Catalysis Today 289 (2017) 20–28

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

The multiple benefits of glycerol conversion to acrolein and acrylic

acid catalyzed by vanadium oxides supported on micro-mesoporous
MFI zeolites
Luiz G. Possato, Thiago F. Chaves, Wellington H. Cassinelli, Sandra H. Pulcinelli,
Celso V. Santilli, Leandro Martins ∗
Instituto de Química, UNESP − Univ. Estadual Paulista, Rua Prof. Francisco Degni 55, 14800-060, Araraquara, SP, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: The ZSM-5 zeolite (MFI structure, Si/Al = 40) was treated using NaOH and either oxalic acid or HCl to
Received 12 April 2016 obtain hierarchical materials with different characteristics, followed by impregnation with vanadium
Received in revised form 20 July 2016 oxides (V2 O5 ) to generate redox-active sites. The impact of the multiple treatments on the efficiency and
Accepted 2 August 2016
stability of the catalysts in the conversion of glycerol to acrolein and acrylic acid was investigated and
Available online 16 August 2016
correlated with catalyst porosity, acidity, and chemical composition. The treated and impregnated V2 O5
catalysts were subjected to XRD, 27 Al NMR, nitrogen physisorption, TPD-NH3 , TG, and UV–Vis analyses,
Keywords:
in order to associate the properties of the catalysts with their activities. The studies showed that the
Glycerol oxidehydration
V2 O5 /MFI catalyst
catalytic performance of the materials depended on the acidic and textural properties of the zeolites,
Acrolein which influenced both the dispersion of V2 O5 and its interaction with the acid sites of the supporting
Acrylic acid zeolites. All the catalysts provided conversion values exceeding 65%, even after 6 h on glycerol stream.
Micro-mesoporous zeolites The distribution of products strongly reflected the effects of pore formation, acid treatment with oxalic
acid or HCl, and the presence of vanadium oxide. The effects of these modifications resulted in higher
selectivity to acrolein and acrylic acid, a reduced rate of coke accumulation in the zeolite, and a longer
catalyst lifetime.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction a high conversion and selectivity to glycols has been obtained by

glycerol hydrogenolysis using catalysts based on metallic Ni, Ru,
The growth of greenhouse gas emissions and the limited and Cu [4–7].
reserves of easily extracted fossil fuels have led researchers and Another important conversion of glycerol into petrochemical-
industry to pursue new alternatives to replace, even partially, the type compounds is the formation of acrolein and acrylic acid,
use of fossil fuels. As a result, the use of compounds derived from which are used in the manufacture of resins. Acrolein can be
biomass, such as biogas, ethanol from sugarcane, and biodiesel has obtained by gas phase glycerol dehydration on acid catalysts such
increased in recent years. In the case of biodiesel, production has as heteropolyacids [8], impregnated phosphate groups on metal
increased and consequently there has been an increase in the con- oxides [9], sulfated zirconia [10], Nb2 O5 [11], mixed oxides [12],
comitant formation of coproduced glycerol [1–3]. zeolites [13–15], functionalized mesoporous silica [16,17], and
Glycerol is a focus of green catalytic processes, because vanadium-silicates [18]. The glycerol can also be dehydrated in
this molecule offers interesting chemical versatility that can be liquid phase [19,20]. Acrylic acid is obtained from the oxidation
exploited for the formation of compounds that are currently of acrolein. The one-step conversion of glycerol into acrylic acid
provided by the petrochemical industry. An example of glycerol val- using bifunctional catalysts with acid and redox sites occurs accord-
orization is the synthesis of 1,2-propanediol and 1,3-propanediol, ing to Scheme 1 [21–26]. An interesting aspect of coupling the
which are used as antifreeze fluids in automobiles. In industry, both two reactions is the mutually supporting endothermic dehydra-
compounds are obtained from the hydration of propene. However, tion of glycerol (H0 = 3.04 kcal/mol) and exothermic oxidation of
acrolein (H0 = − 61.02 kcal/mol).
For glycerol dehydration using zeolites, maximum performance
∗ Corresponding author. of the catalyst is achieved by combining Brønsted acid sites of
E-mail address: leandro@iq.unesp.br (L. Martins). medium strength (strong acid sites lead to severe coke formation,

http://dx.doi.org/10.1016/j.cattod.2016.08.005
0920-5861/© 2016 Elsevier B.V. All rights reserved.
 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28 21

OH
-H O -H O + ½ O2 O ously with glycerol dehydration to acrolein. Byproducts include
HO OH 2 O OH 2 O
acetaldehyde, acetol, and acetic acid, as well as very harmful and
OH
glycerol 3-hydroxy propanal acrolein acrylic acid
deactivating coke molecules. After catalytic experiments with bare
zeolites, the polymerization of bulky molecules on the surfaces
acid sites redox sites of the catalysts led to coke formation and a characteristic black
appearance. However, in the previous work it was found that when
Scheme 1. Glycerol oxidehydration (dehydration combined with oxidation) on
a V2 O5 /zeolite catalyst was used, the coke was continuously oxi-
bifunctional acid and redox active sites.
dized due to the presence of well-dispersed vanadium oxides on
the zeolite surface, which maintained the catalytic sites active for
while weak acid sites are less capable of converting glycerol) [27], longer periods.
porosity (which enhances the diffusion of glycerol and acrolein), The aim of the present work was to explore further the multi-
and high specific area (which increases access to catalytic sites). ple benefits of porous V2 O5 /MFI catalysts in the one-step glycerol
For instance, members of the lamellar MWW zeolite family, which conversion to acrylic acid. The work focused on zeolite supports
includes microporous MCM-22, pillared MCM-36, and delaminated prepared by sequential processes of desilication (in NaOH solu-
ITQ-2, offer advantageous characteristics for glycerol dehydration tion) and dealumination (in HCl or oxalic acid solutions) in order
[22]. Following pillarization and delamination of the MWW struc- to tailor the pores and the quality of acid sites derived from either
ture, the strengths of acid sites decrease, but the increases in aluminum in tetrahedral coordination in the zeolite or from EFA.
mesopores and specific area raise the overall performance of the Improved transformation of glycerol was achieved on the micro-
catalyst [28]. mesoporous V2 O5 /MFI zeolites, due to higher catalytic conversion,
Despite the attraction of lamellar zeolites for use in glyc- improved selectivity to acrolein and acrylic acid, extended catalyst
erol dehydration, the laborious multiple steps and the expense stability, and decreased coke formation.
associated with catalyst preparation are notable disadvantages.
Alternatively, the desilication of commercially available zeolites by 2. Experimental
treatment with sodium hydroxide solution seems to be more prac-
tical [29,30]. The alkaline process is simple, with hydroxyl groups 2.1. Preparation of zeolite supports
attacking and removing silicon atoms from the zeolite structure,
creating randomly distributed pores in the zeolite crystals. The Zeolite of MFI structure (Si/Al mole ratio of 40) was kindly
diameter and volume of the pores can be tuned by adjusting the provided by Zeolyst (USA). The sample was submitted to alka-
concentration of the alkaline solution and by varying the exposure line treatment at 60 ◦ C for 1 h using an aqueous solution of NaOH
time of the zeolite (usually a few minutes) and the desilication (0.6 mol/L). Detailed information concerning the desilication proce-
temperature (which normally ranges from room temperature to dure is provided elsewhere [30,39]. The desilicated zeolite was then
a few tens of degrees Celsius) [31–37]. The broad distribution submitted to two different acid treatments using aqueous solu-
of mesopore families results in catalytic performance in glycerol tions of hydrochloric or oxalic acids. The acidic treatments were
dehydration similar to that of the MWW zeolites. performed under reflux using 0.1 mol/L acid solution. The MFI sup-
A disadvantage of the desilication method is that during the zeo- ports were denoted A (parent and untreated MFI zeolite), B (after
lite treatment process, aluminum atoms are removed as well as alkaline treatment), C (after treatment using H2 C2 O4 ), and D (after
silicon atoms. Silicon species are mostly found in the alkaline liq- treatment using HCl). After the sequential alkaline and acidic treat-
uid phase, but aluminum tends to form insoluble oligomeric species ments, all the supports (A, B, C, and D) were ion exchanged three
that can precipitate on the catalyst surface as extra-framework times with NH4 NO3 solution, at room temperature. The exchanged
aluminum atoms (EFA). Consequently, the mesopores created are NH4 + cations were thermally decomposed by heating the samples
obstructed due to an alkali-induced alumination of the external sur- for 3 h in a conventional muffle furnace, in air atmosphere, from
faces of the crystals, and the nature of the acid sites of the zeolite 25 ◦ C to 500 ◦ C at a heating rate of 10 ◦ C/min.
shifts from Brønsted to Lewis acid sites. This catalytic acid behav-
ior must be considered in the design of catalysts by desilication, 2.2. Preparation of the catalysts
because EFA sites are selective in converting glycerol into undesir-
able byproducts. However, the EFA can be removed from the zeolite V-MFI catalysts were obtained by incipient wetness impregna-
by acid leaching; as a result, the selectivity to acrolein is enhanced tion of the supports using an aqueous solution of vanadyl sulfate
and the diffusion of chemicals through the pores is increased due (0.05 mol/L). Approximately 50 mL slurry of vanadyl sulfate solu-
to the removal of aluminum species. tion and zeolite was stirred for 1 h at 25 ◦ C. The water was allowed
In the second step of glycerol conversion to acrylic acid to evaporate under vacuum at 40 ◦ C, and the samples were dried
(Scheme 1), redox active sites are required. Vanadium oxides are overnight at 100 ◦ C. Finally, the catalysts were subjected to a ther-
strong candidates for this purpose because they possess a very mal treatment for 2 h in an air atmosphere, with heating from 25 ◦ C
important redox characteristic, namely the capacity to adopt mul- to 500 ◦ C at a rate of 5 ◦ C/min. The contents of V2 O5 in all the sam-
tiple oxidation states. On these catalysts, acrolein is oxidized by ples was 10 wt.%.
removing a surface oxygen atom from V2 O5 , giving rise to acrylic
acid and an oxygen vacancy in V2 O5-x . In a subsequent step, the 2.3. Characterization of samples
catalytic site is oxidized and reestablished by feeding an excess
of molecular O2 in the stream (V2 O5-x +1 /2 O2 → V2 O5 ). This redox X-ray diffractograms of the supports and catalysts were
mechanism and the changes in V5+ /V4+ oxidation states during the obtained at the XPD beamline of the Brazilian Synchrotron
catalytic reaction are known as the Mars-Van Krevelen mechanism Light Laboratory (LNLS), using a Huber 4 + 2 circle diffractometer
[38]. equipped with an Eulerian cradle (model 513) placed approx-
In a recent publication, we described additional useful features imately 13 m from the double-bounce Si(111) monochromator
of the V2 O5 /zeolite catalytic system [20]. Besides the advan- (␭ = 1.377494 Å) [40]. The data were collected in high-resolution
tages mentioned above, vanadium oxides supported on zeolites mode, employing a Si(111) analyzer crystal and a Mythen detec-
were much less susceptible to deactivation, compared to the bare tor. Structural parameters were determined by the Rietveld profile
zeolites. Several parallel and unknown reactions occur simultane- method, using GSAS-EXPGUI software [41,42]. The scale factors,
22 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28

zero shifts, and backgrounds of the peak profiles, together with retention times of all the compounds were compared with those of
the lattice parameters, were refined by fitting with a sixth order authentic standards. The glycerol conversions (Xglycerol ) and prod-
Chebyshev polynomial. Pseudo-Voigt functions were employed for uct selectivities (S) were calculated using the following equations:
the peak profile refinements. Other parameters were not refined.
The textural properties of the supports and impregnated cata-
nGl input − nGl output ni
lysts were determined by means of nitrogen adsorption-desorption Xglycerol (%) = × 100 S (%) =
isotherms obtained at −196 ◦ C with a Micromeritics ASAP 2010 nGl input nGl input − nGl output
Zi
instrument, using a relative pressure (P/P0 ) interval of 0.001-0.998. × × 100
ZGl
The samples were previously decontaminated by degassing for 12 h
at 100 ◦ C under a vacuum of 1 × 10−5 Pa. The t-plot method was where:
used to distinguish between the micro- and mesopore contribu- nGI input and nGl output are the input and output flows of glycerol;
tions. Finally, the mesopore distribution was estimated using the ni is the flow of the products i (mol/min);
BJH method [24]. ZGl = 3 (the number of carbon atoms in the glycerol molecule);
The 27 Al NMR spectra were acquired using a Varian INOVA 500 Zi = number of carbon atoms in the products.
spectrometer equipped with a 7 mm probe and operated at 4.5 kHz. The deactivation (D) of the catalysts used in the glycerol dehy-
The experiments were performed using a spinning rate of 78.2 MHz, dration was calculated using:
acquisition time of 15.4 ms, pulse width of 2.4 ␮s, and recycle delay
of 0.1 s. The 27 Al chemical shifts were referenced to an aqueous (Xglycerol )at time zero − (Xglycerol )after 6h
D (%) = × 100
solution of Al(NO3 )3 (1 mol/L). Each spectrum was the result of 256 (Xglycerol )at time zero
scans.
Thermogravimetric analyses of the used catalysts (after the glyc- The glycerol oxidehydration reaction was conducted following
erol reaction) were performed with a TA Instruments SDT Q600 the same protocol as glycerol dehydration, but with the reaction
thermobalance. The samples were heated from 30 to 900 ◦ C, at temperature shifted to 350 ◦ C and with 20% of oxygen fed into the
10 ◦ C/min, under a 100 mL/min flow of synthetic air. stream (6 mL/min of O2 and 24 mL/min of N2 ). The O2 /glycerol mole
Scanning electron microscopy (SEM) analyses of the samples ratio of 4.5 ensured that an excess of oxygen was used.
were performed using an Inspect S50 microscope (FEI) operated at
10 kV, with a secondary electron detector and working distance of
10 mm. After 1 h in an ultrasonic bath, a suspension consisting of the
sample and acetone was deposited on an aluminum stub specimen
mount. The Si/Al mole ratio was determined by energy dispersive
X-ray spectroscopy (EDS), using the same microscope. The spectra
were measured in five different regions, and the composition was
determined by averaging the results obtained for each sample.
The acid sites of the calcined catalysts were determined by
temperature programmed desorption of ammonia (TPD-NH3 ). Pre-
viously, a 200 mg portion of each sample was degassed at 300 ◦ C
for 1 h under a 60 mL/min flow of helium. The temperature was
then decreased to 100 ◦ C and the sample was exposed for 1 h to a
flow of 1% ammonia in helium (60 mL/min). After surface satura-
tion, the sample was submitted to helium treatment at 100 ◦ C for
1 h to remove physisorbed ammonia. The TPD-NH3 analysis was
performed from 100 ◦ C to 600 ◦ C (at 10 ◦ C/min), under a 60 mL/min
flow of helium. The desorbed ammonia was monitored and quan-
tified with a Pfeiffer vacuum mass spectrometer connected to the
outlet stream of the tubular reactor.

2.4. Catalytic tests

The bare zeolite and V-MFI catalysts were tested for glyc-
erol dehydration and oxidehydration, respectively, at atmospheric
pressure. Previously, 100 mg portions of the samples were heated
at 300 ◦ C for 15 min, under a 30 mL/min flow of nitrogen. The glyc-
erol dehydration reaction was performed in a fixed-bed reactor
at 300 ◦ C for 6 h, using a solution of 10 wt.% glycerol in water. A
flow of 0.05 mL/min of this liquid solution was introduced to a
heated line using a syringe pump (KD Scientific), and the glycerol
was transported to the reactor using a 30 mL/min flow of nitrogen.
The unconverted glycerol and the reaction products were con-
densed continuously in a gas-liquid separator kept at 1 ◦ C. The
liquid was periodically collected, weighed, and injected into a gas
chromatograph (Model GC-2014, Shimadzu) equipped with a cap-
illary column (Rtx-1, 30 m, 0.32 mm, 1 ␮m) and a FID detector. For
each injection, a known mass of n-butanol was used as internal
Fig. 1. Rietveld plots of samples A (parent MFI), B (NaOH), C (H2 C2 O4 ), and D (HCl),
standard. Four drops of standardized solution was diluted in iso-
using ␭ = 1.3775 Å: experimental (black line), calculated (red line), and difference
propanol (1 mL) to prevent excess water from the chromatographic plot (blue line). (For interpretation of the references to colour in this figure legend,
column. The GC analyses were performed in triplicate, and the the reader is referred to the web version of this article.)
 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28 23

3. Results and discussion atoms in the structure tended to distort. The same effect led to
reduced crystallinity of samples B, C, and D.
3.1. Characterization of the zeolite supports The relative intensities of the peaks in the NMR spectra were
used for approximate calculation of the amounts of EFA (Table 1).
The XRD patterns of the MFI zeolites before treatment (sam- Comparison of the results for Si/Alglobal and the percentage of EFA
ple A in Fig. 1), after the alkaline treatment (sample B), and after showed that the oxalic acid treatment provided greater selectivity
the acidic treatments (samples C and D) revealed the presence of in removal of EFA without severely affecting the aluminum frame-
the main reflections related to the MFI structure: (011), (200), and work. The micropore volume only changed slightly, from 0.23 to
(051). The samples submitted to the treatments showed decreases 0.21 cm3 /g, while the mesopore volume showed a more signifi-
in peak intensity, consistent with the reduction of long-range order cant change, from 0.41 to 1.25 cm3 /g (Table 1). The reactivity of
caused by the extraction of silicon and aluminum atoms from the HCl was evidenced by a more pronounced reduction of the micro-
MFI structure [35]. The oxalic acid and HCl treatments also affected pore volume to 0.18 cm3 /g. Dealumination of zeolites with oxalic
the crystallinity of the samples, as clearly shown in the diffraction acid occurs because oxalic acid acts as a hydrolyzing agent by both
patterns of catalysts C and D, respectively. solubilizing and removing aluminum oxohydroxide species, and
The results of structural analysis by Rietveld refinement indi- forming aluminum oxalate complexes [47]. These complexes have
cated unit cell expansion after the alkaline treatment [43] (Table 1). a maximum size of around 0.64 nm, which is compatible with the
The removal of silicon (which has a mean Si-O bond length of pore dimensions of zeolites and enables mobility of the complexes
0.162 nm) and the consequent increase in the concentration of within the pores [44].
aluminum in the structure (mean Al-O bond length of 0.175 nm) The N2 adsorption-desorption isotherms of the MFI zeolites are
caused unit cell expansion from 5236.9 nm3 to 5258.1 nm3 [44,45]. shown in Fig. 3a. Sample A exhibited a Type I isotherm typical of a
The unit cell expansion was additionally influenced by the pres- pure MFI microporous structure [48]. After the alkaline treatment,
ence of aluminum atoms that remained in oligomeric form inside a hysteresis loop appeared and the capacity of this sample to adsorb
the pores of the zeolites. The subsequent acid treatments caused N2 increased, especially at relative pressure (P/P0 ) above 0.8, due
partial leaching of the EFA, and the structural analyses of these sam- to the creation of intra-crystalline mesopores. Additionally, the dis-
ples indicated changes in unit cell volume. The decreased unit cell ruption of the micropore structure of the zeolite in alkaline solution
volumes of the acid treated samples (C and D) revealed that the led to a decrease of the adsorption plateau at low P/P0 characteris-
treatments were effective. The global chemical analyses of the sup- tic of nitrogen adsorption limited by the micropore size. Analyses
ports showed growth in the aluminum fraction from sample A to of the BJH pore size distribution (Fig. 3b) showed the creation of
sample B (Si/Al ratios of 40 and 18, respectively), and growth in the mesoporosity in the supports submitted to alkaline and acidic treat-
silicon fraction from sample B to samples C and D. ments. The pore size distributions broadened after use of the acid
Samples C and D were (Table 1) dissimilar in terms of both unit treatments to clean the zeolite surface, giving rise to additional
cell volume and aluminum fraction, due to the differences in the porosity.
strengths of the acids used (the pKa values of hydrochloric acid The acid sites of the samples were characterized by
and oxalic acid are −6.3 and 1.2, respectively) and the mechanism of temperature-programmed desorption, using ammonia as a
EFA leaching. The action of HCl involved solubilization of oligomeric basic probe molecule. The TPD-NH3 curves obtained for the bare
aluminum oxide species, while oxalic acid acted as both an acid and and impregnated zeolites are shown in Fig. 4. The bare zeolite
a chelating agent. showed well-resolved NH3 desorption peaks, one in the low
The formation and leaching of EFA was followed using 27 Al MAS- temperature (LT) region at around 200 ◦ C, and one in the high
NMR (Fig. 2). The chemical shift peak centered at 54 ppm was due temperature (HT) region at around 420 ◦ C (Fig. 4a). The reference
to tetrahedral aluminum atoms in the zeolite structure, while the sample (␥-Al2 O3 ) presented a low amount of desorbed NH3 and a
signal between 0 and −20 ppm was due to the octahedrally coordi- wide desorption temperature range, characteristic of the presence
nated aluminum of EFA [46]. As mentioned before, the desilication of Lewis acid sites. The amounts of NH3 desorbed from the samples
process was not only responsible for the extraction of silicon atoms are given in Table 1. The NaOH treatment resulted in increased
from the structure, but also for aluminum removal, as is clearly NH3 desorption, due to the decrease in the Si/Al ratio (Table 1), in
shown in the detail of the decomposed lines for desilicated sample agreement with the results of the 27 Al NMR and chemical analyses.
B. In contrast, the parent zeolite contained only a minor amount Treatment of zeolite sample B with oxalic acid led to a decrease in
of EFA. A rigorous comparison of the NMR spectra showed that the quantity of acidic sites, due to the removal of EFA, as found in
the peak at 54 ppm became slightly broadened after the chemical the 27 Al NMR analyses. The treatment with HCl was less effective
processes, because the tetrahedral arrangement of the aluminum in removing the EFA, so the acidity was greater than for the sample
treated with oxalic acid [49].

Table 1
Chemical and textural properties of the bare and impregnated zeolite samples.

Samples Vmicro Vmeso ␮mol of Si/Alglobal a % EFAb Unit cell

(cm3 /g) (cm3 /g) NH3 /g volume (nm3 )

A 0.33 0 388 40 12 5236.9

B − NaOH 0.23 0.41 464 18 29 5258.1
C − H2 C2 O4 0.21 1.25 406 24 5 5250.0
D − HCl 0.18 1.30 529 26 11 5236.6
Reference Al2 O3 0 0.89 110 n/ac n/a n/a
V-A 0.26 0 579 n/a n/a n/a
V-B 0.18 0.21 545 n/a n/a n/a
V-C 0.21 0.24 427 n/a n/a n/a
V-D 0.20 0.24 581 n/a n/a n/a
a
Silicon to aluminum mole ratio determined by EDX (global chemical composition).
b
% EFA determined by deconvolution of the 27 Al NMR spectrum (contribution of tetrahedral aluminum species in the zeolite framework).
c
Not applicable.
24 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28

a 54 ppm
b Sample B

Intensity (arb. unit)

Intensity (arb. unit)

D

C
-1 ppm
B -13 ppm

A
80 60 40 20 0 -20 80 60 40 20 0 -20
Chemical Shift (ppm) Chemical Shift (ppm)
Fig. 2. (a) 27 Al NMR spectra of the parent and treated zeolites, and (b) deconvolution of the sample B spectrum.

2.0
A
B
a b
Adsorbed vol. (cm3/g)

dV/dlog(D) (cm³/g)
C 1.5
D
500
1.0

0.5

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1 10 100
Relative pressure (P/P0) Pore diameter (nm)

Fig. 3. (a) Nitrogen adsorption-desorption isotherms (filled and empty points correspond to nitrogen adsorption and desorption, respectively), and (b) BJH mesopore size
distributions from the desorption branches.

Fig. 4. Temperature-programmed desorption of ammonia (TPD-NH3 ) curves for (a) the bare zeolites, and (b) the impregnated zeolites.

The TPD-NH3 curve profiles of the vanadium oxide-impregnated 3.2. Characterization of the vanadium oxide-impregnated zeolites
samples (Fig. 4b) were significantly different, compared to those
of the bare zeolites (Fig. 4a), with the LT and HT peaks shifted to Quantitative phase analyses of the diffraction patterns for the
lower temperatures. The LT peak increased in intensity, while the impregnated VA, VB, VC, and VD catalysts (Fig. S1) revealed small
HT peak ended at around 450 ◦ C, instead of at the value of ca. 550 ◦ C quantities of crystalline V2 O5 species. In terms of zeolite crys-
observed for the bare zeolite. This could be explained by coverage of tallinity, the XRD patterns were similar to those for the bare
the strongest acid sites by vanadium oxide species. The quantities of zeolites. In the treatment of the XRD data, the crystallography infor-
desorbed NH3 were higher for all the samples containing vanadium mation file (CIF) from the International Zeolite Association (IZA)
oxide, reflecting a predominance of weak Lewis acid sites [50,51]. was adopted for the MFI zeolite, while the 29140 standard from the
ICSD database was used for V2 O5 . From comparison of the samples,
a crystalline V2 O5 mass fraction of 9.0% was obtained for the parent
 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28 25

VD the previous finding of greater dispersion of V2 O5 on the porous

zeolites.
VC 373 nm
VB
f(R) (arb. unit) 3.3. Catalytic activity
VA
335 nm
3.3.1. Step 1 − glycerol dehydration
The glycerol dehydration reaction occurs in two sequen-
tial reactions. In the first reaction step, glycerol dehydrates on
acid sites, giving rise to a water molecule and a very reactive
3-hydroxypropene intermediate that rapidly isomerizes via a keto-
enol rearrangement to 3-hydroxypropanal. The 3-hydroxypropanal
molecule then dehydrates, releasing water and acrolein. This reac-
tion sequence occurs at protonic Brønsted acid sites [12,30,54]. In
200 300 400 500
contrast to the performance of the acid sites of the zeolites, a plot
Wavelenght (nm) obtained using a reference alumina (which mainly contained Lewis
acid sites) as catalyst reflected low levels of activity and stability in
Fig. 5. UV–Vis diffuse reflectance spectra of samples containing vanadium oxide.
the reaction (Fig. 6, Table 2).
The catalytic results obtained for samples A, B, C, and D are
summarized in Fig. 6. Regardless of the sample or the reaction
catalyst, which was very close to the 10% mass percentage used in time, the main product of glycerol dehydration was acrolein. Glyc-
preparation of the samples, while V2 O5 mass fractions between 4% erol conversions of 88% and 87% were achieved using sample A
and 6.5% were obtained for the desilicated and acid treated samples. and desilicated sample B, respectively. The beneficial effect of the
This effect of dispersion on the support, with production of poorly micro-mesoporosity of sample B was shown by a decrease in cata-
crystallized V2 O5 particles, could be attributed to the creation of lyst deactivation from 19% to 7%.
mesopores and the capacity to cover the zeolite surface during the The acid treated samples (C and D) both showed glycerol con-
wet impregnation process. Comparison of the mesopore volumes version exceeding 90% during the first hour of reaction. However,
of the bare and impregnated zeolites (Table 1) confirmed the dis- there was more pronounced deactivation of the sample treated
persion of V2 O5 in the mesopores. The important influence of V2 O5 with hydrochloric acid (sample D), with deactivation values of 5.2%
dispersion on catalytic activity is discussed below. and 18.4% for samples C and D, respectively. These findings indi-
The diffuse reflectance spectroscopy measurements in the cated that both acids caused leaching of EFA (Table 1), and that
UV–Vis region (Fig. 5) showed distinct results for the impregnated HCl was less selective than oxalic acid, because it also removed
samples. The shapes of the spectra in this region were influenced part of the structural aluminum, damaging the zeolite structure and
by the local structures of the vanadium atoms. For example, the consequently affecting the activity of the catalyst in the reaction.
spectral range 333–500 nm provides information about the trans- As mentioned before, all the samples were more selec-
fer of low energy electrons between O and V5+ for vanadium atoms tive to acrolein, compared to other reaction products such as
in octahedral coordination [52,53]. On the other hand, the range acetaldehyde, 3-hydroxypropanal, and acetol (Fig. 7). The prod-
285–333 nm is related to vanadium in tetrahedral coordination. uct distributions obtained for sample A after 1 and 6 h (Fig. S2)
The green dashed rectangle highlighted in Fig. 5 indicates a region were indicative of decreased selectivity to acetaldehyde (2 carbon
influenced by the vanadium oxide environment. Bands at 335 nm atoms) and increased selectivity to 3-hydroxypropanal (3 carbon
and 373 nm were due to electron charge transfer of V5+ with ter- atoms). This suggested that decomposition of the primary molecule
minal or bridging oxygen anions, respectively. In other words, the occurred at the beginning of the reaction, and that this decomposi-
highlighted region was associated with terminal and supported tion was associated with coke formation and COx release [30]. The
vanadium oxides. The main peak occurred at 384 nm, although decomposition was related to contact of the glycerol molecule and
shifts to higher energy (or lower wavelength) could occur, depend- products with strong acid sites on the catalyst surface that became
ing on the degree of vanadium oxide coordination by water. Hence, progressively covered with coke, resulting in deactivation of the
the systematic increase in intensity of this band complemented strong acid sites. This was also reflected in the lower selectivity to

Glycerol dehydration Glycerol oxidehydration

100 100
Glycerol Conversion (%)
Glycerol Conversion (%)

80 80

60 60
VA - 3 mL/h
A VA
40 B 40
VB
C
20 VC
D 20
VD
Al2O3
0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Time (h) Time (h)

Fig. 6. Glycerol dehydration and oxidehydration of samples without (left plot) and with vanadium oxide (right plot). The filled and empty symbols correspond to flows of
3.0 mL/h and 6.0 mL/h of glycerol solution, respectively.
26 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28

Table 2
Reactant flow, conversion at time zero, catalyst deactivation, and %wt. of coke after 6 h on glycerol stream: (1) glycerol dehydration, and (2) glycerol oxidehydration.

Reaction Catalyst 10 wt.% glycerol Conversion at Catalyst % wt. of

solution flow (mL/h) time zero (%) deactivation (%) coke

(1) A 3 90 19.4 12.6

B 89 7.0 15.5
C 100 5.2 17.8
D 100 18.4 14.1
Al2 O3 65 68.5 55.5
(2) V-A 6 100 42.4 (0.0)a 12.8 (4.7)
V-B 89 22.4 (0.0) 12.2
V-C 93 20.9 (0.0) 10.4
V-D 100 20.9 (0.0) 14.4
a
In brackets are indicated the results of catalyst deactivation under a flow of 3.0 mL/h of glycerol solution.

A pockets created by desilication with NaOH and acid leaching, main-

60 taining the micropores available for glycerol [24].
B
Selec. Condens. Produc. (%)

50 C
D
3.3.2. Steps 1 and 2 − glycerol oxidehydration
40 As mentioned before, the two steps of glycerol oxidehydration
are (i) the dehydration of glycerol to acrolein on acid sites [2,30],
30
and (ii) the conversion of acrolein to acrylic acid on redox sites [55].
Therefore, in the impregnated catalysts, the catalytic sites respon-
20
sible for glycerol dehydration were the Brønsted acid sites, while
V2 O5 (better represented as V2 O5-x ) was responsible for acrylic
10
acid formation. The V2 O5 impregnation affected the acid sites of
0 the catalysts, increasing the contribution of the Lewis acid sites
and covering the strong Brønsted acid sites, as observed from the
60 VA
TPD-NH3 analyses. As reported previously, the amount of impreg-
VB
Selec. Condens. Produc. (%)

nated V2 O5 significantly affected the proportions of acid and redox

50 VC
sites, leading to substantial alterations in catalyst activity [26]. The
VD
main difference between the two groups of samples concerned cat-
40 alyst stability. No deactivation at all was observed for the samples
impregnated with vanadium oxide. In order to be able to observe
30 the deactivation and compare samples B, C, and D, the flow of glyc-
erol fed to the reactor was increased from 3 to 6 mL/h (see VA results
20
in Fig. 6, open and closed symbols). This change enabled detection
of 42.4% deactivation of the VA catalyst. Despite the deactivation
10
under this new experimental condition, the presence of vanadium
0 oxide reduced the loss of catalytic activity, compared to the bare
zeolites, demonstrating the multifunctional characteristics of these
l

al

l.

c.

e
e

l
n

d
na

to

nc
yd

lA
ci
ei

ci

cA
an

ce
A
ol

A
pa

la
eh

lly
op

catalysts in promoting conversion, with selectivity to acrylic acid

A
cr

oi

Ba
tic
ic
ro

ld

an
Pr
A

yl

ce
yp

ta

n
cr

op

and stability of catalytic activity over longer periods.

ce

bo
ox

Pr
A

ar
dr

The desilicated and impregnated sample (VB) showed decreased

C
hy
3-

deactivation in the dehydration reaction, compared to the VA cat-

Fig. 7. Selectivity towards condensable products and carbon balance in dehydration alyst, similar to the behavior observed for the bare zeolites. The
(top) and oxidehydration (bottom) of the catalysts during 1 h of reaction at 300 ◦ C coke was deposited in the formed mesopores, avoiding obstruction
and 350 ◦ C, respectively. of the micropores. In addition, the removal of EFA by the action of
oxalic acid, followed by impregnation (catalyst VC), resulted in a
further small reduction in deactivation, which decreased to 20.9%.
An interesting result was obtained for the sample treated with
acetic acid after 6 h of reaction (Fig. S2), since acetic acid is a prod- hydrochloric acid and impregnated with V2 O5 (sample VD). This
uct formed from acetaldehyde oxidation. Furthermore, and most zeolite and the reference alumina showed the highest levels of
importantly, there was increased selectivity to acrolein. deactivation in the dehydration of glycerol. However, in the oxide-
Different behavior was observed for samples B, C, and D, with hydration reaction, the VD catalyst presented deactivation very
greater amounts of carbonaceous deposits detected, compared close to that of the VC catalyst, because the action of vanadium
to the parent zeolite, despite lower deactivation. The thermo- oxide helped to reduce coke formation by oxidizing smaller coke
gravimetric analyses of the spent catalysts (Fig. S3) enabled molecules before oligomerization. Furthermore, the mesoporosity
establishment of relationships between the mesopore volumes and helped in dispersing vanadium oxide on the surface of the zeolite,
the amounts of carbonaceous compounds deposited on the cata- as confirmed by the UV–vis analysis that showed an increase in the
lysts, based on the weight loss. In the case of the purely microporous signal at 335–373 nm.
sample A, the coke was mainly deposited within straight and sinu- The distribution of the products was also affected by the treat-
soidal micropores, or in the entrances of pores, blocking the access ments applied to the samples (Fig. 7 and S3). Use of sample VA
of new glycerol molecules. The effect of coke on catalytic perfor- resulted in 5% of acrylic acid after the first hour on stream, with
mance was more noticeable, even for low coke contents. In porous selectivity to acrylic acid increasing to 25% after 6 h. These sam-
zeolites, the coke is preferentially located in the intra-crystalline ples contained an average of 12 wt.% coke, so the shift in the acrylic
 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28 27

Fig. 8. TG (a) and DTA (b) curves of the catalysts impregnated with vanadium oxide after 6 h of reaction.

acid yield could be explained by the coverage by coke of the more position of carbonaceous species on the catalyst surface. The high
deactivating sites. It should be noted that the acrylic acid selectivity porosity of this sample (Table 1) was likely to have contributed
showed a dependence on the glycerol flow. A flow of 3.0 mL/h was to the oxidation of coke precursors at low temperatures. For com-
not only beneficial to catalyst stability, but also enabled selectiv- parison, the thermogram of the most coked sample A is present in
ity to acrylic acid of 17% to be achieved in the first hour. Conditions Fig. 8a . The largest DTA signal (Fig. 8b) represents the two species
providing greater selectivity could therefore be obtained by adjust- of coke formed.
ing the glycerol contact time. The initial acrylic acid selectivity of As mentioned, despite the Lewis acid sites are formed in large
the VB catalyst (14%) was higher than obtained for the VA sample. amounts after V2 O5 impregnation, condition that could damage
Other products of glycerol oxidehydration are acetic acid and the catalytic activity as shown for ␥-Al2 O3 , the maintenance of
acetaldehyde, which both arise from the bifunctional nature of the Brønsted acid sites with the coverage of the strong Brønsted acid
catalyst. The acid sites crack the initial three carbon atoms of the sites by the impregnation was an important aspect for maintaining
glycerol molecule, resulting in acetaldehyde and one molecule of catalytic activity for a prolonged time.
COx . The presence of O2 then leads to the oxidation of acetalde-
hyde to acetic acid. The selectivities to these two products were
lower than 10% in the first hour of reaction, and decreased after
6 h due to deactivation of the sites. This behavior was the oppo- 4. Conclusions
site of that observed for the acrylic acid selectivity, suggesting that
the same sites were involved. The carbon balance between the glyc- The alkaline treatment of ZSM-5 zeolite with NaOH was effec-
erol feed and the outflowing condensable liquid products increased tive for obtaining a micro-mesoporous zeolite by partial disruption
with the glycerol flow and in the presence of V2 O5 . At glycerol solu- of the crystalline framework due to silicon removal. A portion of the
tion flow rates of 3.0 mL/h (for the bare zeolites) and 6.0 mL/h (for framework aluminum atoms in tetrahedral coordination was also
the impregnated samples), the average carbon balances were 30% removed, but due to their low solubility, these atoms remained
and 55%, respectively. on the zeolite as EFA (extra-framework aluminum). The removal
The activity results showed that the mesopores in the catalysts of EFA was achieved by subsequent acid treatment using either
assisted the mass transport of reactant and products, and there- hydrochloric acid or oxalic acid. The latter was more effective,
fore promoted the catalytic activity. In addition, the formation of because although hydrochloric acid acted to dissolve oligomeric
the extra porosity assisted in accommodation of the much less aluminum oxide species, it also attacked the crystalline zeolite
volatile polymeric coke molecules, which remained deposited in structure, while oxalic acid additionally acted as a chelating agent.
the intra-crystalline pockets within the zeolite mesopores, hence The micro-mesoporous zeolites were subsequently impregnated
maintaining access to the micropores. Two types of coke are formed with vanadyl sulfate solution and calcined to form supported V2 O5 .
in glycerol reactions: (1) polyaromatic compounds formed in rad- Spectroscopic measurements in the UV–Vis region indicated that
ical reactions, and (2) polyglycols formed by the condensation V2 O5 was well dispersed on the porous zeolites, due to the high
of neighboring glycerol molecules [26]. Previous characterization pore volumes. The dispersion of V2 O5 increased the quantity of acid
studies [20–22] have shown that both types are present in bare sites on the zeolites. The beneficial effects of micro-mesoporosity
zeolites, while the polyglycol type is suppressed following impreg- and V2 O5 dispersion could be seen in the performance of the cat-
nation with V2 O5 . alysts during glycerol conversion. The selectivities to acrolein and
The results of the thermogravimetric analyses of the spent cat- acrylic acid were enhanced, and the smaller amounts of coke led to
alysts are shown in Fig. 8. The DTA curves of the catalysts with reduced deactivation.
vanadium oxide showed two exothermic peaks (Fig. 8b). The first
peak, at lower temperature (∼350 ◦ C), was associated with the re-
oxidation (exothermic event) of vanadium species (V4+ to V5+ ) in
the case of the spent V2 O5 /ZSM-5 catalysts [20]. The second peak, at Acknowledgements
about 500 ◦ C, was due to polyaromatics (known as stubborn coke)
[8,56,57]. In the first temperature region, some samples showed a This work was supported by the Brazilian agencies CNPq
slight increase in mass (Fig. 8a), associated with the oxidation of V4+ (grants 473456/2012-5 and 401679/2013-6) and FAPESP (grants
to V5+ (highlighted in Fig. 8b) [26]. The VD sample showed a small 2013/10204-2, 2013/50023-7 and 2014/11952-5). The authors also
peak near 200 ◦ C, which could have been related to the decom- thank the Brazilian Synchrotron Light Laboratory (LNLS) in Camp-
inas for use of the XPD beamline (proposal XPD-17839).
28 L.G. Possato et al. / Catalysis Today 289 (2017) 20–28

Appendix A. Supplementary data [26] L.G. Possato, W.H. Cassinelli, T. Garetto, S.H. Pulcinelli, C.V. Santilli, L. Martins,
Appl. Catal. A: Gen. 492 (2015) 243–251.
[27] L.H. Vieira, K.T. Carvalho, E.A. Urquieta-González, S.H. Pulcinelli, C.V. Santilli,
Supplementary data associated with this article can be found, L. Martins, J. Mol. Catal. A: Chem. (2015), http://dx.doi.org/10.1016/j.molcata.
in the online version, at http://dx.doi.org/10.1016/j.cattod.2016.08. 2015.12.019 (in press).
005. [28] M.V. Rodrigues, C. Vignatti, T. Garetto, S.H. Pulcinelli, C.V. Santilli, L. Martins,
Appl. Catal. A: Gen. 495 (2015) 84–91.
[29] H. Decolatti, B. Dalla Costa, C. Querini, Microporous Mesoporous Mat. 204
References (2015) 180–189.
[30] L.G. Possato, R.N. Diniz, T. Garetto, S.H. Pulcinelli, C.V. Santilli, L. Martins, J.
[1] B. Katryniok, S. Paul, V. Belliere-Baca, P. Rey, F. Dumeignil, Green Chem. 12 Catal. 300 (2013) 102–112.
(2010) 2079–2098. [31] J.C. Groen, J.C. Jansen, J.A. Moulijn, J. Perez-Ramirez, J. Phys. Chem. B 108
[2] B. Katryniok, S. Paul, F. Dumeignil, ACS Catal. 3 (2013) 1819–1834. (2004) 13062–13065.
[3] C.A.G. Quispe, C.J.R. Coronado, J.A. Carvalho Jr., Renew. Sustainable Energy [32] J.C. Groen, L.A.A. Peffer, J.A. Moulijn, J. Perez-Ramirez, Microporous
Rev. 27 (2013) 475–493. Mesoporous Mat. 69 (2004) 29–34.
[4] M.A. Dasari, P.P. Kiatsimkul, W.R. Sutterlin, G.J. Suppes, Appl.Catal. A-Gen. 281 [33] J.C. Groen, T. Bach, U. Ziese, A. Donk, K.P. de Jong, J.A. Moulijn, J.
(2005) 225–231. Perez-Ramirez, J. Am. Chem. Soc. 127 (2005) 10792–10793.
[5] T. Miyazawa, Y. Kusunoki, K. Kunimori, K. Tomishige, J. Catal. 240 (2006) [34] J.C. Groen, L.A.A. Peffer, J.A. Moulijn, J. Perez-Ramirez, Chem. A Eur. J. 11
213–221. (2005) 4983–4994.
[6] J. Chaminand, L. Djakovitch, P. Gallezot, P. Marion, C. Pinel, C. Rosier, Green [35] J.C. Groen, J.A. Moulijn, J. Perez-Ramirez, J. Mater. Chem. 16 (2006) 2121–2131.
Chem. 6 (2004) 359–361. [36] J.C. Groen, W. Zhu, S. Brouwer, S.J. Huynink, F. Kapteijn, J.A. Moulijn, J.
[7] P.K. Vanama, A. Kumar, S.R. Ginjupalli, V.R.C. Komandur, Catal. Today 250 Perez-Ramirez, J. Am. Chem. Soc. 129 (2007) 355–360.
(2015) 226–238. [37] J. Perez-Ramirez, C.H. Christensen, K. Egeblad, C.H. Christensen, J.C. Groen,
[8] H. Atia, U. Armbruster, A. Martin, J. Catal. 258 (2008) 71–82. Chem. Soc. Rev. 37 (2008) 2530–2542.
[9] F. Wang, J.-L. Dubois, W. Ueda, Appl. Catal. A-Gen. 376 (2010) 25–32. [38] P. Mars, D.W. Van Krevelen, Chem. Eng. Sci. 3 (1954) 41–59.
[10] F. Cavani, S. Guidetti, L. Marinelli, M. Piccinini, E. Ghedini, M. Signoretto, Appl. [39] D. Verboekend, S. Mitchell, M. Milina, J.C. Groen, J. Perez-Ramirez, J. Phys.
Catal. B-Environ. 100 (2010) 197–204. Chem. C 115 (2011) 14193–14203.
[11] N.R. Shiju, D.R. Brown, K. Wilson, G. Rothenberg, Top. Catal. 53 (2010) [40] H. Canova, A. Fontoura, R. Neuenschwander, B. Diaz, C. Rodella, Upgrades to
1217–1223. the XRD1 beamline optics and endstation at the LNLS, J. Phys. Conf. Series IOP
[12] J. Deleplanque, J.L. Dubois, J.F. Devaux, W. Ueda, Catal. Today 157 (2010) Publ. (2014) (p. 012004).
351–358. [41] B. Toby, J. Appl. Crystallogr. 34 (2001) 210–213.
[13] Y.T. Kim, K.-D. Jung, E.D. Park, Appl. Catal. A-Gen. 393 (2011) 275–287. [42] A.F. Gualtieri, S. Ferrari, M. Leoni, G. Grathoff, R. Hugo, M. Shatnawi, G. Paglia,
[14] C.S. Carriço, F.T. Cruz, M.B. Santos, D.S. Oliveira, H.O. Pastore, H.M.C. Andrade, S. Billinge, J. Appl. Crystallogr. 41 (2008) 402–415.
A.J.S. Mascarenhas, J. Catal. 334 (2016) 34–41. [43] V. Paixao, R. Monteiro, M. Andrade, A. Fernandes, J. Rocha, A.P. Carvalho, A.
[15] W.Y. Pan, L. Huang, F. Qin, Y. Zhuang, X.M. Li, J.X. Ma, W. Shen, H.L. Xu, Acta Martins, Appl. Catal. A-Gen. 402 (2011) 59–68.
Phys. Chim. Sin. 31 (2015) 965–972. [44] R. Giudici, H.W. Kouwenhoven, R. Prins, Appl. Catal. A-Gen. 203 (2000)
[16] J.P. Lourenco, M.I. Macedo, A. Fernandes, Catal. Commun. 19 (2012) 105–109. 101–110.
[17] J.A. Cecilia, C. Garcia-Sancho, J.M. Merida-Robles, J.S. González, R. [45] J.T. Smith, S. Bailey, Acta Crystallogr. 16 (1963) 801–811.
Moreno-Tost, P. Maireles-Torres, Appl. Catal. A-Gen. 516 (2016) 30–40. [46] Z.M. Yan, M.A. Ding, J.Q. Zhuang, X.C. Liu, X.M. Liu, X.W. Han, X.H. Bao, F.X.
[18] A.S. Paula, L.G. Possato, D.R. Ratero, J. Contro, K. Keinan-Adamsky, R.R. Soares, Chang, L. Xu, Z.M. Liu, J. Mol. Catal. A-Chem. 194 (2003) 153–167.
G. Goobes, L. Martins, J.G. Nery, Microporous Mesoporous Mater. 232 (2016) [47] M.R. Apelian, A.S. Fung, G.J. Kennedy, T.F. Degnan, J. Phys. Chem. 100 (1996)
151–160. 16577–16583.
[19] L. Shen, H. Yin, A. Wang, Y. Feng, Y. Shen, Z. Wu, T. Jiang, Chem. Eng. J. 180 [48] S. Storck, H. Bretinger, W.F. Maier, Appl. Catal. A-Gen. 174 (1998) 137–146.
(2012) 277–283. [49] A.W. Chester, E.G. Derouane, Zeolite Charact. Catal. A Tutorial (2009).
[20] L. Shen, H. Yin, A. Wang, X. Lu, C. Zhang, F. Chen, Y. Wang, H. Chen, J. Ind. Eng. [50] M. Petras, B. Wichterlova, J. Phys. Chem. 96 (1992) 1805–1809.
Chem. 20 (2014) 759–766. [51] S.S.R. Putluru, A. Riisager, R. Fehrmann, Appl. Catal. B-Environ. 97 (2010)
[21] M. Dolores Soriano, P. Concepcion, J.M. Lopez Nieto, F. Cavani, S. Guidetti, C. 333–339.
Trevisanut, Green Chem. 13 (2011) 2954–2962. [52] X.T. Gao, I.E. Wachs, J. Phys. Chem. B 104 (2000) 1261–1268.
[22] A. Chieregato, F. Basile, P. Concepcion, S. Guidetti, G. Liosi, M. Dolores Soriano, [53] L.J. Burcham, G. Deo, X.T. Gao, I.E. Wachs, Top. Catal. 11 (2000) 85–100.
C. Trevisanut, F. Cavani, J.M. Lopez Nietoc, Catal. Today 197 (2012) 58–65. [54] B. Katryniok, S. Paul, M. Capron, C. Lancelot, V. Belliere-Baca, P. Rey, F.
[23] A. Chieregato, M. Dolores Soriano, F. Basile, G. Liosi, S. Zamora, P. Concepcion, Dumeignil, Green Chem. 12 (2010) 1922–1925.
F. Cavani, J.M. Lopez Nieto, Appl. Catal. B-Environ. 150 (2014) 37–46. [55] D. Vitry, Y. Morikawa, J.L. Dubois, W. Ueda, Appl. Catal. A-Gen. 251 (2003)
[24] L. Shen, H. Yin, A. Wang, X. Lu, C. Zhang, Chem. Eng. J. 244 (2014) 168–177. 411–424.
[25] A. Chieregato, M.D. Soriano, E. Garcia-Gonzalez, G. Puglia, F. Basile, P. [56] C.-J. Jia, Y. Liu, W. Schmidt, A.-H. Lu, F. Schueth, J. Catal. 269 (2010) 71–79.
Concepcion, C. Bandinelli, J.M.L. Nieto, F. Cavani, ChemSusChem 8 (2015) [57] Y.T. Kim, K.-D. Jung, E.D. Park, Appl. Catal. B-Environ. 107 (2011) 177–187.
398–406.