Acetic acid conversion reactions on basic and acidic catalysts under biomass fast pyrolysis conditions

Article abstract of DOI:10.1016/j.mcat.2018.12.012

The objective of this study was to investigate the reaction pathways of acetic acid, the most abundant organic acid in bio-oils, over acidic and basic catalysts under typical biomass fast pyrolysis conditions. The pyrolysis/cracking experiments were performed in a fluidized bed reactor at 500 °C over a ZSM-5 zeolite formulation, as the acidic catalyst, and MgO as the basic catalyst. Both exhibit high activity (≥ 70 wt.% conversion) but distinctly different product selectivity. The primary reaction scheme on both acid and basic sites involves the ketonization of acetic acid to acetone, water and CO₂, via different elementary routes. However, the consecutive reactions of acetone differ significantly depending on the nature of the active sites, leading to different product selectivity over the two types of catalysts. Aromatics, phenols, light olefins and CO were the main by-products over ZSM-5, while MgO favored the formation of higher ketones. TPD studies on acetic acid and acetone-saturated catalysts indicated that acetone undergoes self-condensation and oligomerization reactions at temperatures > 300 °C on MgO, yielding mainly CO₂ and coke. On ZSM-5, both the direct dehydration of acetic acid and the secondary cracking of acetone condensation products play an equally important role, leading to the production of light olefins and small aromatic hydrocarbons. Phenols were also produced over ZSM-5 via acetone condensation and cyclization routes.

Full text of DOI:10.1016/j.mcat.2018.12.012

MolecularCatalysis465(2019)33–42
Contents lists available at ScienceDirect
Molecular Catalysis
journal homepage: www.elsevier.com/locate/mcat
Acetic acid conversion reactions on basic and acidic catalysts under biomass
fast pyrolysis conditions
A.C. Psarrasa, C.M. Michailofa, E.F. Iliopouloua, K.G. Kalogiannisa, A.A. Lappasa,
⁎
⁎
E. Heracleousa,b, , K.S. Triantafyllidisa,c,
a Laboratory of Environmental Fuels and Hydrocarbons (LEFH), Chemical Process and Energy Resources Institute (CPERI) – Centre for Research and Technology Hellas
(CERTH), 6th km Charilaou-Thermi Road, P.O. Box 60361, 57001 Thessaloniki, Greece
b School of Science & Technology, International Hellenic University (IHU), 14th km Thessaloniki – Moudania, 57001 Thessaloniki, Greece
c Department of Chemistry, Aristotle University of Thessaloniki, University Campus, P.O. Box 116, 54124 Thessaloniki, Greece
A R T I C L E I N F O
A B S T R A C T
Keywords:
The objective of this study was to investigate the reaction pathways of acetic acid, the most abundant organic
acid in bio-oils, over acidic and basic catalysts under typical biomass fast pyrolysis conditions. The pyrolysis/
cracking experiments were performed in a fluidized bed reactor at 500 °C over a ZSM-5 zeolite formulation, as
the acidic catalyst, and MgO as the basic catalyst. Both exhibit high activity (≥ 70 wt.% conversion) but dis-
tinctly different product selectivity. The primary reaction scheme on both acid and basic sites involves the
ketonization of acetic acid to acetone, water and CO2, via different elementary routes. However, the consecutive
reactions of acetone differ significantly depending on the nature of the active sites, leading to different product
selectivity over the two types of catalysts. Aromatics, phenols, light olefins and CO were the main by-products
over ZSM-5, while MgO favored the formation of higher ketones. TPD studies on acetic acid and acetone-sa-
turated catalysts indicated that acetone undergoes self-condensation and oligomerization reactions at
temperatures > 300 °C on MgO, yielding mainly CO2 and coke. On ZSM-5, both the direct dehydration of acetic
acid and the secondary cracking of acetone condensation products play an equally important role, leading to the
production of light olefins and small aromatic hydrocarbons. Phenols were also produced over ZSM-5 via acetone
condensation and cyclization routes.
Biomass fast Biomass
Bio-oil
Acetic acid
ZSM-5
MgO
Reaction pathways
1. Introduction
cases completely replaced by the in situ upgrading of biomass pyrolysis
vapours, i.e. as soon as they are formed via thermal pyrolysis, by cat-
Despite its great potential for the large-scale production of biofuels
and chemicals in an efficient and cost-effective manner, fast pyrolysis of
lignocellulosic biomass suffers from the low quality of the produced oil,
the so-called bio-oil or pyrolysis oil. Bio-oil typically contains sig-
nificant quantities of water and hundreds of oxygen-containing com-
pounds which render it unstable, corrosive and immiscible with pet-
roleum-based fuels [1]. Hence, further upgrading is required, targeting
to elimination of undesirable compounds, such as organic acids, or to
even deeper deoxygenation towards aromatics (BTX) or hydrocarbon
transportation fuels. Several downs-stream catalytic reactions/pro-
cesses have been investigated as means of bio-oil upgrading, including
pyrolysis/cracking, esterification, ketonization, hydrotreating/hydro-
deoxygenation, etc [2]. These processes can be combined or in some
alytic fast pyrolysis (in situ CFP). In in situ CFP a heterogeneous catalyst
is used instead of an inert heat carrier (i.e. silica sand) in mixtures with
the biomass particles within the pyrolysis reactor. Alternatively, a se-
parate catalytic reactor can be coupled with the fast pyrolysis reactor
(ex situ CFP) for the catalytic upgrading of the produced biomass pyr-
olysis vapours [2,3]. CFP is a relatively simple and efficient process that
avoids the use of hydrogen and aims to promote depolymerization of
the initially formed oligomers and to tune product selectivity depending
on the nature of the catalyst used.
Over the last two decades, numerous types of catalysts have been
investigated as candidates for the CFP process, such as zeolites, meso-
porous aluminosilicate materials with uniform pore size distribution
(MCM-41, MSU, SBA-15, etc.), hierarchical zeolites with combined
⁎ Corresponding authors at: Chemical Process and Energy Resources Institute (CPERI) – Centre for Research and Technology Hellas (CERTH), 6th km Charilaou-
Thermi Road, P.O. Box 60361, 57001 Thessaloniki, Greece (E. Heracleous) and Department of Chemistry, Aristotle University of Thessaloniki, University Campus,
P.O. Box 116, 54124 Thessaloniki, Greece (K.S. Triantafyllidis).
E-mail addresses: eheracle@cperi.certh.gr (E. Heracleous), ktrianta@chem.auth.gr (K.S. Triantafyllidis).
https://doi.org/10.1016/j.mcat.2018.12.012
Received 3 November 2018; Received in revised form 13 December 2018; Accepted 17 December 2018
2468-8231/©2018ElsevierB.V.Allrightsreserved.

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
micro- and mesoporosity, metal doped zeolites as well as various oxides
with acid-base properties [1,4,5]. Focus has been largely placed on
catalysts with acidic properties that favor dehydration, decarbonyla-
tion, cracking and aromatization reactions towards BTX, with zeolite
ZSM-5 being the most widely studied material [1,4,6]. Acid catalysts,
however, typically suffer from fast deactivation due to acid site poi-
soning and extensive coking that leads to pore blockage and loss of
reactivity and selectivity to the desired products [7]. The application of
basic materials in CFP has been much scarcer. Studies from our group
have highlighted the promising potential of materials with basic
properties. In a screening study of various acidic and basic catalysts for
the in-situ catalytic upgrading of biomass pyrolysis vapors in a fixed
bed reactor, it was shown that basic catalytic materials lead to high CO2
yields and produce a bio-oil with decreased acid concentration and high
content of ketones via ketonization and aldol condensation reactions
[8]. In a subsequent study it was shown that MgO catalysts, derived
from natural magnesite mineral, exhibit, despite the lack of acidity,
comparable activity in the catalytic pyrolysis of biomass to that of a
diluted, industrial ZSM-5 zeolite catalyst, with enhanced deoxygenation
due to acids ketonization and CeC coupling reactions [9]. Pütün [10]
employed MgO as catalyst for the fast pyrolysis of cotton seeds and
reported the production of bio-oil with much lower oxygen content
than the thermal bio-oil. Similar results were also published by Lin et al.
using CaO as catalyst for the pyrolysis of white pine biomass [11]. MgO
has also been considered as a suitable dopant for zeolite-based biomass
pyrolysis catalysts. The incorporation of 10 wt.% MgO in lamellar and
pillared ZSM-5 zeolites allowed tailoring the zeolite activity to avoid
the excessive cracking of bio-oil in the in-situ catalytic upgrading of
eucalyptus woodchips fast pyrolysis vapors. This in turn resulted in
higher yield of organics and decreased the formation of undesired
polyaromatic hydrocarbons and coke [12]. Analogous findings were
reported also for MgO-doped hierarchical ZSM-5 and Beta zeolites. The
enhanced performance of the MgO-loaded catalysts was attributed to
the adequate balance of Lewis acid and basic sites [13].
impurities) produced via beneficiation and rotary kiln calcination
(800–1200 °C) of natural magnesite minerals, kindly provided by Gre-
cian Magnesite S.A. [9]. Prior to characterization and testing the cat-
alysts were calcined at 500 °C for 2 h in air.
Powder X-ray diffraction (XRD) was applied to verify the crystal
structure of ZSM-5 zeolite and MgO using a Siemens Diffractometer
D5000 equipped with Cu Kα X-ray radiation and a curved crystal gra-
phite monochromator operating at 45 kV and 100 mA; counts were
accumulated in the range of 5-75° 2θ every 0.02° (2θ) with counting
time 2 s per step. The crystal size (L) was also determined by measuring
the width at half maximum, β1/2, of the MgO’s main peak at 2θ = 42.9°
as an input to the Scherrer equation (L [nm] = K λ/β1/2 cosθ).
N2 adsorption-desorption experiments at −196 °C were performed
on an Automatic Volumetric Sorption Analyzer (Autosorb-1,
Quantachrome) for the determination of surface area (BET method),
total pore volume at P/Po = 0.99, micropore volume (t-plot method),
and pore size distribution (BJH method) of the samples that were
previously outgassed at 150 °C for 16 h under 5 × 10−9 Torr vacuum.
Fourier-Transform Infrared (FT-IR) spectroscopy experiments,
combined with in situ adsorption of pyridine were performed on a
Nicolet 5700 FTIR spectrometer (resolution 4 cm−1) using the OMNIC
software for the determination of the Brønsted (band at 1545 cm−1
attributed to pyridinium ions) and Lewis (band at 1450 cm−1 attributed
to pyridine coordinated to Lewis acid sites) type acid sites of the cat-
alysts. Data processing was carried out via the GRAMS software and the
quantitative determination of acid sites was performed by adopting the
molar extinction coefficients proposed by Emeis [18]. The samples were
initially outgassed at 450 °C under high vacuum (10−6 mbar) for 1 h,
followed by adsorption of pyridine (added in pulses) for 1 h at 1 mbar
equilibrium pressure. All spectra were collected at 150 °C in order to
eliminate the possibility of pyridine condensation.
Basicity was determined by CO2-temperature programmed deso-
rption (TPD-CO2). In a typical experiment, 0.2 g of the sample were
loaded in a fixed bed quartz reactor and pretreated at 600 °C in He for
1 h, followed by cooling to 80 °C under He flow and subsequent treat-
ment with a flow of 40% CO2/He for 1 h at 80 °C. Flushing with pure He
at 80 °C for 3 h was then applied to remove the physisorbed CO2. TPD
analysis was carried out from 80 to 600 °C at a heating rate of 10 °C/
min and a He flow rate of 50 cm3/min. The composition of the exit gas
was monitored online by a quadrupole mass analyzer (Omnistar,
Balzer). Quantitative analysis of the desorbed CO2 was based on the
fragment m/z = 44.
Despite the intense research efforts, biomass pyrolysis and bio-oil
upgrading still remain challenging, owing mainly to the complex nature
of bio-oil. Depending on the feedstock, pyrolysis conditions and cata-
lyst, biomass pyrolysis oil consists of oxygenates corresponding to dif-
ferent families: alcohols, phenols, aldehydes, ketones, acids, and esters
[14]. The determination of optimum conditions for the catalytic
transformation of bio-oil requires detailed knowledge of the reactivity
and the mechanistic pathways followed during the upgrading reactions.
According to a recent review on the mechanism of thermal and catalytic
lignocellulosic biomass pyrolysis [15], further research is needed to
explore the fundamental reaction mechanisms of the reactants on the
catalyst’s active sites. To this end, model compounds, being present in
biomass fast pyrolysis oils, can be very useful in understanding and
identifying the reaction steps involved in the upgrading of bio-oils.
In this context, the present study aims to elucidate reaction path-
ways of acetone conversion, a representative model compound of the
carboxylic acids contained in bio-oil, over acidic and basic catalysts
under typical biomass fast pyrolysis conditions. Tests were conducted
over typical catalysts with well-defined acidic and basic properties,
namely ZSM-5 and MgO, in a small-scale fluidized bed reactor unit at a
wide range of residence times. Temperature-programmed desorption
(TPD) studies on acetic acid and acetone-saturated samples were also
undertaken in an effort to elucidate the stepwise formation and trans-
formation of primary and secondary products.
2.2. Acetic acid conversion experiments
The pyrolysis experiments were carried out in a fully automated
bench-scale microactivity (MAT) test unit, utilizing a fluidized bed re-
actor configuration. Details on the design of the unit can be found in
previous communications [19]. All experiments were conducted at
constant temperature of 500 °C. Variation of the conversion level was
attained by modifying the catalyst-to-reactant (acetic acid) mass ratios,
corresponding to a WHSV range of 16 to 48 h−1. In all tests, the liquid
and gaseous products were separated, collected and analyzed.
Gas products were analyzed on a gas chromatograph (HP-6890)
equipped with two thermal conductivity detectors (TCD). The liquid
products were analyzed on a HP5890II GC with a polar DB-WAX
column and equipped with a flame ionization detector (FID). The
qualitative and quantitative determination of the compounds was per-
formed by means of calibration lines using external standard solutions
of acetone, cyclopentanone, hydroxyacetone, acetic acid and phenol (in
order of elution) at five concentration levels. The liquid samples were
also analyzed by GC–MS (5975C Mass spectrometer with 7890 Gas
Chromatograph using a HP-5MS 5% Phenyl Methyl Silox column) in an
attempt to identify the compounds related to unassigned peaks ob-
served in the GC-FID chromatograms. To this end, the aqueous samples
were dispersed in CH2Cl2 at proper dilution ratio. In order to avoid the
2. Experimental
2.1. Catalysts and catalyst characterization
The catalysts used were a commercial equilibrium ZSM-5 zeolite
formulation diluted with silica-alumina (containing 30 wt. % crystalline
zeolite) [16,17] and a MgO material (with < 2% SiO2 and CaO
34

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
formation of emulsions and the introduction of water in the apolar
GC–MS column, Na2SO4 was added to the mixture. The samples were
filtered through a 0.2 μm nylon filter prior to injection in the GC–MS.
The water content of the liquid products was quantified via volumetric
Karl Fischer titration (ASTM E203). Finally, the amount of coke de-
posited on the catalyst surface was measured on a LECO CS400 ele-
mental analyzer.
that may act as Lewis acid sites [16,17,22]. With regard to the MgO
catalyst, it shows an appreciable surface area, being attributed solely to
textural/intercrystal meso/macroporosity [9]. As expected, it exhibits
only basic surface properties and no acidity.
3.2. Acetic acid conversion under biomass fast pyrolysis conditions
The performance of the two investigated catalysts in the fast pyr-
olysis of a common woody lignocellulosic biomass, i.e. beech wood
sawdust, was typical of an acidic ZSM-5 zeolite and a basic MgO
[9,16,17]. Representative data on product yields and selectivity to se-
lected types of compounds is shown in Table S1 (Supporting informa-
tion). It can be clearly seen that the acidic ZSM-5 catalyst induces the
deoxygenation of the organics in bio-oil via dehydration, dec-
arbonylation and decarboxylation reactions, while it favors the pro-
duction of aromatics. On the other hand, the basic MgO causes also
substantial deoxygenation of the bio-oil, but with negligible formation
of aromatics and increased formation of ketones. The CO2/CO ratio is
also significantly higher with MgO compared to the performance of the
acidic zeolite. Furthermore, the acids in the organic bio-oil, which
comprise mainly of acetic acid derived from the acetyl groups in
hemicellulose as well as from the decomposition of sugars [23], de-
crease on both catalysts (Table S1). It has been previously suggested,
without direct experimental proof, that carboxylic acids may contribute
to the formation of aromatics on ZSM-5 and the formation of ketones on
MgO and other basic catalysts [24]. These differences in product yields
and selectivities imply the different reaction pathways that are being
catalyzed by the acidic ZSM-5 and basic MgO during the in situ up-
grading of biomass pyrolysis vapors.
2.3. Temperature-programmed desorption (TPD) studies
Temperature-programmed desorption (TPD) studies of acetic acid
and acetone-saturated catalysts were performed on a home-made TPD
unit comprising of a quartz fixed bed reactor, heating and gas flow
controllers and on-line quadrupole mass analyzer (Omnistar, Balzers).
The fresh ZSM-5 and MgO samples (100 mg) were first saturated with
either acetic acid or acetone (2 cm3) and then dried overnight in an
oven at 40 °C. The saturated catalyst samples were placed in the fixed
bed reactor and were heated to 800 °C with a heating rate of 5 °C/min
under a flow of He (40 cm3/min). Mass spectra were recorded using a
SEM amplifier operating at 1400 V and an ionization potential of
−50 eV. The signals of the following mass-to-charge (m/z) were re-
corded: 60 (CH3COOH), 58 (CH3OCH3), 44 (CO2), 18 (H2O), 56 (i-
C4H8), 41 (C3H6) and 26 (C2H4). The intensity of the signal for each
reactant/product was normalized with respect to helium and corrected
for contributions from other components, based on the cracking pat-
terns of each molecule.
3. Results and discussion
3.1. Catalyst properties
Thus, in this work, we investigated the conversion of acetic acid on
ZSM-5 and MgO under representative biomass fast pyrolysis conditions
in a fluidized bed reaction unit at 500 °C in a WHSV range of 16–48
h−1. Fig. 1 shows the conversion of acetic acid as a function of space
velocity over the two materials. High conversion levels ranging be-
tween 70–90 wt.% were recorded over both catalysts, with ZSM-5 ap-
pearing slightly more active than MgO. However, the turnover fre-
quencies (TOFs) normalized per active site (considering all acid and
basic sites for ZSM-5 and MgO respectively) and determined at iso-
conversion show an order of magnitude higher reactivity for ZSM-5 (see
inset of Fig. 1). Although the accessibility to the active sites might differ
over the two materials, the superior reactivity of the acidic sites is in-
disputable and it is clear that the various reaction steps proceed much
faster than on basic sites. This has been further verified by comparing
the performance of the commercial ZSM-5 formulation with that of a
pure ZSM-5 zeolite with higher total acidity as well as higher ratio of
Brønsted to Lewis acid sites, with the pure zeolite being significantly
more reactive compared to the commercial equilibrium ZSM-5 for-
mulation of the present study (Fig. S-1, Supporting information).
The two catalysts, i.e. commercial acidic ZSM-5 zeolite formulation
and basic MgO, have been previously investigated by our group in
biomass fast pyrolysis [9,16,17]. Detailed physicochemical character-
ization has been reported in these studies, while a summary of the most
important properties is also shown in Table 1. The commercial ZSM-5
catalyst exhibits relatively low surface area and micropore volume,
compared to a pure crystalline ZSM-5 zeolite, as it is diluted with
amorphous silica-alumina matrix with enhanced meso/macroporosity
(average pore size of 4 nm). The number of Brønsted acid sites and the
Brønsted to Lewis acid sites ratio are also significantly lower compared
to those of a pure H-ZSM-5 zeolite due to the dilution and the nature of
silica-alumina which possesses mainly Lewis acid sites [20,21]. In ad-
dition, being in equilibrium state, i.e. having been subjected to multiple
cycles of biomass pyrolysis - regeneration in a recirculating fluid bed
reactor, the high temperature steaming conditions during regeneration
induce dealumination of the zeolitic framework, thus lowering the
Brønsted acid sites and generating extra-framework aluminum species
Table 1
Physicochemical characteristics of the ZSM-5 and MgO catalysts.
Catalysts
Porosity characteristics (N2 porosimetry)
Acidity
Basicity
(FTIR - pyridine)
(TPD-CO2)
BET areab
(m2/g)
Micropore volumec
(cm3/g)
Meso/macropore volumed
(cm3/g)
Meso-pore size
(nm)e
Brønsted sites
(μmol/g)
Lewis
(μmol/g)
sites
(μmol/g)
ZSM-5a
MgO
138
64
0.037
–
0.071
0.360
4.0f
36.5
–
18.1
–
∼0
244
28.9
a
Commercial equilibrium ZSM-5 zeolite formulation in silica-alumina matrix (contains ∼30 wt. % crystalline zeolite).
b
c
d
e
f
Multi-point BET method.
t-plot method.
Difference of total pore volume (at P/Po = 0.99) minus the micropore volume.
BJH analysis using adsorption data.
Due to the mesoporosity of the silica–alumina matrix, in addition to the typical microporous structure of ZSM-5 zeolite with ∼0.55 nm channels.
35

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
Fig. 1. Acetic acid conversion as a function of WHSV at 500 °C (inset: TOF at 74 wt.% iso-conversion).
Fig. 2. Product selectivities at 74 wt.% acetic acid iso-conversion.
Selectivity to the different liquid, gaseous and solid reaction pro-
except acetone, water and CO2, is considerably different over the two
catalysts. In the presence of ZSM-5, C2-C4 light olefins and CO are
formed as by-products in the gas phase. In addition, GC–MS analysis of
the organic liquid phase revealed, besides acetone, a considerable
amount of phenolic and aromatic compounds, such as methyl- and di-
methyl- phenols, xylenes, alkyl-benzenes etc. (Fig. 3 and Table S2 in
Supporting information). On the contrary, over basic MgO the non-
gaseous major by-products consist of higher ketones (pentenones,
hexenones etc.), coke and a very small amount of phenols and aromatics
(almost exclusively tri-methyl benzenes). It should be mentioned that
coke production is slightly higher over MgO than ZSM-5 (Fig. 2). These
indications clearly point out that the reaction pathways over acidic
catalysts are different in comparison to those on basic catalysts, being
numerous and more complex on the former.
ducts at constant conversion is presented in Fig. 2. The main liquid
products over both catalysts (MgO and ZSM-5) are acetone and water,
with the main gaseous product being CO2. Thus, the reaction pathway
that prevails on both acid and basic materials is the ketonization of
acetic acid with production of acetone, water and CO2 as primary
products. When the equilibrium ZSM-5 formulation of the present work
is compared to the pure ZSM-5 zeolite (Fig. S-1, Supporting informa-
tion), it can be seen that the selectivity (or yield) to the various pro-
ducts has similar trend for the two catalysts but is different (case of
CO2) or changes more steeply (case of acetone and water) with the
increase of conversion, thus indicating differences either in the intrinsic
reactivity of their active (acid) sites and/or diffusional and confinement
effects of microporosity (pure ZSM-5) versus the combined micro- and
meso/macroporosity (due to the silica – alumina matrix of the ZSM-5
formulation).
The yield of the major products as a function of acetic acid con-
version for the two catalysts is shown in Fig. 4. Such graphs/correla-
tions are typical in the field of catalytic pyrolysis/cracking of biomass
or petroleum fractions, depicting the effect of all reaction parameters
on product yields and conversion. In the present work, conversion was
varied by progressive increase of WHSV (via increase of catalyst to feed
ratio). Acetone production monotonously decreases with increasing
conversion, confirming that it undergoes successive reactions leading
ultimately to the production of coke, H2O and CO2 over both basic and
Stoichiometrically, the ketonization reaction leads to an iso-mole-
cular mixture of the three components, corresponding to a weight based
selectivity of 48.3 wt.% to acetone, 15 wt.% to H2O and 36.7.% to CO2.
However, the results shown in Fig. 2 deviate from this stoichiometry;
the acetone selectivity is much lower while the concentration of H2O is
notably higher, evidencing the occurrence of extensive secondary re-
actions of acetone. The distribution of the rest of the products, i.e.
36

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
Fig. 3. Concentration (GC–MS peak area, %) of major by-products present in the organic phase of the liquid products, at 74 wt.% acetic acid iso-conversion.
Fig. 4. Yields of acetone and H2O (top left), coke (top right), CO2 (bottom left) and other gases (bottom right) as a function of acetic acid conversion at 500 °C.
acid sites, in addition to CO and light olefins that are only formed on the
acid sites of ZSM-5. Comparatively, MgO appears more selective to
acetone than ZSM-5; at the same time, the decreasing rate of acetone
yield is also slightly lower than that with ZSM-5. This confirms the
lower reactivity of MgO towards further acetone conversion. Despite
the more extensive decomposition reactions on ZSM-5 (as evidenced by
the formation of light olefins whose concentration increases with con-
version), the coke yield is greater on MgO.
3.3. Temperature-programmed desorption (TPD) studies
The ZSM-5 and MgO catalyst samples were saturated with acetic
acid or acetone and were then heated in a temperature programmed
manner under helium flow in order to obtain information on the pri-
mary and secondary transformations that occur on the catalyst surface.
Fig. 5 shows the TPD profile (MS spectra) obtained for the acetic-acid
saturated MgO sample. At temperature < 150 °C, water and small
37

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
Fig. 5. Temperature-programmed desorption profile of acetic acid-saturated MgO.
amounts of acetic acid were detected that were likely due to the des-
orption of catalyst’s moisture and evaporation of physisorbed acetic
acid, respectively. The main transformation reactions occur in the
temperature range of 275–400 °C and lead almost exclusively to the
production of acetone, CO2 and water. No acetic acid desorption was
observed at the temperature range of ca. 150 to 400 °C, pointing out to
strong interaction of the carboxylic acid with the MgO surface that
inhibits its evaporation and allows its transformation on the catalyst
surface sites. The evolved gaseous product exhibited a doublet with
maxima at ˜305 °C and 330 °C. Although all products were detected at
both temperatures, the evolution of acetone was more pronounced at
305 °C, whereas formation of CO2 prevailed at 330 °C indicating the
simultaneous occurrence of secondary acetone conversion reactions
that potentially have higher activation energy than the ketonization
reaction. This observation is supported by the TPD profile obtained on
the acetone-saturated MgO, shown in Fig. 6, which is dominated by
large production of H2O and CO2 at ˜ 335 °C. MgO has been reported to
exhibit promising performance in the gas phase aldol condensation of
acetone for the synthesis of mesityl oxide and isophorone [25–27]. Self-
aldol condensation of acetone initially forms diacetone alcohol which
subsequently dehydrates to form mesityl oxide and water. At > 200 °C,
aldol condensation of acetone was reported to show more complex
reaction pathways to yield acyclic, cyclic and aromatic trimers [28].
Therefore, the large production of water observed at 335 °C in the TPD
of the acetone-saturated sample could be associated with self-con-
densation and oligomerization reactions of acetone. This temperature
coincides with that of the high-temperature peak in the TPD of the
acetic acid-saturated MgO and can thus be attributed to secondary
conversion of acetone that primarily forms on MgO from the carboxylic
acid.
The corresponding TPD profiles for the acetic acid and acetone-sa-
turated ZSM-5 catalyst are presented in Figs. 7 and 8 respectively and
exhibit considerable differences in comparison with MgO. At low
temperature (< 200 °C), the profile is dominated by large desorption of
water, together with some CO2 and unreacted physisorbed acetic acid,
at ˜130 °C. This initial dehydration step could be related with the
Fig. 6. Temperature-programmed desorption profile of acetone-saturated MgO.
38

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
formation of acyl intermediates via the interaction of acetic acid with
the strong Brønsted acid sites of the zeolite, as previously suggested by
Gumidyala et al. [29] on the basis of similar studies on HZSM-5. Fol-
lowing this initial dehydration step, the reaction zone is broader on the
acidic catalyst compared to MgO and extends from ˜250 to 500 °C.
There is a single desorption peak of acetone at 310 °C, accompanied by
evolution of CO2 and H2O. Ketonization probably occurs via the inter-
action of the formed acyl species with either acetates or carboxylates, as
discussed in detail in the following section. The products profile is
however dominated by the formation of light olefins, namely isobutene,
propene and ethene, in accordance with the results of the reactivity
tests presented above. No formation of aromatics and phenols is ob-
served, as the mass-to-charge (m/z) ratios monitored during the TPD
studies was < 100. It is also unlikely to observe oligomerization/cy-
clization products in such studies due to the low residence time at each
temperature and the low partial pressures of the reactants (acetic acid/
acetone). The production of the light olefins exhibits a doublet, with
temperature maxima at 310 °C (together with the formed acetone) and
350 °C. Temperature-programmed desorption studies of acetone pre-
adsorbed on HZSM-5 showed that acetone is converted at low tem-
peratures by zeolitic acid sites to mesityl oxide that undergoes further
condensation and cracking at higher temperature, releasing mainly
isobutene in the gas phase [30]. Significant sequential condensation of
acetone was also observed at temperature > 300 °C during acetic acid
ketonization tests on ZSM-5 [29]. Considering that similar evolution of
olefins was also observed at similar temperature range in the TPD
profile of the acetone-saturated ZSM-5 (Fig. 8), it can thus be deduced
that acetone is formed via ketonization of acetic acid on the zeolite acid
sites and then undergoes acid-catalyzed condensation to higher mole-
cular weight products that finally crack to yield olefins and light gases.
pathway. Several reaction routes have been proposed in literature. One
possible pathway is via the adsorption of the acid as CH3COO− and H+
on an electron–rich surface and the subsequent formation of the ketone
via interaction of two such surface acetate ions (Reaction 3) [34,35]. It
is however anticipated that the strongly acidic protons of bridging
hydroxyls in the zeolitic framework promote the formation of surface
acyl species during adsorption of acetic acid to form CH3CO+ and OH−
[31]. Acetone would then be formed via interaction with an acetate ion
as shown in Reaction 4. Alternatively, the recombination of a surface
acyl with a carboxylate could form β-ketoacid that decomposes to yield
the ketone through a similar route to the mechanism of ketonization on
metal oxide surfaces [31,36].
Reaction 3: Ketonization via surface acetate ions
2Zeol-OH + 2 CH3COOH → 2 Zeol+ — CH3COO− +2 H2O
2Zeol+ – CH3COO− → CH3COCH3 + CO2 + ZeolO− + Zeol+
ZeolO− + Zeol+ + H2O → 2Zeol-OH
Reaction 4: Ketonization via acylium ion formation
Zeol-OH+ + CH3COOH → Zeol-O− — CH3CO+ + H2O
Brønsted sites Acylium ion
Zeol-OH + CH3COOH → Zeol+ — CH3COO− + H2O
Acetate ion
Zeol-O− — CH3CO+ + Zeol+ — CH3COO− → CH3COCH3 + CO2
Zeol-O− + Zeol+ + H2O → Zeol-OH+ + Zeo+l-OH
Zeol -OH+ + CH3 COOH
Zeol -O --- CH3CO + H2O
+Acylium ion
4. Mechanistic pathways of acetic acid conversion on acidic and
basic sites
Brønsted sites
Zeol -OH + CH3 COOH
Zeol --- CH3COO + H2O
Acetate ion
Despite the obvious differences between the active sites of the two
investigated catalysts, being acidic in nature on ZSM-5 and basic on
MgO, the primary reaction in the conversion of acetic acid is the ke-
tonization to the corresponding ketone, acetone, accompanied by the
formation of CO2 and H2O, as evidenced by the results presented in the
previous sections. The reaction follows however a different underlying
mechanistic pathway over the two materials. It has been well estab-
lished that oxides with low lattice energy (or very high basicity) such as
alkali and alkali earth metal oxides, including MgO, CaO, BaO, SrO and
CdO, interact very strongly with acetic acid. This interaction results in
the formation of bulk carboxylate salts that decompose upon thermal
treatment to acetone, water and CO2 according to Reaction (1) [31]. As
demonstrated by the fluidized bed experiments and the TPD study of
this work, the initially formed acetone undergoes secondary reactions
over MgO, with the main final products being higher ketones, water and
coke. Based on the GC–MS analysis, these ketones are mainly pente-
nones and hexenones (Table S2) that probably form from acetone oli-
gomerization via aldol self-condensation reactions. The ketones may
evolve to coke through cyclization and condensation reactions (Reac-
tion 2) [32,33].
Zeol -O --- CH3 CO+ + Zeol+ --- CH3COO
CH3COCH3 + CO2
Zeol -O + Zeol+ + H2O
Zeol -OH+ + Zeol -OH
The performance of ZSM-5 and MgO was largely differentiated
concerning the distribution of the secondary products, as shown in the
present study. The formation of CO, C2-C4 olefins, phenols and aro-
matics that was mainly observed with ZSM-5 indicates the occurrence
of more complex reaction scheme of acetone over the acidic active sites.
Deoxygenation could be progressing over the ZSM-5 through the pro-
duction of i-butene from acetone, by dimerization with water abstrac-
tion (aldol condensation) and successive cracking of the mesityl oxide
which is an intermediate product (Reaction 5) [30,36,37]. The i-butene
molecule may further oligomerize to higher olefins, which ultimately
form aromatic molecules, such as alkyl benzenes and xylene (Reaction
6), possibly via a Diels Alder mechanism or other cyclization and de-
hydrogenation reactions [14,38,39]. The cracking of higher olefins to-
wards ethylene and propylene which may also be converted to aro-
matics on ZSM-5 is also a possible pathway. The detection of propylene
and other C4 hydrocarbons in the reactions of acetic acid and acetone
over ZSM-5 in the present study may also be attributed to the formation
of alkyl aromatics and their consecutive dealkylation [14]. It should be
noted that, although the reaction of acetone to i-butene and succes-
sively to aromatics is negligible over MgO (based on the observed
products), the initial aldol condensation to mesityl oxide and its limited
thermal cracking cannot be completely excluded.
Reaction 1: Bulk Ketonization
2CH3COOH + MgO → Mg(CH3COO)2 + H2O
Mg(CH3COO)2 → MgO + CH3COCH3 + CO2
Reaction 2: Aldol Condensation and Polymerization to higher
ketones and coke
Reaction 5: i-Butene production
CH3COCH3 + CH3COCH3 → (CH3)2C = CH(CO)CH3 + H2O
Aldol condensation
CH3COCH3 → pentenone, hexenone → coke
Ketonization has been much less studied on strong acidic materials,
such as the ZSM-5 zeolite. As opposed to MgO and other low lattice
energy oxides, ketonization follows in this case a surface catalyzed
2(CH3)2C = CH(CO)CH3
+
2Zeol-OH+
→
2(CH3)2C = CH2
+
2ZeolO− — CH3CO+
39

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
Fig. 7. Temperature-programmed desorption profile of acetic acid-saturatedZSM-5.
Fig. 8. Temperature-programmed desorption profile of acetone-saturated ZSM-5.
2ZeolO− — CH3CO+ + H2O → 2ZeolOH+ + CH3COCH3 + CO2
2(CH3)2C = CH(CO)CH3 → 2(CH3)2C = CH2 + CH3COCH3 + CO2
The phenolic molecules, detected in the product distribution of the
ZSM-5 catalyzed reaction, are expected products of acetone poly-
merization [40,41]. As reported above (see reaction 5), acetone is
converted at low temperatures by zeolitic acid sites to mesityl oxide
[30], which can undergo cyclization reactions to isophorone [42] as
shown in Reaction Scheme 7. Alternatively, the reaction of mesityl
oxide with acetone can yield phorone that has been reported to convert
Mesityl oxide cracking
Reaction Scheme 6: i-Butene reaction network [adapted from
14]
40

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
to isophorone over zeolite catalysts [43]. Isophorone is a key inter-
mediate for the production of 3,5-xylenol via catalytic rearrangement
[44]. The dimethyl-phenol can further react to methyl-phenols with loss
of methane, in agreement with the formation of methane that was de-
tected in the reaction products of the acetone reaction over the ZSM-5
catalyst in this study.
5. Conclusions
Acetic (or other) acids in bio-oil can be effectively removed during
in situ or ex situ upgrading via pyrolysis with either acidic (ZSM-5) or
basic (MgO) catalysts via different reaction pathways. The primary
route involves in all cases the deoxygenation of acetic acid via ketoni-
zation towards acetone, CO2 and water. The successive reactions are
Reaction Scheme 7: Phenols production [adapted from 42]
The CO detected in the ZSM-5 catalyzed reaction is unlikely to
originate from the decarbonylation of acetic acid, as this route has been
reported to be insignificant compared to decarboxylation and dehy-
dration under similar conditions and catalysts as this work [14].
However, decarbonylation was found to have an important contribu-
tion in the transformation of acetone over HZSM-5 and therefore CO
could be a product of the secondary decarbonylation of the acetone
produced in the process [14]. There is also a possibility for direct de-
hydration of acetic acid to ketene and successive decomposition of
ketene to methylene radicals and CO [45]. The ethylene observed in the
gaseous products over ZSM-5 supports the direct ketene decomposition
route, as ethylene is produced by the combination of two methylene
radicals (Reaction 8) [37]. Ethylene can then be readily converted to
benzene or alkylated benzene via oligomerization, cyclization and de-
hydrogenation on the Brønsted acid sites of ZSM-5. The polymerization
of ketene to coke species could also contribute to the CO yield [33].
Reaction 8: Dehydration to Ketene, decomposition to ethylene
and CO, and oligomerization-cyclization-dehydrogenation of
ethylene to aromatics
however different, altering significantly the final product distribution
on the two catalysts. Higher ketones are mainly produced on MgO,
while aromatics, phenolic compounds and light olefins are observed in
the products of ZSM-5 catalyzed reaction. Temperature-programmed
desorption studies on acetic acid and acetone-saturated catalyst amples
show that at high temperature, acetone undergoes secondary self-con-
densation and oligomerization reactions yielding mainly CO2 and coke
on MgO. On ZSM-5, both the direct dehydration of acetic acid and the
cracking of the acetone condensation products play an important role,
leading to the production of light olefins which are then converted to
phenols and aromatics. The obtained catalytic results and the reaction
mechanisms proposed in the present work can be utilized for the design
of improved catalysts in bio-oil upgrading via catalytic pyrolysis/
cracking, selecting between the production of a bio-oil rich in aromatics
and phenolics or a bio-oil that contains mainly higher ketones and re-
lated compounds that could be hydrodeoxygenated towards gasoline or
diesel range alkanes.
Acknowledgments
This research was co-financed by European Union (European Social
Fund) and Greek national funds through Operational Program
"Education and Lifelong Learning" of the National Strategic Reference
Framework (NSRF)-Research Funding Program: THALES-Investing in
knowledge society through the European Social Fund (MIS 380405).
CH3COOH → CH2CO + H2O
CH2CO → :CH2 + CO
2[:CH2] → C2H4
3 C2H4 → C6H6 + 3 H2
Appendix A. Supplementary data
Finally, the presence of methane in the product slate of ZSM-5 could
be the result of the decomposition of surface acetates to methane [38],
according to reaction 9, supporting the surface acetate formation route.
Reaction 9: Acetate Decomposition
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.mcat.2018.12.012.
References
CH3COO− + zeol-OH→ CH4 + CO2 + zeol-O−
The above described reaction pathways, proposed on the basis of
the obtained experimental data, the specific characteristics of the acidic
ZSM-5 zeolite and the basic MgO, and the available literature regarding
related mechanistic sutdies of various individual reactions, can be uti-
lized in order to elucidate the product yields and distribution in the
upgrading (in situ or downstream) of bio-oil under typical catalytic
pyrolysis conditions. More specifically, the aim and future work is fo-
cused on the development of catalysts that can effectively eliminate
acetic acid (and other acids) present in the bio-oil and at the same time
tune the selectivity towards the desired product in each case.
[1] A.A. Lappas, K.G. Kalogiannis, E.F. Iliopoulou, K.S. Triantafyllidis, S.D. Stefanidis,
WIREs: Energy Environ. 1 (2012) 285–297.
[2] D.A. Bulushev, J.R. Ross, Catal. Today 171 (2011) 1–13.
[3] E.F. Iliopoulou, K.S. Triantafyllidis, A.A. Lappas, WIREs Energy Environ. (2018),
https://doi.org/10.1002/wene.322;
S. Wan, Y. Wang, Front. Chem. Sci. Eng. 8 (2014) 280–294.
[4] E.F. Iliopoulou, P.A. Lazaridis, K.S. Triantafyllidis, Nanocatalysis in the fast pyr-
olysis of lignocellulosic biomass, chapter 27, pp. 655–714, in: B. Sels, M.l Van de
Voorde (Eds.), Nanotechnology in Catalysis: Applications in the Chemical Industry,
Energy Development, and Environment Protection, Wiley, 2017.
[5] C. Liu, H. Wang, A.M. Karim, J. Suna, Y. Wang, Chem. Soc. Rev. 43 (2014)
7594–7623.
[6] A. Zheng, L. Jiang, Z. Zhao, Z. Huang, K. Zhao, G. Wei, H. Li, WIREs: Energy and
41

A.C. Psarras et al.
MolecularCatalysis465(2019)33–42
Environ. (2016), https://doi.org/10.1002/wene.234.
[24] J.D. Adjaye, R.K. Sharma, N.N. Bakhshi, J.S. Kevin, C.S. Emerson (Eds.), Stud. Surf.
[7] M. Guisnet, P. Magnoux, D. Martin, Stud. Surf. Sci. Catal. 111 (1997) 1–19.
[8] S.D. Stefanidis, K.G. Kalogiannis, E.F. Iliopoulou, A.A. Lappas, P.A. Pilavachi,
Bioresour. Technol. Rep. 102 (2011) 8261–8267.
Sci. Catal. vol. 73, Elsevier, 1992, pp. 301–308.
[25] C.X. Ma, G. Liu, Z.L. Wang, Y.F. Li, J. Zheng, W.X. Zhang, M.J. Jia, React. Kinet.
Catal. Lett. 98 (2009) 149–156.
[9] S.D. Stefanidis, S.A. Karakoulia, K.G. Kalogiannis, E.F. Iliopoulou, A. Delimitis,
H. Yiannoulakis, T. Zampetakis, A.A. Lappas, K.S. Triantafyllidis, Appl. Catal. B 196
(2016) 155–173.
[26] Z. Liu, F. Peng, X. Liu, Adv. Mater. Res. 550–553 (2012) 424–428.
[27] M. Leon, L. Faba, E. Diaz, S. Bennici, A. Vega, S. Ordonez, A. Auroux, Appl. Catal. B
147 (2014) 796–804.
[10] E. Putun, Energy 35 (2010) 2761–2766.
[28] L. Wu, T. Moteki, A.A. Gokhale, D.W. Flaherty, D. Toste, Chem 1 (2016) 32–58.
[29] A. Gumidyala, T. Sooknoi, S. Crossley, J. Catal. 340 (2016) 76–84.
[30] L. Kubelkova, J. Novakova, Zeolites 11 (1991) 822–826.
[31] T.N. Pham, T. Sooknoi, S.P. Crossley, D.E. Resasco, ACS Catal. 3 (11) (2013)
2456–2473.
[11] Y. Lin, C. Zhang, M. Zhang, J. Zhang, Energy Fuels 24 (2010) 5686–5695.
[12] J. Fermoso, H. Hernando, P. Jana, I. Moreno, J. Prech, C. Ochoa-Hernandez,
P. Pizarro, J.M. Coronado, J. Cejka, D.P. Serrano, Catal. Today 277 (1) (2016)
171–181.
[13] H. Hernando, I. Moreno, J. Fermoso, C. Ochoa-Hernández, P. Pizarro,
J.M. Coronado, J. Čejka, D.P. Serrano, Biomass Conver. Bioref. 7 (3) (2017)
289–304.
[32] R. Martinez, M.C. Huff, M.A. Barteau, J. Catal. 222 (2004) 404–409.
[33] M. Watanabe, H. Inomata, R. Lee Smith Jr., K. Arai, Appl. Catal. A Gen. 219 (2001)
149–156.
[14] A.G. Gayubo, A.T. Aguayo, A. Atutxa, R. Aguado, M. Olazar, J. Bilbao, Ind. Eng.
Chem. Res. 43 (2004) 2619–2626.
[34] E. Kukulska-Zając, K. Góra-Marek, J. Datka, Microporous Mesoporous Mater. 96
(2006) 216–221.
[15] W. Shurong, D. Gongxin, Y. Haiping, L. Zhongyang, Prog. Energy Combust. Sci. 62
(2017) 33–86.
[35] A.G. Gayubo, A.T. Aguayo, A. Atutxa, B. Valle, J. Bilbao, J. Chem. Technol.
Biotechnol. 80 (2005) 1244–1251.
[16] E.F. Iliopoulou, S.D. Stefanidis, K.G. Kalogiannis, A. Delimitis, A.A. Lappas,
K.S. Triantafyllidis, Appl. Catal. B 127 (2012) 281–290.
[36] A. Pulido, B. Oliver-Thomas, M. Renz, M. Boronat, A. Corma, ChemSusChem 6 (1)
(2013) 141–151.
[17] E.F. Iliopoulou, S.D. Stefanidis, K.G. Kalogiannis, A.C. Psarras, A. Delimitis,
K.S. Triantafyllidis, A.A. Lappas, Green Chem. 16 (2014) 662–674.
[18] C.A. Emeis, J. Catal. 141 (1993) 347–354.
[37] C.D. Chang, A.J. Silvestri, J. Catal. 47 (1977) 249–259.
[38] M.A. Hasan, M.I. Zaki, L. Pasupulety, Appl. Catal. A Gen. 243 (2003) 81–92.
[39] A. Corma, G.W. Huber, Angew. Chem. Int. Ed. 46 (2007) 7184–7201.
[40] S. Lippert, W. Baumann, K. Thomke, J. Mol. Catal. 69 (1991) 199–214.
[41] K.V. Ramanamurty, G.S. Salvapati, J. Sci. Ind. Res. 59 (2000) 339–349.
[42] K.V. Ramanamurty, G.S. Salvapati, Indian J. Chem. 38B (1999) 24–28.
[43] O.I. Kusnetsov, G.M. Panchenkov, A.M.D. Guseim, T.A. Chasova, Nefte- pererab
Neftekhzm 3 (1973) 28.
[19] A.C. Psarras, E.F. Iliopoulou, L. Nalbandian, A.A. Lappas, C. Pouwels, Catal. Today
127 (2007) 44–53.
[20] V.G. Komvokis, S. Karakoulia, E.F. Iliopoulou, M.C. Papapetrou, I.A. Vasalos,
A.A. Lappas, K.S. Triantafyllidis, Catal. Today 196 (2012) 42–55.
[21] P. Lanzafame, S. Perathoner, G. Centi, E. Heracleous, E.F. Iliopoulou,
K.S. Triantafyllidis, A.A. Lappas, ChemCatChem 9 (2017) 1632–1640.
[22] A.A. Gusev, A.C. Psarras, K.S. Triantafyllidis, A.A. Lappas, P.A. Diddams, Molecules
22 (2017) 1784–1804.
[44] K.V. Ramanamurty, G.S. Salvapati, M. Janardanarao, J. Mol. Catal. 54 (1989) 9–30.
[45] X. Li, S. Wang, Y. Zhu, G. Yang, P. Zheng, Int. J. Hydrogen Energy 40 (2015)
330–339.
[23] A. Demirbas, Energ. Convers. Manage. 41 (2000) 633–646.
42