Hierarchical NaY Zeolites for Lactic Acid Dehydration to Acrylic Acid

Article abstract of DOI:10.1002/cctc.201600102

The industrial viability of heterogeneous catalysts for the gas-phase conversion of lactic acid to acrylic acid will strongly depend on their selectivity and durability. Here, we initially screened various aluminum-rich zeolites confirming that NaY is the most efficient catalyst for this reaction. This material was modified by sequential dealumination and alkaline treatment attaining a solid featuring a hierarchical distribution of micro- and mesopores, reduced Lewis acidity, and increased basicity owing to the presence of well-dispersed sodium ions interacting with external siloxy groups, as evidenced by in-depth characterization. These properties were crucial for determining higher selectivity (75 %) and minimizing the activity loss in a 6 h run. Remarkably, the catalyst gained in selectivity and stability upon reuse in consecutive cycles. This was ascribed to the depletion of the stronger basic sites and the clustering of sodium into oxidic particles upon the intermediate calcination.

Full text of DOI:10.1002/cctc.201600102

DOI: 10.1002/cctc.201600102
Full Papers
Hierarchical NaY Zeolites for Lactic Acid Dehydration to
Acrylic Acid
Giacomo M. Lari+, BegoÇa PuØrtolas+, Matthias S. Frei, Cecilia Mondelli,* and Javier PØrez-
Ramírez*[a]
The industrial viability of heterogeneous catalysts for the gas-
phase conversion of lactic acid to acrylic acid will strongly
depend on their selectivity and durability. Here, we initially
screened various aluminum-rich zeolites confirming that NaY is
the most efficient catalyst for this reaction. This material was
modified by sequential dealumination and alkaline treatment
attaining a solid featuring a hierarchical distribution of micro-
and mesopores, reduced Lewis acidity, and increased basicity
owing to the presence of well-dispersed sodium ions interact-
ing with external siloxy groups, as evidenced by in-depth char-
acterization. These properties were crucial for determining
higher selectivity (75%) and minimizing the activity loss in
a 6 h run. Remarkably, the catalyst gained in selectivity and sta-
bility upon reuse in consecutive cycles. This was ascribed to
the depletion of the stronger basic sites and the clustering of
sodium into oxidic particles upon the intermediate calcination.
Introduction
The replacement of petroleum with biomass-derived feed-
stocks has a growing impact in today’s energy mix and drives
the development of environmentally friendly processes for the
production of commodity chemicals. Among bio-based plat-
forms, lactic acid (LA) has gained an increasing importance in
recent years. LA is produced through sugar fermentation, but
a more scalable, greener, and less expensive two-step route
starting from waste glycerol has been recently proposed as an
alternative method to sustain the fast growth of its market
(+15% per year between 2013 and 2020).[1] LA can be trans-
formed into various derivatives, of which poly(lactic acid), a bio-
degradable and biocompatible plastic, and acrylic acid (AA), an
important bulk chemical (ꢀ4”106 taÀ1, 2500 $tonÀ1) used in
the manufacture of paint additives, adhesives, absorbents, and
textiles,[2] are the most relevant. The production of the latter
from LA only consists of an acid- or base-catalyzed dehydra-
tion.[3] Thus, an LA-based process could be economically and
environmentally more attractive than the current industrial
preparation, which relies on the partial oxidation of the costly
propylene. Still, the design of efficient heterogeneous catalysts
for the emerging process is a highly challenging task.[4] Firstly,
selectivity is an issue due to the competitive transformations
that can take place at the temperature required for the dehy-
dration of LA (573–673 K) (Scheme 1).[4,5] In this respect, the
Scheme 1. Different transformations observed in the gas-phase conversion
of LA over acidic and basic catalysts.
best solids among those investigated, that is, alkaline-earth
phosphates,[6] supported phosphate salts,[4] calcium hydroxy-
apatites,[7] inorganic salts,[8] and zeolites (mainly NaY)[9] modi-
fied with alkali or alkaline-earth metals or alkali phosphates,
have been found to contain both mild acid and basic sites.[6,8]
In the case of alkali phosphates-modified NaY, the formation of
an alkali lactate by the interaction of LA with the metal cations
appeared critical to enable the formation of AA.[5a,10] In spite of
these findings, the speciation and the chemical environment
of the active sites remain unclear. Secondly, rapid catalyst de-
activation has been observed even in short-term runs and, es-
pecially, for the more selective materials. This has driven some
researchers to assess the catalytic performance upon reaction–
regeneration cycles.[4,11] Interestingly, for alkali phosphates-
modified NaY zeolites, regenerated catalysts have displayed
a higher activity owing to an improved dispersion of the phos-
phate species and a stronger interaction between sodium lac-
tate and the zeolite.[4] However, the exact mechanism govern-
[a] G. M. Lari,+ Dr. B. PuØrtolas,+ M. S. Frei, Dr. C. Mondelli,
Prof. J. PØrez-Ramírez
Institute for Chemical and Bioengineering
Department of Chemistry and Applied Biosciences
ETH Zurich
Vladimir-Prelog-Weg 1, 8093 Zurich (Switzerland)
Fax: (+41)44-6331405
E-mail: cecilia.mondelli@chem.ethz.ch
jpr@chem.ethz.ch
[+] These authors contributed equally to this work.
ChemCatChem 2016, 8, 1507 – 1514
1507
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ing the different behavior of the material after regeneration is
still unknown. As the high deactivation rates have primarily
been attributed to the formation of carbonaceous deposits
and/or the strong adsorption of products arising from the
polymerization of LA,[5b] the introduction of a secondary meso-
porous network by the base-assisted removal of silicon frame-
work atoms appears attractive to generate materials with
longer lifetime. Indeed, post-synthetically modified zeolites ex-
hibiting improved accessibility have demonstrated superior in
traditional reactions afflicted by coking.[12] Additionally, previ-
ous works on high-silica faujasites have demonstrated the ef-
fectiveness of the alkaline treatment in simultaneously gener-
ating basic sites.[13]
Herein, we perform a screening of zeolites with different
framework topologies and Si/Al ratios in the dehydration of LA
to AA. Having identified Al-rich NaY as the best performer, we
investigate the development of hierarchical zeolites with tuned
acidic and basic properties from this material by the sequential
application of dealumination and base treatments. Thereafter,
we explore the effect of consecutive reaction-regeneration
cycles on the catalytic properties of the solid attaining the
highest AA yield. In-depth characterization of selected fresh
and used samples by Ar sorption, temperature-programmed
desorption of CO2 (CO2-TPD), 23Na and 27Al magic-angle-spin-
ning nuclear magnetic resonance (MAS NMR) and diffuse-re-
flectance infrared Fourier transform (DRIFT) spectroscopies,
and electron microscopy provide a detailed understanding of
the nature and location of the active sites, the role of meso-
porosity, and the changes in structure and basicity occurring
upon the repeated use and regeneration phases.
Figure 1. (a) Product selectivity at 30 min of reaction, and (b) evolution of
the LA conversion during 2 h on stream over sodium-form zeolites with dif-
ferent framework topologies and Si/Al ratios.
level dramatically decreased over most of the zeolites (Fig-
ure 1b). In view of the better retained activity compared to
the other selective material (LTL-2.9), FAU-3 (hereon referred to
as NaY) was selected for further investigation.
Post-synthetic modification of NaY by sequential acid and
base treatments
The parent NaY zeolite was dealuminated by treatment in an
acidic medium (aqueous H4EDTA, EDTA=ethylenediamine tet-
raacetate) attaining, after ion exchange to its sodium form, a hi-
erarchical analog coded NaY-DA. This material was further
modified by exposure to NaOH solutions of different alkaline
strength. These samples are labeled as NaY-DA-y, where y
refers to the molar NaOH concentration applied. The composi-
tional, textural, and acidic/basic properties of all these catalysts
are summarized in Table 1. As expected, the dealumination led
to the development of additional mesoporous surface area,
which was accompanied by a drop in crystallinity and micropo-
rous volume. In addition, the Si/Al ratio increased with the
consequent reduction of the amount of charge-compensating
Na+ cations. These parameters were not further significantly
modified by the alkaline treatment except for the mesoporosi-
ty, which developed to a greater extent, as highlighted by TEM
(Figure 2). The straight edges and clear crystalline fringes of
NaY corroborate its highly crystalline nature, whereas intracrys-
talline mesopores and indented edges are visualized for NaY-
DA-0.15.
Results and Discussion
LA dehydration performance of zeolites with different
framework topologies
To date, the study of LA dehydration to AA on zeolites has
been mostly limited to FAU materials. Furthermore, little data
exist comparing different catalysts under the same reaction
conditions. Accordingly, we firstly assessed the performance of
commercial zeolites in sodium form with distinct framework
topology and Si/Al ratio in the gas-phase conversion of LA at
623 K (Figure 1a). After 1 h on stream, FAU zeolites yielded AA
with variable selectivity: the highest value (62%) was mea-
sured for the sample featuring a Si/Al ratio of 3 (FAU-3) and
the lowest for that with a ratio of 385 (FAU-385). The selectivity
to the acetaldehyde (AD) byproduct followed a reverse trend,
being the highest (80%) for the Al-poor sample. MFI zeolites
were unselective irrespective of their Si/Al ratio, which was
varied between 15 and 1000. The AA selectivity was progres-
sively higher over BEA-12.5 (Si/Al=12.5), MOR-10 (Si/Al=10),
and LTL-2.9 (Si/Al=2.9), in that order, likely owing to the in-
creasing Al content in the materials. Indeed, a higher amount
of Lewis acid sites, which are believed to catalyze the dehydra-
tion of LA,[6,8] is typically present in zeolites with a low Si/Al
ratio. While the selectivity did not change substantially in the
subsequent hour on stream over all solids, the LA conversion
Negligible amounts of Brønsted-acid sites were detected by
FTIR spectroscopy of adsorbed pyridine for the parent and the
hierarchical zeolites, whereas NaY contained a remarkable con-
centration of Lewis acid sites, which reduced to less than half
upon the post-synthetic treatments. This is in line with the de-
crease of the number of Na+ cations, which have been identi-
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1508
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Characterization data of the NaY zeolite catalysts.
[b]
[c]
[c]
[d]
[e]
[f]
Catalyst
Si/Al[a]
(À)
Na[a]
[wt%]
Vpore
Vmicro
Smeso
SBET
cLewis
cbase
Cryst.[g]
[%]
Coke[h]
[wt%]
[cm3 gÀ1
]
[cm3 gÀ1
]
[m2 gÀ1
]
[m2 gÀ1
]
[mmolgÀ1
]
[a.u.]
NaY
NaY-DA
NaY-DA-0.05
NaY-DA-0.10
3.6
7.3
6.8
6.6
5.7
5.7
5.5
5.6
4.9
8.5
5.1
5.7
5.8
7.2
6.8
7.0
6.9
6.3
0.29
0.21
0.28
0.35
0.27
0.07
0.13
0.07
0.27
0.25
0.10
0.09
0.09
0.12
0.00
0.04
0.00
0.10
43
128
190
203
89
45
57
37
122
680
321
443
460
406
45
171
58
386
161
n.d.[i]
81
65
78
2.3
2.4
8.3
18.2
45.9
21.7
18.9
16.8
15.7
100
51
46
46
44
n.d.
n.d.
n.d.
38
20.4
12.3
10.3
4.4
10.3
n.d.
n.d.
n.d.
9.5
NaY-DA-0.15
NaY-DA-0.15 used 1x
NaY-DA-0.15 used 1 x calc.
NaY-DA-0.15 used 2 x
NaY-DA-0.20
60
n.d.
n.d.
45
[a] Si, Al, and Na contents determined by X-ray fluorescence (XRF) spectroscopy. [b] Volume adsorbed at p/p0 =0.99. [c] t-plot method. [d] BET method.
[e] Fourier transform infrared (FTIR) spectroscopy of adsorbed pyridine. [f] Integrated area of the CO2-TPD curves, normalized by the sample weight.
[g] Crystallinity determined by X-ray diffraction (XRD) analysis, ASTM standard D3906. [h] Carbon content determined by thermogravimetric analysis (TGA)
in the temperature range from 523 to 1023 K. [i] Not determined.
Figure 2. HAADF–STEM images, EDS mapping of Na (red) and Si (green), and TEM images (from left to right) of (a) NaY and NaY-DA-0.15 in (b) fresh and
(c) used forms.
fied as the Lewis acid sites in NaY zeolites.[14] 27Al MAS NMR
spectroscopy of the parent sample evidenced an intense and
very symmetric peak centered at 61.9 ppm, specific to tetra-
coordinated aluminum species in the zeolite framework, and
a weak signal at 7.3 ppm, attributed to octahedral aluminum
species (Figure 3a). Upon the acid treatment, the intensity of
the first peak decreased, reflecting the reduction in the alumi-
num content, and a broad contribution appeared at higher
fields (55.0 ppm), which is assigned to aluminum sites with dis-
torted tetrahedral geometry.[15] The spectrum of the sample re-
trieved after desilication did not show any further substantial
alteration. Based on this analysis, it can be concluded that oc-
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1509
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
groups present in defect-rich regions (i.e., the external surface
of the crystal or the mesopore walls), leading to stronger basic
sites similar to those observed for high-silica faujasites.[13b,c] The
presence of the latter species was corroborated by DRIFT spec-
troscopy, which evidenced the appearance of the silanol band
at 3735 cmÀ1 upon dealumination of the parent sample due to
the formation of defects in a protic environment and its partial
depletion upon the subsequent deprotonation of these groups
during the alkaline treatment (Figure 5).
Figure 3. (a) 27Al and (b) 23Na MAS NMR spectra of the fresh parent and hier-
archical NaY zeolites.
Figure 5. DRIFT spectra of the fresh parent and hierarchical NaY zeolites.
tahedral Al atoms only marginally contribute to the Lewis acid-
ity of these materials.
The chemical environment of the alkali metal atoms was fur-
ther analyzed by 23Na MAS NMR spectroscopy (Figure 3b). In
NaY zeolites, three distinct crystallographic sites occupied by
Na+ cations have been identified, that is, SII in the supercage,
SI’ in the sodalite cages, and SI in the center of the hexagonal
prisms, which exhibit characteristic resonances at À2.0, À5.0,
and À7.0 ppm, respectively.[16] Our NaY sample mainly exhibit-
ed sodium in the first two types of locations. Upon dealumina-
tion, the alkali metal atoms migrated from the supercages to
the sodalite cages. After the base treatment, a feature centered
at À3.4 ppm was observed, which could be related to Na+ cat-
ions at siloxy groups, as previously observed.[13b,c] Still, a further
reorganization of sodium between supercages and sodalite
cages cannot be excluded. In line with these spectroscopic evi-
dences, sodium mapping by energy-dispersive X-ray spectros-
copy (EDS) revealed a homogeneous dispersion of the metal in
both the NaY and NaY-DA-0.15 zeolites (Figure 2). Note that
the lower concentration of sodium in the modified sample re-
flects its lower amount in the material (Table 1).
The basicity of the catalysts was assessed by CO2-TPD
(Figure 4 and Table 1). The parent and the dealuminated zeo-
lites produced no significant desorption, confirming their negli-
gible basicity. In contrast, two major peaks between approxi-
mately 363 and 553 K became evident in the curves of the al-
kaline-treated zeolites, indicating the presence of basic sites
with a moderate character. The total amount of basic centers
in these samples followed a volcano-like trend with a maximum
for NaY-DA-0.15. Additionally, this solid also contained a higher
fraction of slightly stronger basic sites. As these evidences do
not correlate with the sodium loading, Na+ cations likely com-
pensate negative charges of different nature. A fraction is pres-
ent at ion exchange positions, to balance the framework
charge caused by the isomorphous substitution of silicon with
aluminum atoms. These sites have the same nature as the one
found in the parent NaY and possess very mild basicity. Anoth-
er part is believed to substitute the protons in the silanol
Catalytic performance of hierarchical NaY zeolites
The mesoporous NaY zeolites prepared through the post-syn-
thetic treatments were tested as catalysts in the gas-phase de-
hydration of LA under the same conditions as applied to the
screening of commercial zeolites presented above and their
performance was compared after 15 min on stream. In all
cases, AA and AD were the only products detected. The dealu-
minated sample attained an AA yield of only 10% (not shown).
This is explained by the presence of Brønsted acid sites formed
upon ion exchange of part of the sodium atoms with the pro-
tons released by the formation of the metal-EDTA chelate,
which catalyze the decarbonylation of LA to AD (Scheme 1).
When the dealuminated sample was ion exchanged to its
sodium form (NaY-DA), higher activity and selectivity were ob-
tained, which largely exceeded those of the parent NaY zeolite
Figure 4. CO2 desorption profiles of the fresh parent and hierarchical NaY
zeolites.
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1510
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tained by all sequentially dealuminated and alkaline-treated
samples follows the same trend observed for their amount of
basic sites. This hints that basic sites also lead to a more dura-
ble LA conversion than Lewis acid sites.
Notably, the modified zeolites appear less stable than other
materials previously reported,[3,5,8b,17] but their catalytic per-
formance has been evaluated at much higher weight hourly
space velocities (WHSV=0.1–1 hÀ1 versus 6 hÀ1 in the literature
and in this study, respectively) to simulate more industrially rel-
evant conditions and observe deactivation in a much shorter
time scale.
Catalyst deactivation, regeneration, and reuse
Coke formation and the strong adsorption of polymeric deriva-
tives have been claimed to encompass the main routes for cat-
alyst deactivation in the gas-phase dehydration of LA. Charac-
terization of the used catalysts by TGA confirmed the deposi-
tion of a conspicuous amount of coke, which was about
double in the starting NaY with respect to all modified samples
(Table 1). Based on these data, the faster deactivation of NaY-
DA compared to NaY is surprising. Hierarchically structured
porous materials have evidenced higher stability compared to
their microporous counterparts in catalytic processes in which
deactivation is caused by fouling owing to the possibility to
accommodate coke onto the mesopores.[12b] Our result can
only be explained assuming that the Lewis acid sites responsi-
ble for the desired reaction and coke formation are predomi-
nantly located in the micropores, which are present in a lower
quantity in the dealuminated sample. NaY-DA-0.15 was further
characterized to gain more understanding of the benefits of
the alkaline treatment. The porosity and acidity–basicity analy-
ses uncovered the full depletion of the micropores, the halving
of the mesopores and a reduction in the concentration of
basic as well as Lewis acid sites (Figure 7a and Table 1). 27Al
NMR MAS spectroscopy evidenced a broadening of the peak
originally centered at approximately 61.9 ppm, indicating a re-
arrangement of the tetrahedral aluminum centers (Figure 7b).
Based on these data, as only a minor fraction of the original
catalytic activity was lost upon use, it is likely that the advant-
age of the hierarchical material lies in the additional presence
of basic centers and their distinct preferential placement com-
pared to the Lewis acid sites, i.e., the active basic centers pres-
ent at the external surface of the zeolite crystals are only partly
affected by the formation of carbonaceous deposits promoted
by the Lewis acid sites mainly in the micropores.
Figure 6. (a) LA conversion and product selectivity after 15 min and (b) evo-
lution of the AA yield during 6 h on stream over the parent and hierarchical
NaY zeolites.
(Figure 6a). As the amount of basic sites for NaY-DA is similar
to that of NaY (Figure 4 and Table 1), the improved catalytic
performance is likely the result of the augmented external sur-
face area, which enhances the access of the reactant to the
active sites, suppressing its conversion through side reactions.
Catalytic testing of the sequentially dealuminated and alka-
line-treated samples evidenced a greater AA selectivity for cat-
alysts prepared using NaOH solutions with progressively
higher concentration reaching a maximum for the NaY-DA-0.15
zeolite. Harsher alkaline treatments led to an inferior per-
formance (NaY-DA-0.2). This volcano-like trend is in line with
the variation in amount of basic sites determined by CO2-TPD
and points to a major catalytic role of the basic sites formed
upon exchange of the protons of silanol groups by sodium in
LA dehydration with respect to the Lewis acid centers. This is
corroborated by the fact that the latter are more abundant in
the less performing parent material and present in rather simi-
lar amount in all dealuminated and alkaline-treated solids.
The stability of the parent and modified zeolites was evalu-
ated in a 6 h test. The data are expressed in terms of AA yield
(Figure 6b) after normalization by the values measured after
15 min of reaction for each catalyst and effectively indicate an
alteration in their activity since the selectivity remained almost
constant in all of the cases (Figure 6a). NaY continuously deac-
tivated and retained only 20% of the initial AA yield at the end
of the run. The same result was found for NaY-DA but this ma-
terial exhibited an even faster deactivation rate than the previ-
ous. NaY-DA-0.05, NaY-DA-0.10, and NaY-DA-0.20 deactivated
more rapidly than NaY within the first 2 h on stream but their
AA yield stabilized at different levels in the remainder of the
test. The most active and selective catalyst, NaY-DA-0.15,
better preserved its activity in terms of both deactivation rate
and overall loss in AA yield. The final fraction of AA yield re-
The nature of the fouled species was explored by DRIFT
spectroscopy (Figure 8). A band centered at approximately
1600 cmÀ1 was detected along with three additional absorp-
tions at 2974, 2938, and 2874 cmÀ1. The first spectral feature
corresponds to the stretching of the C=O group in sodium lac-
tate, whereas the second set of signals arises from the stretch-
ing of CÀH bonds, supporting the generation of adsorbed
polymeric species.[3] Inspection of the OÀH stretching region in
the spectra of the fresh and used material revealed a significant
depletion of the sharp silanol band at 3735 cmÀ1, suggesting
that at least part of the adsorbed species interacted with these
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1511
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
nally present in the large sodalite or supercages (À3.4 and
À1.1 ppm respectively), while the hexagonal prism seemed to
comprise a site offering more stability to the alkali metal cat-
ions.
The implementation of reaction–regeneration cycles has
been proposed as a strategy to extend the long-term use of
the catalysts. Herein, the activity of the best performing zeolite
(NaY-DA-0.15) was monitored over three consecutive runs be-
tween which the catalyst was thermally treated in air to simu-
late a typical regeneration procedure. Interestingly, the previ-
ously described slight activity loss in the first cycle became less
pronounced in the second and third tests and the AA selectivi-
ty was fully retained, or even moderately improved (Figure 9).
Figure 7. (a) CO2 desorption profiles and (b) 27Al and (c) 23Na MAS NMR spec-
tra of NaY-DA-0.15 in fresh, used, and regenerated forms.
Figure 9. (a) LA conversion and (b) AA selectivity over NaY-DA-0.15 upon
three subsequent catalytic cycles.
CO2-TPD analysis indicated that calcination of the sample re-
trieved after the first cycle provoked a further reduction of the
basic sites, especially of the stronger type (Figure 7a). The ab-
sence of the very intense and broad CO2 desorption signal ob-
served from 573 K, attributed to the decomposition of the or-
ganics present on the surface of the catalyst prior to the coke
burn-off, confirms the effectiveness of the regeneration with
respect to the fouling issue. In line with this, the signals attrib-
uted to adsorbed (poly)lactate species in the DRIFT spectrum
disappeared, and the band at 3735 cmÀ1 specific to terminal si-
lanols was restored (Figure 8). The 23Na MAS NMR spectrum of
the calcined sample evidenced a decrease in intensity for the
low-field portion of the signal and a strengthening of the
high-field contributions (Figure 7c). This indicates that the de-
pletion of sodium atoms interacting with carboxylic group
owing to the removal of the carbonaceous deposits led to the
formation of more sodium oxide particles. The latter is sup-
ported by the EDS map acquired for the material after the
third cycle, which shows well-defined sodium-rich regions (Fig-
ure 2c). The generation of these oxidic aggregates likely is the
origin of the beneficial changes in basicity determining the im-
Figure 8. DRIFT spectra of NaY-DA-0.15 in fresh, used, and regenerated
forms.
surface groups.[18] Furthermore, the broad band centered at
approximately 3400 cmÀ1 became more intense, indicating
a higher content of hydrogen-bonded hydroxyl groups, includ-
ing the contribution from both the surface groups–adsorbate
and adsorbate–adsorbate interactions.
These evidences were corroborated by 23Na MAS NMR spec-
troscopy (Figure 7c). Indeed, the depletion of the resonance at
À3.4 ppm and the appearance of a wide signal ranging from
À30 to 6 ppm agree with the interaction of sodium with car-
boxylate groups of lactate and polylactate species.[19] Addition-
ally, they indicate the clustering of the metal cations into
sodium oxide particles.[19] The sodium atoms that underwent
redistribution mostly comprised those having a lower interac-
tion with the framework basic oxygen atoms, i.e., those origi-
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1512
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
in the sodium form and used as received. Post-synthetic dealumi-
nation of NaY was performed at 373 K in a 0.15m solution of
H4EDTA (Fluka, >99%, 24 h, 15 cm3 per gram of zeolite). The re-
sulting slurry was filtered, washed with deionized water (3”,
ꢀ30 cm3 gzeoliteÀ1), and dried at 338 K for 16 h. A portion of this
sample was converted into the sodium form by ion exchange in
aqueous NaNO3 as detailed above, yielding the NaY-DA catalyst.
Another portion was treated in aqueous NaOH (Sigma–Aldrich,
97%, 0.05–0.2m, 338 K, 30 min, 30 cm3 gzeoliteÀ1) using an Easymax
102 setup (Mettler Toledo). After quenching in an ice bath and fil-
tering, the zeolites were washed with deionized water (3”,
ꢀ30 cm3 gzeoliteÀ1) and dried at 338 K for 16 h. The resulting cata-
lysts were labeled with the suffix y, which stands for the molar con-
centration of the NaOH solution. All catalysts were calcined in
static air as mentioned before.
proved catalyst performance upon cycling. Since the character-
ization data obtained for the catalyst after the first and the
second cycle are comparable (Figure 7 and 8), it seems likely
that the catalyst reaches equilibration already after the first re-
action-regeneration cycle.
Conclusions
In this study, we addressed the identification of zeolite-based
catalysts displaying enhanced stability in the gas-phase dehy-
dration of lactic acid to acrylic acid. Initially, we confirmed that
NaY zeolite is the best catalyst for a selective conversion of the
substrate among commercial aluminosilicates with different
framework topologies and Si/Al ratios. Thereafter, we demon-
strated that sequential dealumination and alkaline treatment
with a specific concentration of sodium hydroxide enable us to
produce a hierarchical NaY zeolite showing higher selectivity
and durability than the parent material. Based on in-depth
characterization, we could unravel the interplay between po-
rosity and site speciation at the basis of the superiority of the
modified catalyst. The starting NaY, only featuring Lewis acid
sites mostly in the micropores, suffers from rapid deactivation
through pore blockage with coke. In contrast, the dealuminat-
ed and base-treated sample, additionally containing a secon-
dary mesoporous network and basic sites therein located, re-
mains active and selective for longer time because fouling
takes place further away from the sites mostly responsible for
the desired catalytic process. The latter are found to likely
comprise siloxy groups counterbalanced by sodium cations.
Evaluation of this material in consecutive reaction-regeneration
tests revealed improved selectivity and stability upon cycling,
which was associated with a partial redistribution of sodium
from ion exchange positions into clustered oxide species and
with a decrease of the basicity. Our findings indicate the supe-
riority of basic over Lewis acid sites, providing new insights
into the parameters that govern the selectivity to acrylic acid
in the reaction under study. Additionally, they highlight the
benefits of an improved understanding of the phenomena re-
lated to deactivation and regeneration for the design of effi-
cient catalysts for lactic acid dehydration. For instance, to mini-
mize the formation and irreversible adsorption of polymeric
species on the zeolite surface, the utilization of methyl lactate
as an alternative substrate could be envisaged.
Catalyst characterization
The Na, Al, and Si contents in the samples were determined by
XRF spectroscopy using an EDAX Orbis Micro-XRF analyzer
equipped with a Rh source operated at a voltage of 35 kV and
a current of 500 mA. Powder XRD was performed using a PANalytical
X’Pert PRO-MPD diffractometer with Ni-filtered CuKa radiation (l=
0.1541 nm), acquiring data in the 5–708 2q range with a step size
of 0.058 and a counting time of 8 s per step. Ar sorption at 77 K
was conducted using a Micromeritics 3Flex analyzer. Prior to the
measurements, the samples were evacuated at 573 K for 3 h. CO2-
TPD was carried out using a Micromeritics Autochem II chemisorp-
tion analyzer coupled with a MKS Cirrus 2 quadrupole mass spec-
trometer. The catalyst (0.05 g) was pretreated in He flow
(20 cm3 minÀ1) at 823 K for 2 h, exposed to CO2 (50 pulses of 1 cm3
in a He flow of 10 cm3 minÀ1) at 323 K, and purged in He at the
same temperature for 1 h. Thereafter, the catalyst was heated up
to 973 K (10 KminÀ1) to monitor the desorption of CO2. 23Na and
27Al MAS NMR spectroscopy was conducted using a Bruker Avance
700 spectrometer operated at 185.2 and 182.4 MHz, respectively,
using 4 mm ZrO2 rotors spun at 10 kHz. For Na, spectra were ac-
quired using 2048 accumulations with a pulse length of 1 ms, a re-
cycle delay of 1 s, and solid NaCl (Sigma, 99.9%) as a reference
(d=7.21 ppm). In the case of Al, 512 scans were accumulated and
(NH4)Al(SO4)2 was used as a reference (d=0.00 ppm). Prior to the
analysis, the samples were dried in vacuum. FTIR spectroscopic
studies of adsorbed pyridine were performed using a Bruker IFS66
spectrometer equipped with a liquid-N2 cooled mercury cadmium
telluride (MCT) detector. Spectra were recorded in the range of
650–4000 cmÀ1 with a resolution of 4 cmÀ1 and co-addition of
32 scans. Zeolite wafers (ꢀ1 cm2, 20 mg) were evacuated
(10À3 mbar) for 4 h at 693 K, prior to adsorbing pyridine (Sigma Al-
drich, >99%) at room temperature. Gaseous and weakly adsorbed
molecules were subsequently removed by evacuation at 473 K for
30 min. The concentrations of Lewis acid sites were calculated
from the area of the bands at 1454 cmÀ1, using the extinction coef-
ficient eLewis =2.94 cmmmolÀ1. DRIFT spectroscopy was performed
using a Vertex 70 spectrometer equipped with a liquid-N2 cooled
MCT detector and a high-temperature cell. Prior to the measure-
Experimental Section
Catalyst preparation
FAU (CBV 720, Zeolyst International, bulk Si/Al=15; HSZ-390HUA,
Tosoh Corporation, bulk Si/Al=385), MFI (CBV 3024E, Zeolyst Inter-
national, bulk Si/Al=15; CBV 8014, Zeolyst International, Zeolyst
International, bulk Si/Al=40), BEA (CP814E, Zeolyst International,
bulk Si/Al=12.5), LTL (ZEOcat L, Zeochem, bulk Si/Al=2.9), and
MOR (CBV 21 A, Zeolyst International, bulk Si/Al=10) zeolites were
converted into their sodium form by ion exchange in aqueous
NaNO3 (0.1m (1m for MOR), 6 h, 298 K, 100 cm3 gzeoliteÀ1, 3 consecu-
tive treatments) followed by calcination in static air for 5 h at 823 K
(5 KminÀ1). FAU zeolite (NaY, Zeochem, bulk Si/Al=3) was obtained
ments, the samples were dried at 573 K in N2 flow (60 cm3 minÀ1
)
for 2 h. Spectra were recorded in the range of 600–4000 cmÀ1 with
a resolution of 2 cmÀ1. TGA of the used catalysts was conducted
using a Mettler Toledo TGA/DSC1 instrument. A catalyst sample of
15 mg was heated in the range of 323–1023 K (10 KminÀ1) under
air flow (40 cm3 minÀ1). The weight loss between 523–1023 K was
ascribed to the removal of deposited coke. TEM and high-angle an-
nular dark field (HAADF) STEM images and EDS elemental maps
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1513
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
were acquired using an FEI Talos instrument operated at 200 kV.
Powdered samples were deposited onto Mo grids. HAADF images
were collected before and after the EDS measurements to corrobo-
rate the absence of morphological changes.
Center for Energy Research (SCCER BIOSWEET, project number
KTI.2014.0116). Dr. S. Mitchell is acknowledged for the microscop-
ic analyses. The Micromeritics Grant Program is thanked for the
award of the 3Flex instrument.
Catalytic testing
Keywords: acidity · basicity · biomass · dehydration · zeolites
The gas-phase dehydration of LA to AA was studied at ambient
pressure in a continuous-flow fixed-bed reactor comprising: (i) a
mass flow controller for feeding N2 (PanGas, 99.999%), (ii) a syringe
pump for the injection of liquids, (iii) a tubular quartz micro-reactor
(i.d.=6 mm) heated in an oven, and (iv) a liquid-gas separator im-
mersed in an ice bath (Scheme 2). The catalyst (0.2 g, particle
size=0.2–0.4 mm) was loaded into the reactor, which was then
filled with 2 mm glass spheres to facilitate the evaporation of the
liquid feed. The system was heated at the reaction temperature
(623 K) in N2 flow (100 cm3 minÀ1) and allowed to equilibrate for
10 min. Thereafter, a 10 wt% aqueous LA solution (Fisher, 88%)
was fed at a rate of 0.2 cm3 minÀ1. Three consecutive catalytic tests
were performed with NaY-DA-0.15. The amount of catalyst and the
LA flow were constant in all the experiments to ensure a similar
contact time in all cases. After each catalytic test, part of the cata-
lyst was regenerated under an air flow (823 K, 5 h, 5 KminÀ1) prior
to the following test. For selected catalytic runs, a fraction of the
catalyst before and after the regeneration step was kept for char-
acterization. Liquid samples were periodically collected from the
condenser and analyzed by high-performance liquid chromatogra-
phy using an HPLC system (Merck LaChrome) equipped with an
HPX-87H column kept at 308 K and a refractive index detector. A
0.005m H2SO4 aqueous solution flowing at 0.6 cm3 minÀ1 was used
as the eluent. Calibration curves were obtained using LA (ABCR,
98%), AA (ABCR, 99%), and acetaldehyde (Sigma–Aldrich,
>99.5%). The conversion was calculated as the ratio between the
moles of LA reacted per mole of LA fed, the selectivity to i as the
mole of product i formed per mole of LA reacted, and the yield of
i as the product between LA conversion and selectivity to i. The
carbon balance was calculated as the ratio between the moles of
carbon in the liquid and the moles of carbon fed. The experimental
error, determined on the basis of three repetitions of selected
tests, was within 7%.
[1] a) M. Morales, P. Y. Dapsens, I. Giovinazzo, J. Witte, C. Mondelli, S. Papa-
dokonstantakis, K. Hungerbühler, J. PØrez-Ramírez, Energy Environ. Sci.
2015, 8, 558–567; b) G. M. Lari, C. Mondelli, J. PØrez-Ramírez, ACS Catal.
2015, 5, 1453–1461; c) R. K. Pazhavelikkakath Purushothaman, J. van
Haveren, D. S. van Es, I. Melian-Cabrera, H. J. Heeres, Green Chem. 2012,
14, 2031–2037; d) M. A. Abdel-Rahman, Y. Tashiro, K. Sonomoto, Bio-
technol. Adv. 2013, 31, 877–902.
[2] a) H. Danner, M. Ürmçs, M. Gartner, R. Braun, Appl. Biochem. Biotechnol.
1998, 70, 887–894; b) R. Datta, M. Henry, J. Chem. Technol. Biotechnol.
2006, 81, 1119–1129; c) E. Blanco, S. Loridant, C. Pinel in Reaction Path-
ways and Mechanisms in Thermocatalytic Biomass Conversion II (Eds.: M.
Schlaf, Z. C. Zhang), Springer, Singapore, 2016, pp. 39–62.
[3] M. Dusselier, P. Van Wouwe, A. Dewaele, E. Makshina, B. F. Sels, Energy
Environ. Sci. 2013, 6, 1415–1442.
[4] J. Zhang, Y. Zhao, M. Pan, X. Feng, W. Ji, C.-T. Au, ACS Catal. 2011, 1,
32–41.
[5] a) M. S. Tam, G. C. Gunter, R. Cracuin, D. J. Miller, J. E. Jackson, Ind. Eng.
Chem. Res. 1997, 36, 3505–3512; b) G. Näfe, M.-A. López-Martínez, M.
Dyballa, M. Hunger, Y. Traa, Th. Hirth, E. Klemm, J. Catal. 2015, 329,
413–424.
[6] E. Blanco, P. Delichere, J. M. M. Millet, S. Loridant, Catal. Today 2014,
226, 185–191.
[7] V. C. Ghantani, S. T. Lomate, M. K. Dongare, S. B. Umbarkar, Green Chem.
2013, 15, 1211–1217.
[8] a) R. E. Holmen, U.S. Patent 2859240, 1958; b) J. Peng, X. Li, C. Tang, W.
Bai, Green Chem. 2014, 16, 108–111.
[9] a) P. Sun, D. Yu, K. Fu, M. Gu, Y. Wang, H. Huang, H. Ying, Catal.
Commun. 2009, 10, 1345–1349; b) J. Yan, D. Yu, P. Sun, H. Huang, Chin.
J. Catal. 2011, 32, 405–411.
[10] D. C. Wadley, M. S. Tam, P. B. Kokitkar, J. E. Kackson, D. J. Miller, J. Catal.
1997, 165, 162–171.
[11] J. Zhang, Y. Zhao, X. Feng, M. Pan, J. Zhao, W. Ji, C.-T. Au, Catal. Sci. Tech-
nol. 2014, 4, 1376–1385.
[12] a) D. Verboekend, T. C. Keller, S. Mitchell, J. PØrez-Ramírez, Adv. Funct.
Mater. 2013, 23, 1923–1934; b) L. G. Possato, R. N. Diniz, T. Garetto, S. H.
Pulcinelli, C. V. Santilli, L. Martins, J. Catal. 2013, 300, 102–112.
[13] a) T. C. Keller, S. Isabettini, D. Verboekend, E. G. Rodrigues, J. PØrez-Ram-
írez, Chem. Sci. 2014, 5, 677–684; b) T. C. Keller, K. Desai, S. Mitchell, J.
ˇ
PØrez-Ramírez, ACS Catal. 2015, 5, 5388–5396; c) T. C. Keller, M. Polozij,
B. PuØrtolas, H. V. Thang, P. Nachtigall, J. PØrez-Ramírez, J. Phys. Chem. C
2016, 120, 4954–4960.
[14] G. Martra, S. Coluccia, P. Davit, E. Gianotti, L. Marchese, H. Tsuji, H. Hat-
tori, Res. Chem. Intermed. 1999, 25, 77–93.
[15] a) C. P. Grey in Handbook of Zeolite Science and Technology (Eds.: S. M.
Auerbach, K. A. Carrado, P. K. Dutta), CRC Press, Boca Raton, 2003; b) M.
Hunger, G. Engelhardt, J. Weitkamp, Microporous Mesoporous Mater.
1995, 3, 497–510.
[16] a) I. Hannus, I. Kiricsi, P. Lentz, J. B. Nagy, Colloids Surf. A 1999, 158, 29–
34; b) M. Feuerstein, M. Hunger, G. Engelhardt, J. P. Amoureux, Solid
State Nucl. Magn. Reson. 1996, 7, 95–103.
[17] T. Abe, S. Hieda, European Patent 0379691, 1990.
Scheme 2. Continuous-flow fixed-bed reactor used in this study.
[18] G. Larsen, E. Lotero, M. Mµrquez, H. Silva, J. Catal. 1995, 157, 645–655.
[19] a) H. Koller, G. Engelhardt, A. P. M. Kentgens, J. Sauer, J. Phys. Chem.
1994, 98, 1544–1551; b) H. Nakayama, D. D. Klug, C. I. Ratcliffe, J. A. Rip-
meester, J. Am. Chem. Soc. 1994, 116, 9777–9778.
Acknowledgements
This work was sponsored by the Swiss National Science Founda-
tion (Project Number 200020-159760) and the Swiss Competence
Received: January 26, 2016
Published online on March 21, 2016
ChemCatChem 2016, 8, 1507 – 1514
www.chemcatchem.org
1514
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim