative to prevent this problem is to use a
weaker alkali, like Na2CO3 [35].
In this paper we present results obtained after modification of
an H-ZSM-5 zeolite (Si/Al = 40) by alkaline treatment with Na2CO3
aqueous solutions in different concentrations, followed by ionic
exchange with NH4NO3 to obtain acid catalysts. The textural prop-
erties and the acidity of the materials were studied employing
numerous characterization techniques. The activity of the solids
was tested in the glycerol dehydration reaction to obtain acrolein.
The zeolites offer several advantages. These solids are environ-
mentally harmless, noncorrosive, thermal and chemical stable and
possess a high versatility among other things. Moreover, the
shape-selective properties of zeolites related to the presence of
an ordered microporous network can restrict the formation of
undesired products by control of reactant or product diffusion in
the pores as well as the volume available for transition states. Nev-
ertheless, the pure microporous nature of zeolites frequently
involves transport limitations, particularly when bulky molecules
are involved, which adversely affect catalytic performance. ZSM-
5 zeolite is known as a solid acid catalyst for efficient conversion
of fossil and biomass resources to fuels and chemicals [24,25]. In
addition, ZSM-5 has been observed to exhibit excellent catalytic
properties in aromatics compounds synthesis, mainly due to its
shape-selectivity characteristic. However, the microporous chan-
nels present in its structure may in some cases be a major limita-
tion, because mass transport is slow through these narrow
channels, especially when reaction is significantly faster than diffu-
sion in the micropores [26,27]. Scaling down of zeolite crystals
from micrometer to nanometer scale and developing mesoporous
zeolite catalysts have been recognized as important factors in
improving catalytic activity. Emerging mesoporous materials how-
ever generally do not comply with most practical requirements as
a result of limited thermal stability and poor acidic properties.
Consequently, new synthesis procedures for preparation of small
zeolitic crystals [28] or post-treatment procedures to create
extra-porosity [29] are increasingly investigated. Conventional
steaming and acid leaching methods or the less-known treatments
in alkaline media have been applied to modify various properties of
zeolites. The last method removes preferentially Si from the zeolite
framework (desilication) [30], while the former ones lead to dealu-
mination. Desilication was firstly applied to study chemical
changes of MFI crystals upon contact with NaOH solutions, and this
treatment has shown to induce a significant mesoporosity in MFI-
type zeolites [31,32]. Furthermore, this treatment allowed retain-
ing the crystal structure and the acidic properties of the solids.
Desilicated zeolites have been investigated in several reactions,
including cumene cracking, methanol to propylene, methanol to
gasoline, hydroxylation of benzene to phenol, methane aromatiza-
tion, and hexene conversion [21,33]. In most cases, improved
activity, stability, and selectivity have been reported. In general,
the observed improvement has been ascribed to enhanced
diffusion due to the generation of mesoporous channels. The meso-
porous structure is an advantage for the diffusion of reactants
because the external surface area of zeolite is improved and the
catalytically active acid sites become more easily accessible to
reactants.
2. Experimental
2.1. Catalyst preparation
A ZSM-5 zeolite with a nominal Si/Al ratio of 40 (CBV-8020),
supplied by Valfor in an H-form, was used in this study. The cal-
cined sample of this catalyst is labeled in this work as H-ZSM-5.
The alkali treatments were carried out using aqueous solutions of
0.2, 0.3, 0.4, 0.5 and 0.7 mol Lꢀ1 Na2CO3, employing 30 ml of solu-
tion per gram of calcined ZSM-5 zeolite. The treatment was carried
out with vigorous stirring at 65 °C during 0.5 h, in a flask with a
reflux condenser and a water bath, following the procedure previ-
ously established for these materials [36]. The zeolite suspension
was then cooled down immediately using an ice bath. The remain-
ing product was filtered, carefully washed until neutral pH and
finally dried at 100 °C overnight. The alkali-treated samples were
obtained in a Na-exchanged form, and it was necessary to convert
them into the H-form in order to recover the acidity. After the
alkali treatment, two consecutive ion exchanges with 0.5 mol Lꢀ1
NH4NO3 aqueous solutions during 2.5 h in reflux were carried
out to each sample. Then, the samples were calcined in air at
550 °C for 2 h. The modified zeolite catalysts were named H-AT
(M), being M = 0.2, 0.3, 0.4, 0.5 and 0.7 according to the concentra-
tion of Na2CO3 used in the alkali-treatment solution.
2.2. Catalyst characterization
Nitrogen adsorption–desorption isotherms were recorded at
liquid-nitrogen temperature (ꢀ196 °C) and relative pressure (P/
Po) interval between 6 ꢁ 10ꢀ7 and 0.998 in a Quantachrome equip-
ment. Before the adsorption, samples were evacuated at 250 °C
during 3 h under vacuum of 1 ꢁ 10ꢀ5 Pa. The BET model [37] was
used in the relative pressure range 0.01–0.10 to calculate the total
surface area, while the micropores volume and external surface
area were derived from the t-plot, according to Lippens and de Boer
[38]. The pore-size distribution was calculated employing the
Broekhoff and de Boer (BdB) method [39], applied to the adsorp-
tion branch of the isotherm.
X-ray diffraction (XRD) patterns were measured in a Shimadzu
XD-D1 instrument with a monochromator, CuK radiation and a
a
scanning rate of 2° minꢀ1 in the range 2h = 0°ꢀ55°.
Si and Al concentrations in the filtrate obtained upon alkaline
treatment were determined by inductively coupled plasma atomic
emission spectroscopy (ICP-OES). Measurements were performed
in a Perkin Elmer Optima 2100 DV. Also the Al and Si content in
the solids were measured by Energy-dispersive X-ray fluorescence
(XRF), using a Shimadzu equipment, model EDX-720, and working
in the energy dispersion mode.
The versatility of this alkaline treatment opens new avenues to
improve diffusion characteristics in zeolite-catalyzed applications.
Most of the studies include the use of NaOH in different concentra-
tions, varying the temperature and the time of treatment. If the
treatment is very aggressive, although mesoporosity is generated
in the structure, changes in morphology can also be produced, with
a partial collapse of the zeolite grain. M. Ogura et al. [34] observed
changes at micrometric scale combining SEM and TEM techniques,
The morphology of H-ZSM-5 catalyst before and after the alka-
line treatment was studied by Scanning electron microscopy (SEM)
samples at 15 KV using an electron microscope (Phenom PROX). In
order to ensure that the data collected was representative of the
whole sample, scans were made at more than one location, indicat-
ing that the zeolite particles are homogeneous.

18
C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
27Al-MAS-NMR was used to analyze the Al coordination states
the experiment conditions (400 °C, 6% O2, 40 ml minꢀ1 H2 flow), a
100% conversion of COx was reached and in this way the sensitivity
of the technique was greatly improved. The analysis was carried
out with a heating rate of 12 °C minꢀ1 starting from 20 °C. Equip-
ment calibrations were made using pulses of CO2 diluted in He.
on the zeolite samples. Spectra were recorded using a Bruker
Avance AMX-500 spectrometer. Prior to measurements, all zeolites
were stored for 48 h in a desiccator over a saturated Ca(NO3)2
aqueous solution in order to achieve full pore hydration, thus lead-
ing to improved NMR spectral resolution due to minimization of
distortions in the Al tetrahedral [40]. After hydration, samples
were packed at 42 mbar H2O into ZrO2 rotors, which were spun
at 10 kHz. Al(NO3).9H2O was used as reference. For measuring
the 1D spectrum, an excitation pulse with power level of 7 dB
2.3. Catalyst performance tests
The dehydration of glycerol was conducted at atmospheric
pressure in a continuous fixed-bed stainless steel reactor, using
0.4 g of catalyst (particle size 40–80 mesh) without diluents. Prior
to the catalytic tests, catalysts were pretreated at 300 °C in N2 flow
for 1 h to remove adsorbed water. The reaction feed was an aque-
ous solution containing glycerol (20 wt%). It was evaporated in a
preheater at 260 °C, with 50 ml minꢀ1 of N2 as carrier gas. It is well
known that diffusion is driven by the concentration difference
between internal and external space of catalyst [41]. Y. Gu et al.
studied the effect of carrier gas, concluding that the glycerol con-
version and acrolein selectivities were improved when using car-
rier [42]. The authors stated that the introduction of carrier gas
could enlarge the concentration difference by supplying glycerol
(and removing products) continuously to (and from) outer space
of catalyst, which could stimulate the diffusion of glycerol into
channels and products out of channels of H-zeolites. Besides, the
dilution effect of carrier gas decreases the partial pressures of
reactants and products simultaneously, which reduce the rates of
coke formation and catalyst deactivation. In addition, a lower
glycerol partial pressure could also reduce the possibility of inter-
molecular dehydration, which would kinetically affect the catalyst
performance.
Throughout this work, the space velocity used (WHSV = 0.75
hꢀ1) represents the mass of glycerol per gram of catalyst, per hour.
The reactor was placed in an electrically heated furnace, and the
reaction temperature was set at 275 °C. After the reaction, the car-
rier flow was maintained for 1 h, in order to purge the system.
Reaction products together with the unconverted glycerol were
collected in a cold trap maintained at 5 °C at 30 min intervals.
The analysis of the products was carried out offline in an Agilent
7820A gas chromatograph (GC) equipped with a HP-FFAP capillary
and a length of 0.7 ls was applied. Relaxation time was set to
250 ms, and 2400 scans were recorded.
The amount and strength of acid sites were evaluated by pyri-
dine temperature-programmed desorption (TPD). Approximately
10 mg of catalyst, sandwiched by quartz wool, was loaded in a
quartz tube. The solids were pretreated in-situ in N2 flow (30 ml
minꢀ1) at 300 °C during 1 h. After cooling down to room tempera-
ture, three consecutive saturation steps with pure pyridine were
performed. Then, N2 was flowed and the temperature was
increased up to 150 °C, until no physically adsorbed pyridine was
detected. The TPD experiments were carried out heating at 12 °C
minꢀ1 in N2 flow from 150 °C to 800 °C. Pyridine coming out of
the sample cell passed through a methanation reactor. A H2 stream
was also fed to this reactor, quantitatively converting pyridine to
CH4 using a Ni catalyst. The CH4 was continuously measured by a
FID detector.
Pyridine with a diameter of 0.68 nm is at the limit for adsorp-
tion in the 10-ring pores of ZSM-5 with a diameter of 5.5 nm. In
this case, TPD curves not only depend on the properties of the acid
sites, but also on steric limitations of pyridine inside the pores. For
this reason, also experiments with adsorption of ammonia were
performed in order to analyse the acidity of the catalysts. The sam-
ples were pretreated inside the TPD reactor in Ar flow (60 ml
minꢀ1) at 300 °C for 1 h, followed by adsorption of ammonia
(diluted 1% NH3 in He) at 100 °C during 45 min. The removal of
physically adsorbed ammonia was carried out by purging the sam-
ple at 100 °C for 2 h under a flow rate of 60 ml minꢀ1 of Ar. Then,
TPD of ammonia was performed heating the sample from 100 °C
up to 800 °C, at a rate of 12 °C minꢀ1, using the same flow rate of
Ar. The amount of desorbed ammonia of each sample was deter-
mined by using a mass spectrometer Balzers Omnistar.
The Lewis and Brønsted acidity of the samples was studied by
pyridine adsorption followed by infrared spectroscopy (FTIR).
Spectral measurements of pyridine adsorption on the samples
were performed on a JASCO FT-IR 5300 spectrometer equipped
with a DTGS detector. The range and resolution of acquisition were
4600–400 and 4 cmꢀ1 respectively. A self-supporting wafer for
each sample (ꢂ20 mg and 13 mm of diameter) was prepared,
placed in a thermostatized cell with CaF2 windows connected to
a vacuum line, and evacuated for 8 h at 400 °C. The background
spectrum was recorded first after cooling the sample to room tem-
perature. Afterwards, the solid wafer was exposed to pyridine
vapors (Sintorgan, 99% purity) until the system was saturated to
46 mmHg at room temperature; the contact time at this pressure
was 12 h. The IR spectrum for each sample was obtained after pyr-
idine desorption by evacuation for 1 h at 100, 200, 300 and 400 °C.
All the spectra were recorded at room temperature before and after
pyridine adsorption and desorption at each temperature. The dif-
ference spectrum was finally obtained by subtracting the back-
ground spectrum previously recorded.
column (30 m ꢁ 0.32 mm ꢁ 0.25
lm) and a flame ionization detec-
tor (FID). The following products have been identified and quanti-
fied: acrolein, acetaldehyde, propanal, acetol, and methanol. Others
compounds in low concentrations have not been identified. The
following equations were used to calculate glycerol conversion,
product yield, selectivity and the relative deactivation:
moles of glycerol reacted
moles of glycerol fed
Glycerol conversion ð%Þ ¼
ꢁ 100
ð1Þ
ð2Þ
ð3Þ
ð4Þ
moles of product
ꢁ moles of glycerol fed
Product yield ð%Þ ¼
ꢁ 100
m
product yield
glycerol conversion
Product selectivity ð%Þ ¼
Xꢃ ꢀ X
ꢁ 100
D ð%Þ ¼
ꢁ 100
Xꢃ
The coefficient
m represents the number of moles of a given pro-
duct obtained per mole of glycerol, being 1.5 for acetaldehyde and
1 for all the other products. In Eq. (4), D(%) is the percentage of
deactivation, calculated using the initial (X°) and final (X) glycerol
conversion.
The coke in the catalysts after reaction was characterized by
temperature-programmed oxidation (TPO). The gases coming out
of the sample cell passed through a methanation reactor, as in
the pyridine TPD experiments, but in this case the CO and CO2
resulting from coke oxidation were converted to CH4. Again, this
compound was continuously measured using a FID detector. Under
Also, experiments varying the WHSV (1.5 hꢀ1 and 3 hꢀ1) or the
temperature (300 °C) were carried out, in order to study the deac-
tivation phenomena on the catalysts and the products distribution.

C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
19
Van Mao et al. [35] reported a 92% of crystallinity in the modified
zeolite.
3. Results and discussion
3.1. XRD analysis
3.2. SEM analysis
Fig. 1 shows the XRD patterns for the calcined H-ZSM-5 zeolite
and for the different catalysts obtained after the alkaline treat-
ments. The changes in crystallinity of the desilicated zeolite were
calculated based on the relative intensities of the reflection peak
at 23° (2h), taking the crystallinity of the parent sample to be
100%. No appreciable changes were observed in the treated sam-
ples, which showed crystallinity values ranged between 98% and
100%. The diffraction peaks assigned to MFI type materials did
not change in intensity and also the lattice parameters were the
same after alkali-treatment. Therefore, X-ray diffraction analysis
confirmed the preservation of the long-range crystal ordering in
the samples after the alkali treatment.
SEM analysis showed that the zeolite particles size is in the
range of 0.5–2 lm. All catalysts consist of similar particles associ-
ated with small crystallites and large agglomerated crystals, as
other researchers previously observed [34]. After the alkali treat-
ment, no visually detectable changes in the general morphology
of the particles were observed (see Supplementary Data, Fig. S1).
The particles of the alkaline-treated sample have a slightly rougher
surface.
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jcat.2018.07.036.
Ogura et al. [34] observed that the peaks at 23.07° and 23.25°,
respectively assigned to [0 5 1] and [ꢀ5, 0, 1] of MFI, were dis-
tinctly splitted after alkali-treatment with NaOH, indicating that
poor crystallinity parts of ZSM-5 were selectively dissolved. How-
ever, this change was not observed in this work.
It is important to mention that the desilication by alkali treat-
ment with Na2CO3 produces lower modifications in the zeolite
framework in comparison with that observed in a previous work
when NaOH was employed as the alkali [43] and also as was
reported by others authors [44]. For example, in the case of treat-
ing an H-ZSM-5 (Si/Al = 15) with a 0.8 mol Lꢀ1 Na2CO3 solution, Le
3.3. Textural properties
Fig. 2A shows the N2 adsorption-desorption isotherm of the
original H-ZSM-5 (curve a). It is a type I isotherm with an initial
slope due to the adsorption in the micropores of the zeolite, and
then a plateau at high relative pressure. The presence of a type
H4 hysteresis loop was observed, associated to slit-shaped meso-
pores which coexist with the micropores [35,44,45]. These meso-
pores are not cylindrical, and represents an external surface area
of 150 m2 gꢀ1 (Table 1).
After the alkali treatment, the isotherms (curves b–f) practically
did not change in the relative pressure region below P/Po = 0.3,
indicating the preservation of the microporous structure of the
material. The micropores volume decreased slightly in comparison
to the non-treated zeolite, but their values hardly changes among
the different alkali concentrations (Table 1).
2.00
1.75
1.50
At higher relative pressure, the amount of adsorbed N2
increased in the alkaline treated samples, and the hysteresis loop
was modified, indicating in this case the generation of mesoporos-
ity in the solids [46]. This phenomenon is more appreciable when
the intensity of the alkali treatment was increased. To facilitate the
comparison the isotherm of the non-treated H-ZSM-5 zeolite was
copied (dashed line) over imposing it with the isotherms of the
modified catalysts (curves b and f) in Fig. 2A. It can be observed
that the mesopores volume and external surface area were
improved in all cases.
1.25
1.00
0.75
0.50
0.25
0.00
f
e
d
c
b
a
5
10 15 20 25 30 35 40 45 50 55
In addition, some enlargement of the existing mesopores in the
parent material was also observed through the broadening of the
hysteresis loop given by the N2 adsorption desorption isotherms
as indicated in Fig. 2A.
2
Fig. 1. XRD patterns of original zeolite (a) H-ZSM-5 and after alkaline treatment (b)
H-AT(0.2), (c) H-AT(0.3), (d) H-AT(0.4), (e) H-AT(0.5) and (f) H-AT(0.7).
B
A
f
e
d
c
b
a
f
e
d
c
b
a
0
40
80
120
160
200
0.0
0.2
0.4
0.6
0.8
1.0
Relative Pressure (P/P0)
Pore Width (Å)
Fig. 2. (A) N2 adsorption-desorption isotherms (vertically shifted for better observations) and (B) Pore size distribution; for samples (a) H-ZSM-5 (b) H-AT(0.2), (c) H-AT(0.3),
(d) H-AT(0.4), (e) H-AT(0.5) and (f) H-AT(0.7).

20
C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
Table 1
Textural properties, XRF and ICP results.
Sample
SBET (m2 gꢀ1
)
SEXT (m2 gꢀ1
)
Vmicrop. (cm3 gꢀ1
)
Vmesop. (cm3 gꢀ1
)
Si/Ala
%Alb extracted
%Sib extracted
H-ZSM-5
H-AT(0.2)
H-AT(0.3)
H-AT(0.4)
H-AT(0.5)
H-AT(0.7)
406
384
392
421
417
416
150
163
168
195
188
183
0.130
0.113
0.114
0.116
0.117
0.119
0.137
0.159
0.168
0.180
0.185
0.190
41.3
36.0
34.6
34.3
33.6
32.4
–
–
0.20
0.17
0.15
0.13
0.16
5.7
6.9
6.7
8.2
8.4
a
Obtained by XRF analyses of the solids.
Obtained by ICP analyses of the filtrate.
b
Fig. 2B shows the pore size distribution obtained after applying
the BdB method to the N2 adsorption isotherms. For the H-ZSM-5
zeolite, the presence of mesopores with a narrow size distribution
around 35 Å was detected. After the alkali treatment, a small
decrease in the amount of these mesopores was observed, proba-
bly because some of them increased its size. However, the major
change in the pore size distribution is the generation of new pores
with a broad size distribution around 78 Å. An increase in the con-
centration of Na2CO3 for the alkali treatment produced more quan-
tity of these mesopores.
f
e
d
c
b
a
In all samples, an increase of the external surface due to the
mesoporosity generation was observed, with a slowly decrease
of the micropores volume (Table 1). However, it is interesting
to analyze that when a low concentration of the alkali was
employed, the total BET surface area was reduced in comparison
with the non-treated zeolite. A reason for this could be the pore
blockage by deposition of different species. For example, silicon
species dissolved from the zeolite framework precipitated easily
on the surface of the crystals, forming an amorphous silica layer
[47]. This was observed by other authors when using low NaOH
concentrations and low contact time that correspond to mild
treatments conditions [34]. However, in our case, the presence
of amorphous silica in the prepared materials is rule out accord-
ing to the XRD results and the crystallinity values obtained. For
the samples treated with alkali concentrations of 0.4 mol Lꢀ1 or
greater, the increase of the external surface was more apprecia-
ble, and also the BET surface was improved. In this case, the
higher concentration of COꢀ3 2 leads to a more extensive hydrolysis
of these species and the formation of OHꢀ1 ions, which are cap-
able of removing silicon, preventing the precipitation and block-
age of the pores.
1700
1600
1500
1400
Wavenumber (cm-1)
Fig. 3. FTIR spectra of the samples (a) H-ZSM-5, (b) H-AT(0.2), (c) H-AT(0.3), (d) H-
AT(0.4), (e) H-AT(0.5) and (f) H-AT(0.7).
3.5. Si and Al removal during the alkaline treatments
The effectiveness of the alkali treatment for selectively remov-
ing Si and maintaining the Al content was analyzed by XRF mea-
surements in the solids, and ICP determinations in the filtrate.
Table 1 resume the Si/Al ratio which was reduced from 41.3 in
the non-treated zeolite H-ZSM-5 to 36 in the H-AT(0.2). Then, for
the other samples, only a slight decrease of this parameter with
the alkali concentration was observed, being the percentage of
reduction of the Si/Al ratio between 16 and 21%.
When the alkali treatment was conducted under the same con-
ditions to an H-ZSM-5 with lower silicon content (Si/Al = 15), there
were no appreciable changes in the value indicating that the treat-
ment with Na2CO3 was not effective for this Si/Al ratio (result not
shown). In a previous work, it was shown that when the treatment
for this zeolite was conducted using NaOH 0.2 mol Lꢀ1, a decrease
in the Si/Al ratio of 8.5% was observed [43]. Le Van Mao et al. [35]
reported a reduction of 48% in a similar zeolite, but the authors
used more aggressive treatment conditions: three consecutive
treatments with Na2CO3 0.8 mol Lꢀ1, higher temperature (80 °C)
and more contact time (4 h). After comparing the results obtained
for an H-ZSM-5 (Si/Al = 15) with those observed for a Na-Y (Si/Al =
2.4) and a Na-X (Si/Al = 1.2), they concluded that with higher Si/Al
ratio in the material, the alkaline treatment was more effective.
Particularly for H-ZSM-5 zeolites, the optimum Si/Al ratio to
carry out the treatment is ranged between 25 and 50. When the
Si/Al ratio is lower than 25, the desilication is not effective due
to a stabilizing effect of the aluminum in the framework [49,50].
The ICP results obtained in the analyses of the Si and Al contents
in the filtrate are also reported in Table 1. It can be noticed that the
amount of extracted aluminum from the zeolite was very low,
around 0.2% of the initial content. Instead, the percentage of silicon
removed was 8.4% for the sample H-AT(0.7). This is a clear evi-
dence of the high selectivity of the treatment with sodium carbon-
ate regarding the removal of Si compared to Al.
There is another possible reason for the reduction of the BET
area when Na2CO3 is used as alkali, related to the deposition of
COꢀ3 2 species which are not removed by calcination and could be
responsible of a possible total or partial pore blockage, as discussed
in the next section.
3.4. FTIR analysis
Fig. 3 shows the FTIR results in the 800–1800 cmꢀ1 region for
the original H-ZSM-5 zeolite and for the samples after the alkaline
treatment. New signals associated to carbonate species appear in
the modified catalysts (curves b–f). In these spectra the signal at
1560 cmꢀ1 corresponds to the COꢀ3 2 asymmetric stretching (from
pure Na2CO3), while the doublet at 1490–1440 cmꢀ1 is related to
the COꢀ3 2 asymmetric stretching also, but when this specie is
attached to Al+3 atoms (Al(OH)3 + Na2CO3). No appreciable changes
in the intensity of the signals were observed, therefore according to
this technique the amount of deposited species would be similar
for all the samples. Su et al. [48] observed these signals, when they
studied carbonate species adsorbed on aluminum and iron oxides,
employing in situ infrared experiments.

C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
21
3.6. 27Al-MAS-NMR results
ple H-AT(0.5) (Table 1). Accordingly, at equivalent pH values, the
NaOH removes only 10% of the Si extracted with Na2CO3. Further-
more, the NaOH solution removes 69% more Al from the frame-
work, which affects negatively the final acidity of the materials.
This is a very important difference between these two bases. In
summary, using the alkaline treatment with Na2CO3, the extraction
of silicon is maximized with a minimum loss of aluminum.
On the other hand, it is very important to remark, that the
recovery of the starting material after the alkaline treatment with
Na2CO3 was 80%, whereas only 50% of the initial mass was recov-
ered when the treatment was carried out with NaOH. Conse-
quently, due to a better desilication performance, and a lower
loss of the starting material (i.e. better yield), the treatment with
Na2CO3 is advantageous.
The normalized 27Al-MAS-NMR spectra for all the samples are
shown in Fig. 4. The signal at 55 ppm indicates the presence of
tetrahedral coordinated framework Al (FAL). The peak shape is nar-
row and symmetric and the intensities were comparable for all
samples. No signals of dealumination were detected after the alka-
line treatment. Also the signal at ꢀ0.55 ppm was observed, which
corresponds to the presence of octahedral non framework Al
(EFAL) [51]. Patterns recorded for the non-treated and alkaline-
treated H-ZSM-5 zeolites have shown to be very similar, with a
negligible contribution from non-framework aluminum species.
This suggests that the aluminum was predominantly tetrahedrally
coordinated in the zeolite framework upon the alkaline treatment
with Na2CO3. Similar results were reported by others authors when
performed 27Al-MAS-NMR experiments to H-ZSM-5 zeolite modi-
fied by alkaline treatment with NaOH [36].
3.8. Acidic properties
It can be concluded that the conditions adopted in this work for
the alkali treatments were quite moderate, the removal of Si was
highly selective, the generation of mesoporosity was appreciable,
and the crystalline structure of the zeolite was not altered.
Information about the superficial acidity of the solids was
obtained by Pyridine and Ammonia TPD, and by FTIR experiments.
Fig. 5A shows the Pyridine TPD profiles. The area under the
curves is related to the total amount of acid sites, whereas the
peaks desorption temperatures indicate the strength of these acid
sites. The original zeolite H-ZSM-5 (curve a) showed a profile with
two peaks, one at low temperature (240 °C) corresponding to weak
acid sites and another centered at 610 °C associated to strong acid
sites. A total acidity of 0.37 mmol gꢀ1 was detected for this mate-
rial, being 90% strong acid sites.
After the alkali treatment, some modifications were observed in
the acidic properties. First, in all cases, the amount of weak acid
sites decreased. Only the sample H-AT(0.4) showed sites with
weak acidity, while very low amount of weak acid sites were
detected for the other samples.
3.7. Anion effect: comparison of NaOH and N2CO3 during the alkaline
treatment
In the case of the Na2CO3 solutions used in this work to perform
the alkaline treatments, the pH is close to 12 in all cases, due to the
buffer characteristics of these solutions.
In order to evaluate the effect of the anion present in the alkali
(OHꢀ or COꢀ3 2), the H-ZSM-5 zeolite was treated with a 0.01 mol
Lꢀ1 NaOH solution, which has the same pH than the Na2CO3 solu-
tion (pH = 12). The results were compared with those obtained for
the sample H-AT(0.5), because the 0.5 mol Lꢀ1 Na2CO3 has the pH
closer to 12 (pH = 12.009). The percentage of Si and Al extracted
when using NaOH were 0.83% and 0.35% respectively. Instead, by
using Na2CO3 the values were 8.2% of Si and 0.13% of Al in the sam-
The position of the peak at higher temperature was slightly
moved to lower temperature in the case of the samples H-AT
(0.2), H-AT(0.3) and H-AT(0.4). Instead, for the samples treated
with higher concentration of alkali, H-AT(0.5) and H-AT(0.7), the
position for the maximum was maintained without changes as
compared to the untreated catalyst.
Table 2 summarized the acidic values of the catalysts. The
amount of total acid sites determined by pyridine TPD analyses
was notably increased for the treated samples in comparison with
the non-treated zeolite, and the generated acid sites had mainly
high strength, as shown in the pyridine TPD profiles.
In conclusion, the total amount of acid sites determined by pyr-
idine TPD analyses was higher in the zeolite after the alkaline
treatment. The strength and amount of the acid sites were compa-
rable between the modified zeolites, but the samples H-AT(0.5)
and H-AT(0.7) presented the stronger acidity.
f
As a result of the preferential silicon extraction, the Si/Al ratio in
the treated zeolite was lower than in the non-treated sample and
thus the number of potential Brønsted acid sites per unit weight
should be increased.
e
d
Fig. 5B shows the Ammonia TPD results for the untreated mate-
rial and for the samples H-AT(0.2) and H-AT(0.7). In this case, the
TPD profiles were similar, indicating that the strength and total
amount of acid sites in the modified samples were similar to the
original H-ZSM-5. Similar results were obtained for the samples
H-AT(0.3), H-AT(0.4) and H-AT(0.5) (not shown). The acidic values
obtained by NH3 TPD for the modified samples (not shown) are in
agreement with those obtained using pyridine. It is clear that the
size of the ammonia molecule allowed the accessibility to all the
acid sites in the zeolites, also in the original H-ZSM-5 catalyst.
For the pyridine results, the differences in total acidity between
the untreated zeolite and the modified materials was due to the
mesoporosity generated by the alkaline treatment, which facilitate
the accessibility of the pyridine to the acid sites.
c
b
a
80
60
40
20
0
Fig. 4. 27Al-MAS-NMR spectra samples (a) H-ZSM-5 (b) H-AT(0.2), (c) H-AT(0.3), (d)
H-AT(0.4), (e) H-AT(0.5) and (f) H-AT(0.7).

22
C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
0.4
A
B
b
d
0.3
0.2
0.1
c
f
a
a
e
f
b
100 200 300 400 500 600 700 800
100 200 300 400 500 600 700 800
Temperature (°C)
Temperature (ºC)
Fig. 5. (A) Pyridine TPD profiles (a) H-ZSM-5, (b) H-AT(0.2), (c) H-AT(0.3), (d) H-AT(0.4), (e) H-AT(0.5) and (f) H-AT(0.7). (B) Ammonia TPD profiles (a) H-ZSM-5, (b) H-AT(0.2)
and (f) H-AT(0.7).
Table 2
Acidic results.
parison with the non-treated zeolite. This is in agreement with the
pyridine TPD results. The non-treated H-ZSM-5 zeolite presented
a
Sample
Acidity (mmol gꢀ1
)
B/Lb
mainly Lewis acid sites (B/L = 0.53). In addition, a pyridine FTIR
analyses with thermo-desorption at different temperatures was
performed. Fig. 6B shows results obtained with the non-treated
H-ZSM-5 zeolite. It can be observed that the intensity of the Lewis
acid sites signal for this material was too weak and practically it
disappears after evacuation at 200 °C (curve b), while a small
amount of Brønsted acid sites remains after evacuation at 400 °C
(curve d). The Brønsted acid sites in the H-ZSM-5 were more ther-
mally stable. Therefore, it can be concluded that the low tempera-
ture peak in the pyridine TPD profile (Fig. 5, curve a) represents
weak Lewis acid sites, and the high temperature range corresponds
mainly to Brønsted acid sites. This assignment to the regions in the
TPD profile with the nature of the acid sites is valid also for the
treated zeolites.
The pyridine FTIR spectra for the zeolite modified by alkali
treatments followed by two ion exchange steps with NH4NO3 solu-
tion (Fig. 6A, curves b–f) showed an important increase in the size
of the signal at 1544 cmꢀ1. This result indicates that the concentra-
tion of framework Al (FAL) sites on the surface of the treated sam-
ples was enhanced; and consequently the concentration of
Brønsted acid sites was improved. This is because the greater
H-ZSM-5
H-AT(0.2)
H-AT(0.3)
H-AT(0.4)
H-AT(0.5)
H-AT(0.7)
0.37
0.48
0.43
0.47
0.45
0.43
0.53
1.86
2.61
2.10
2.62
2.04
a
Obtained from pyridine TPD results.
Obtained from pyridine FTIR results.
b
The nature of the acid sites was investigated using the Pyridine
FTIR technique. Fig. 6A shows the signals associated to chemically
adsorbed pyridine in the range 1700–1400 cmꢀ1 after evacuation
at 100 °C. The non-treated zeolite H-ZSM-5 (curve a) displayed
the bands characteristic of pyridine adsorbed on Lewis acid sites
at 1622 and 1450 cmꢀ1 (L), the bands attributed to Brønsted acid
sites at 1544 and 1636 cmꢀ1 (B), and the signal corresponding to
pyridine adsorbed on both Lewis and Brønsted acid sites at 1490
cmꢀ1 (L + B) [52].
For the alkaline treated zeolites (curves b–f), there was a very
important increase in the total intensity of the FTIR signal, in com-
C
B
A
d
f
d
e
c
c
d
c
b
b
b
a
L
a
L
L + B
B
B
a
B
L
B
L
1700
1600
1500
1400
1700
1600
1500
1400 1700
1600
1500
1400
Wavenumber (cm-1)
Wavenumber (cm-1)
Wavenumber (cm-1)
Fig. 6. Pyridine FTIR spectra. (A) After desorption at 100 °C for the catalysts (a) H-ZSM-5, (b) H-AT(0.2), (c) H-AT(0.3), (d) H-AT(0.4), (e) H-AT(0.5) and (f) H-AT(0.7). (B) For the
H-ZSM-5 sample after desorption at (a) 100 °C, (b) 200 °C, (c) 300 °C and (d) 400 °C. (C) For the H-AT(0.5) sample after desorption at (a) 100 °C, (b) 200 °C, (c) 300 °C and (d)
400 °C.

C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
23
external surface area and mesoporosity obtained upon the alkaline
treatment, led to a higher exchange of NH+4 that were transformed
in H+ by calcination. Fig. 6C shows the pyridine FTIR results for the
sample H-AT(0.5) after evacuation at different temperatures. These
results demonstrate that the Brønsted acid sites are strong, main-
taining adsorbed pyridine even after evacuation at 400 °C.
The concentrations of Brønsted and Lewis acid sites were esti-
mated from the intensities of the band at 1544 and 1450 cmꢀ1
respectively in Fig. 6A, following the procedure of Emeis [53].
These values were used to calculate the B/L ratio for all the samples
(Table 2). These results showed a higher concentration of Brønsted
acid sites accessible to pyridine on the treated samples. These
Brønsted sites were proposed as the catalytic species where the
dehydration of glycerol to acrolein takes place [20].
It is very important to highlight the high strength of the
Brønsted acid sites in the H-AT(0.5) treated zeolite, that are able
to maintain the pyridine adsorbed even after the treatment at
400 °C (Fig. 6C). On the other hand, the original zeolite showed a
decrease in the amount of pyridine adsorbed as a function of the
desorption temperature in the range 100–400 °C, as indicated in
Fig. 6B. This different strength of the acid sites explains the lack
of a direct correlation between the Si/Al ratio and the amount of
acid sites quantified by pyridine TPD in each catalyst. The pretreat-
ment at 150 °C carried out during this analysis, proportionally des-
orbs more pyridine from the weaker acid sites of the original
zeolite, thus detecting a significant lower acid sites density com-
pared to the zeolite treated with the sodium carbonate.
ier unidentified compounds were found, defined as ‘‘others”. Acet-
als, which are formed by the reaction between glycerol and
acrolein, are included within this group of unknown products.
Dehydration of glycerol has been shown to first produce acetol
and acrolein [54]. Two distinct and independent pathways occurs,
one leading to acrolein through 3-hydroxypropenal (a very unsta-
ble product) and the other forming acetol. The first pathway
implies the removal of the central alcohol function in the glycerol
molecule, whereas the second implies the removal of one of the
two terminal alcohol groups. This mechanism has been observed
on a wide variety of catalysts, including base and acid-supported
catalysts and zeolites [13,16]. The predominant route depends
mainly on the nature of the acid sites. The mechanism for the for-
mation of acrolein over Brønsted acid sites has been previously
suggested, as well as the reaction to acetol over Lewis acid sites
[17]. As it was mentioned in the introduction section, a cooperative
role of Brønsted and Lewis acid sites in the glycerol dehydration to
acetol is also possible [23].
Acetone can be obtained from the hydrodeoxygenation of acetol
and acetaldehyde is presumably derived from the catalytic decom-
position of the feed [16]. Corma et al. [8] found that acetaldehyde
can also be formed from acetol, and proposed a mechanism for
acetaldehyde formation via hydrogen-transfer mechanism, with a
dehydrogenated dione intermediate. This intermediate can also
form propanal [8].
Witsuthammakul and Sooknoi [55], and previously Chai et al.
[13] reported that acrolein is highly reactive and it can undergo
secondary reactions such as condensation and oligomerization,
forming high molecular weight products, namely as ‘‘others”.
These products are cyclic unsaturated and oxygenated compounds
such as phenol, dihydrofuran, cyclopentenone, methyl cyclopen-
tenone and cyclohexenone. They are responsible for coke forma-
tion and cause the catalyst deactivation, particularly in large pore
zeolites [56]. Kim et al. reported XRD results indicating that the
carbon deposited on the catalyst surface exist as amorphous spe-
cies [57]. It was found that the strong acid sites led to low acrolein
selectivity due to a fast coke deposition [13].
3.9. Activity results
3.9.1. Reaction network discussion
Fig. 7 shows a complete reaction scheme adapted from Corma
et al. [8] based on the findings of the present work, including the
main products. Glycerol can participate in numerous consecutive
and parallel reactions as dehydration, cracking, and hydrogen
transfer, all of them catalyzed by acid sites. In addition to acrolein,
others oxygenated compounds produced by glycerol conversion
are acetaldehyde, and hydroxyacetone (acetol). Acetone, propanol
and methanol were detected in trace amounts. A number of heav-
It is also known that glycerol can form oligomers (polyglyc-
erols), which may lead to coking reactions [58]. The same mecha-
Propanal
O
H
OH
O
Acrolein
OH
O
Olefins
Aromatics
Coke
HO
OH
Oligomers
Acetaldehyde
Glycerol
O
OH
Acetol
O
OTHERS
Acetone
Fig. 7. Scheme of reaction for glycerol dehydration.

24
C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
nism is applicable to acetol as well. Coke is also formed from
acetaldehyde [59], and probably acrolein; coke can be formed from
many sources.
c
3.9.2. Catalytic results
b
Fig. 8 shows the catalytic activity results obtained at a space
velocity WHSV = 0.75 hꢀ1 and 275 °C. The glycerol conversion as
a function of time on stream (TOS) is showed in Fig. 8A. A decrease
in the conversion with time on stream was observed in all cases,
due to the deactivation by coke formation, which is the main cause
of deactivation during glycerol dehydration with solid acids [60].
The TPO profiles of deactivated catalysts were similar and the per-
centage of carbon deposits after 7 h of reaction was around 11 wt%
in all samples. Fig. 9 shows the TPO results for the H-ZSM-5, H-AT
(0.5) and H-AT(0.7).
a
100 200 300 400 500 600 700 800
Temperature (ºC)
After 7 h of time on stream the conversion obtained with the H-
ZSM-5 zeolite felt below 50%. In the case of the alkaline treated
zeolites the stability was much better, being the conversion close
to 70% at the same time on stream. It is also observed that in the
case of the original zeolite the conversion continues decreasing,
while it tends to stabilize in the modified catalysts. Possibly, the
deactivation by blocking the mouth of pores with coke is more
important in the original zeolite. The higher conversion reached
for the treated zeolites is associated to the higher amount of acces-
sible acid sites obtained on these solids. The availability of more
active sites allows conversions of practically 100% up to TOS of 3
h, while at the same time the conversion for the non-treated zeo-
lite was below 80%. The changes in the selectivity during the first
hours of the reaction indicate that in spite of maintaining the con-
version in 100%, deactivation by coke deposition is taking place.
The increased mesoporosity, also attenuates the catalyst deacti-
vation by coke formation. The geometry of channels and the
intracrystalline diffusivity have great impact on the formation of
coke [42]. The samples H-AT(0.5) and H-AT(0.7) which were syn-
thesized with the greater external surface area and the narrowed
pore size distribution, showed the better stability in reaction
(70% Conversion at TOS = 7 h). The relative deactivation calculated
according to Eq. (4) were: 49.4%, 35.6%, 36%, 34.8%, 30.2% and
31.1% for samples H-ZSM-5, H-AT(0.2), H-AT(0.3), H-AT(0.4), H-
AT(0.5) and H-AT(0.7) respectively.
Fig. 9. TPO profiles for catalysts after activity tests. (a) H-ZSM-5, (b) H-AT(0.5) and
(c) H-AT(0.7).
there is also a correlation between these variables, since the pyri-
dine FTIR results previously discussed showed a higher B/L ratio in
the treated zeolites and accordingly an improvement in the selec-
tivity to acrolein was obtained.
3.9.3. Effect of temperature and space velocity
Fig. 10 shows the glycerol conversion and acrolein selectivity
with TOS obtained for the catalysts H-ZSM-5 and H-AT(0.7) at
275° C and different space velocities. It can be seen that the cat-
alytic behavior is better for the modified zeolite at each spatial
velocity. If each catalyst is analyzed separately, it is clear that an
increase in the feed rate to the reactor produces a much more pro-
nounced deactivation of the material.
Although acrolein is one of the primary products of glycerol
dehydration, as discussed in Section 3.9.1, if the selectivities values
are compared at TOS = 1 h, it is observed that an increase in the
space velocity produce a decrease both in the glycerol conversion
and in the selectivity to acrolein. It is important to observe that this
correlation (at lower conversion, lower selectivity to acrolein)
occurs for different catalysts at a given space velocity (see Fig. 8)
or comparing data obtained at different space velocities and differ-
ent catalysts (Fig. 10). For example, after 5 h on oil, where the cat-
alytic behavior is rather stable, at a given TOS the selectivity
decreases as the conversion decreases. Therefore, glycerol is
involved in parallel reactions that lead to the formation of other
compounds decreasing the selectivity to acrolein at low conver-
Besides, Fig. 8B shows that the selectivity to acrolein was also
improved for the treated zeolite. Initial selectivities (TOS = 1 h)
close to 75% were obtained, well above the 50% reached for the
non-treated material. The nature and strength of the acid sites
plays a fundamental role in the acrolein selectivity. It was pro-
posed that the Brønsted acid sites were more active than the Lewis
acid sites to convert glycerol to acrolein [13]. In these catalysts,
100
100
A
B
90
90
80
70
60
50
40
30
20
10
0
80
70
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Time on Stream (h)
Time on Stream (h)
Fig. 8. (A) Glycerol conversion and (B) Acrolein selectivity with time on stream at WHSV = 0.75 hꢀ1 and T = 275 °C for the catalysts: ( ) H-ZSM-5, ( ) H-AT(0.2), ( ) H-AT
(0.3), ( ) H-AT(0.4), ( ) H-AT(0.5), ( ) H-AT(0.7).

C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
25
100
100
90
80
70
60
50
40
30
20
10
0
A
B
90
80
70
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Time on Stream (h)
Time on Stream (h)
Fig. 10. (A) Glycerol conversion and (B) Acrolein selectivity at T = 275 °C with time on stream at different WHSV. H-ZSM-5 ( = 0,75 hꢀ1
,
= 3 hꢀ1) and H-AT(0.7) ( = 0,75
hꢀ1, j = 1,5hꢀ1
,
= 3 hꢀ1).
Table 3
Glycerol conversion and products selectivity at T = 275 °C and WHSV = 0.75 hꢀ1
.
Sample
TOS
H-ZSM-5
1 h
H-AT(0.2)
1 h
H-AT(0.3)
1 h
H-AT(0.4)
1 h
H-AT(0.5)
1 h
H-AT(0.7)
7 h
7 h
7 h
7 h
7 h
1 h
7 h
Glycerol conversion (%)
Acrolein yield (%)
Selectivity (%)
Acetaldehyde
Propanal
Acetol
Acrolein
Others
95
53
48
13
100
70
64.4
18.0
100
66
64
14.1
100
72
65.2
19.6
100
74
69.8
20.2
100
72
68.9
22.7
3.8
3.0
15
53
25.2
2.43
0.88
15.2
27.7
53.1
4.3
3.7
15
70
6.7
1.5
4.5
3.8
18
66
7.2
1.76
1.16
19
22
55.3
4.3
3.3
16
72
4.4
1.1
0.73
17
30
50.7
5.0
4.0
15
74
2.5
1.8
0.84
16
29
51.6
5.0
4.7
15
72
3.4
1.87
1.0
15.8
33
0.82
16.1
28
53.1
47.4
100
90
80
70
60
50
40
30
20
10
0
100
WHSV = 0.75 h-1
WHSV = 1.5 h-1
90
80
70
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Time on Stream (h)
Time on Stream (h)
100
90
80
70
60
50
40
30
20
10
0
WHSV = 3 h-1
0
1
2
3
4
5
6
7
8
Time on Stream (h)
Fig. 11. Product distribution at different WHSV for H-AT(0.7) catalyst at 275 °C. Acrolein ( ), acetaldehyde ( ), propanal ( ), acetol ( ), others ( ).

26
C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
100
100
90
80
70
60
50
40
30
20
10
0
A
B
90
80
70
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Time on Stream (h)
Time on Stream (h)
Fig. 12. (A) Glycerol conversion and (B) Acrolein selectivity with time on stream at WHSV = 0.75 hꢀ1 for the catalysts H-AT(0.7) at: T = 275 °C ( ), T = 300 °C (j).
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
A
B
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Time on Stream (h)
Time on Stream (h)
Fig. 13. (A) Glycerol conversion and (B) Acrolein selectivity with time on stream at WHSV = 0.75 hꢀ1 and T = 275 °C for the catalysts: ( ) H-ZSM-5 AT NaOH 0.01 mol Lꢀ1, (d)
H-AT(0.7).
sion. Table 3 summarized the glycerol conversion, yield to acrolein,
and the initial and final selectivities to each compound. Fig. 11
shows the selectivities to the main compounds of this reaction sys-
tem at different space velocities for the H-AT(0.7) catalyst. It can be
observed that as the WHSV increases, the selectivity to acrolein,
propanal and acetaldehyde decreases. The amount of acetol
path is favored via acetaldehyde and the formation of coke precur-
sors. In agreement with results shown above, at lower conversion
lower acrolein selectivity, what indicates that the Brønsted acid
sites involved in the dehydration step deactivates faster than the
Lewis sites.
´
´
remains practically constant; however the quantity of others is
noticeably increased. As above discussed, acrolein is formed by
dehydration of glycerol on Brønsted acid sites, whiles acetol is
formed on the Lewis acid sites. Results shown in Fig. 11 and Table 3
therefore, imply that the Brønsted acid sites deactivate faster than
the Lewis acid sites and for this reason acrolein selectivity
decreases with time while acetol selectivity increases. Since the
3.9.4. Effect of the anion during the alkaline treatment
Fig. 13 shows the activity results of the H-AT(0.7) catalyst and
the results obtained with a H-ZSM-5 zeolite treated with 0.01
mol Lꢀ1 NaOH solution. It can be seen that in addition to the
advantages mentioned above about the material yield in the prepa-
ration, the catalytic performance is much better for the zeolite
modified with Na2CO3. This is attributed to the lower Si/Al ratio
of the latter, in which the Si was selectively removed from the zeo-
lite structure.
´
´
selectivity of the compounds grouped in others increases as the
conversion decreases, it is possible to conclude that these com-
pounds are formed directly from the glycerol, being primary prod-
ucts. For this reason, the reaction scheme shown in Fig. 7 includes
this reaction step. Fig. S2 (Supplementary Material) shows typical
chromatograms at low and high glycerol conversion obtained at
1 h and 7 h on stream respectively. Peaks without labels were
not identified, and corresponds to the compounds grouped in
‘others’. It can be seen that this group contains several compounds
with small area.
4. Conclusions
A procedure to develop mesoporosity in an H-ZSM-5 zeolite
was studied, consisting in alkaline treatment with Na2CO3 aqueous
solutions and ionic exchanges with NH4NO3 solutions followed by
calcinations. The selected concentrations, temperatures and time
of treatment were appropriate to reach a selective removal of Si
from the zeolite framework, maintaining the Al content, and gen-
erating extra mesoporosity. In this manner, the concentration of
Brønsted acid sites was improved in the zeolite, and also the acces-
sibility to the active sites. The activity and selectivity was better
than in the case of the non-treated zeolite, and furthermore the
stability in reaction was increased.
The effect of temperature was also studied (Fig. 12). An increase
of this variable from 275 °C to 300 °C was detrimental to the cat-
alytic performance of the materials. For example, for the catalyst
H-AT(0.7) an important increase in the rate of deactivation was
observed, added to a decrease in the selectivity to acrolein. This
is associated with the fact that at that temperature the reaction

C.D. Lago et al. / Journal of Catalysis 366 (2018) 16–27
27
[22] C.L. Lima, S.J.S. Vasconcelos, J.M. Filho, B.C. Neto, M.G.C. Rocha, P. Bargiela, A.C.
Oliveira, Appl. Catal. A 399 (2011) 50–62.
[23] G. Foo, D. Wei, D. Sholl, C. Sievers, ACS Catal. 4 (2014) 3180–3192.
[24] B. Yilmaz, U. Muller, Top Catal. 52 (2009) 888–895.
[25] E. Taarning, C.M. Osmundsen, X. Yang, B. Voss, S.I. Andersena, C.H. Christensen,
Energy Environ. Sci. 4 (2011) 793–804.
[26] K. Egeblad, C.H. Christensen, M. Kustova, Chem. Mater. 20 (2008) 946–960.
[27] J.C. Groen, L.A.A. Peffer, J.A. Moulijn, J. Perez-Ramírez, Colloids Surf. A 241
(2004) 53–58.
[28] B. Vogel, C. Schneider, E. Klemm, Catal. Lett. 79 (2002) 107–112.
[29] S. van Donk, A.H. Janssen, J.H. Bitter, K.P. de Jong, Catal. Rev. 45 (2003) 297.
[30] R. Le Van Mao, S.T. Le, D. Ohayon, F. Caillibot, L. Gelebart, G. Denes, Zeolites 19
(1997) 270.
The alkaline treatment with Na2CO3 leads to a significantly
higher yield during the preparation procedure as compared to
the treatment with NaOH. Also the activity results obtained with
the material treated with Na2CO3 are much better than the
observed for the zeolite modified with NaOH at equivalent pH.
The Brønsted acid sites deactivate faster than the Lewis sites,
and this behavior leads to a lower acrolein selectivity, and higher
selectivity for acetol formation as catalyst deactivates.
Acknowledgements
[31] J. Perez-Ramirez, F. Kapteijn, J.C. Groen, A. Domenech, G. Mul, J.A. Moulijn, J.
Catal. 214 (2003) 33.
The authors thank Martina Braun for her technical assistant
with measuring the 27Al MAS NMR spectra at the Department of
Chemistry and Catalysis Research Center, Technische Universität
München, Garching, Germany. The authors wish to acknowledge
the financial support received from the following primary funders:
Universidad Nacional del Litoral (Santa Fe, Argentina, grant CAID
PACT 96), ANPCyT (Argentina) (grant PICT 2013-1252) and CONI-
CET (Argenitna) (grant PIO 13320130100209CO).
[32] T. Suzuki, T. Okuhara, Micropor. Mesopor. Mat. 43 (2001) 83.
[33] Y.N. Li, S.L. Liu, Z.K. Zhang, S.J. Xie, X.X. Zhu, L.Y. Xu, Appl. Catal. A 338 (2008)
100.
[34] M. Ogura, S. Shinomiya, J. Tateno, Y. Nara, M. Nomura, E. Kikuchi, M.
Matsukata, Appl. Catal. A: General 219 (2001) 33–43.
[35] R. Le Van Mao, S. Xiao, A. Ramsaran, J. Yao, J. Mater. Chem. 4 (1994) 605–610.
[36] J.C. Groen, L.A.A. Peffer, J.A. Moulijn, J. Pérez-Ramírez, Chem. Eur. J. 11 (2005)
4983–4994.
[37] S. Brunauer, P.H. Emmet, E. Teller, J. Am. Chem. Soc. 60 (1938) 309.
[38] B.C. Lippens, J.H. de Boer, J. Catal. 4 (1965) 319.
[39] J.C.P. Broekhoff, J.H. de Boer, J. Catal. 9 (1967) 8.
[40] J. Jiao, J. Kanellopoulos, W. Wang, S. Ray, H. Foerster, D. Freude, M. Hunger,
PCCP 7 (2005) 3221–3226.
References
[41] H. Chen, T. Sun, D. Sui, J. Dong, Anal. Chim. Acta 698 (2011) 27–35.
[42] Y. Gu, N. Cui, Q. Yu, C. Li, Q. Cui, Appl. Catal. A: General 429–430 (2012) 9–16.
[43] H.P. Decolatti, B.O. Dalla Costa, C.A. Querini, Microp. Mesop. Mat. 204 (2015)
180–189.
[44] V. Rac, V. Rakic, Z. Miladinovic, D. Stosic, A. Auroux, Thermochim. Acta 567
(2013) 73–78.
[45] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T.
Siemieniewska, Pure Appl. Chem. 57 (1985) 603–619.
[46] J.C. Groen, L.A.A. Peffer, J. Pérez-Ramírez, Microporous Mesoporous Mater. 60
(2003) 1–17.
[1] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502.
[2] D.L. Klass, Biomass for Renewable Energy, Fuels, and Chemicals (Chapter 1),
Elsevier Inc., Amsterdam, 1998, pp. 1–27.
[3] M. Ayoub, A. Abdullah, Renew. Sust. Energ. Rev. 16 (2012) 2671–2686.
[4] R.L. Maglinao, B.B. He, Ind. Eng. Chem. Res. 50 (2011) 6028–6033.
[5] M. Pagliaro, R. Ciriminna, H. Kimura, M. Rossi, G. Della Pina, Angew. Chem., Int.
Ed. 46 (24) (2007) 4434–4440.
[6] C.H. Zhou, J.N. Beltramini, Y.X. Fan, G.O. Lu, Chem. Soc. Rev. 37 (2008) 527–549.
[7] Y. Moro-Oka, W. Ueda, Adv. Catal. 40 (1994) 233–273.
[8] A. Corma, G. Huber, L. Sauvanauda, P. O’Connor, J. Catal. 257 (2008) 163–171.
[9] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Applied Catal. A: Gen. 353 (2009) 213–
222.
[10] J.L. Dubois, C. Duquenne, W. Hölderich, US Patent 7.396.962, 2008.
[11] K. Omata, S. Izumi, T. Murayama, W. Ueda, Catal. Today 201 (2013) 7–11.
[12] S. Ramayya, A. Brittain, C. DeAlmeida, W. Mok, M.J. Antal, Fuel 66 (10) (1987)
1364–1371.
[13] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Green Chem. 9 (2007) 1130–1136.
[14] L.Z. Tao, S.H. Chai, Y. Zuo, W.T. Zheng, Y. Liang, B.Q. Xu, Catal. Today 158 (2010)
310–316.
[15] J. Deleplanque, J.L. Dubois, J.F. Devaux, W. Ueda, Catal. Today 157 (2010) 351–
358.
[16] E. Tsukuda, S. Sato, R. Takahashi, T. Sodesawa, Catal. Commun. 8 (2007) 1349–
1353.
[17] A. Alhanash, E.F. Kozhevnikova, I.V. Kozhevnikov, Appl. Catal. A: Gen. 378
(2010) 11–18.
[47] A. Cizmek, B. Subotic, I. Smit, A. Tonejc, R. Aiello, F. Crea, A. Nastro,
Microporous Mater. 8 (1997) 159.
[48] C. Su, D.L. Suarez, Clays Clay Miner. 45 (1997) 814–825.
[49] S. Abell, A. Bonilla, J. Perez-Ramirez, Appl. Catal. A 364 (2009) 191–198.
[50] J.C. Groen, J.A. Moulijn, J. Perez-Ramirez, J. Mater. Chem. 16 (2006) 2121–2131.
[51] S.M.T. Almutairi, B. Mezari, G.A. Filonenko, P.C.M.M. Magusin, M.S. Rigutto, E.A.
Pidko, E.J.M. Hensen, Chem. Cat. Chem. 5 (2013) 452.
[52] J.P. Marques, I. Gener, P. Ayrault, J.C. Bordado, J.M. Lopes, F. Ramoa Ribeiro, M.
Guisnet, Microporous Mesoporous Mater. 60 (2003) 251–262.
[53] C.A. Emeis, J. Catal. 141 (1993) 347–354.
[54] M.R. Nimlos, S.J. Blanksby, X. Qian, M.E. Himmel, D.K. Johnson, J. Phys. Chem. A
110 (2006) 6145.
[55] A. Witsuthammakul, T. Sooknoi, Appl. Catal. A 413–414 (2012) 109–116.
[56] D.M. Bibby, R.F. Howe, G.D. McLellan, Appl. Catal. A 93 (1992) 1–34.
[57] Y.T. Kim, K.D. Jung, E.D. Park, Appl. Catal. A 393 (2011) 275–287.
[58] J. Barrault, J.M. Clacens, Y. Pouilloux, Top. Catal. 27 (2004) 137–142.
[59] A.G. Gayubo, A.T. Aguayo, A. Atutxa, R. Aguado, J. Bilbao, Ind. Eng. Chem. Res.
43 (2004) 2610.
[18] B. Katryniok, S. Paul, M. Capron, C. Lancelot, V. Bellière-Beca, P. Rey, F.
Dumeignil, Green Chem. 12 (2010) 1922–1925.
[19] C.J. Jia, Y. Liu, W. Schmidt, A.H. Lu, F. Schüth, J. Catal. 269 (2010) 71–79.
[20] A.S. de Oliveira, S.J.S. Vasconcelos, J.R. de Sousa, F.F. de Sousa, J.M. Filho, A.C.
Oliveira, Chem. Eng. J. 168 (2011) 765–774.
[60] Y.T. Kim, K.D. Jung, E.D. Park, Microporous Mesoporous Mater. 131 (2010) 28–
36.
[21] H. Atia, U. Armbruster, A. Martin, Appl. Catal. A 393 (2011) 331–339.