Kinetics of Valeric Acid Ketonization and Ketenization in Catalytic Pyrolysis on Nanosized SiO2, γ-Al2O3, CeO2/SiO2, Al2O3/SiO2 and TiO2/SiO2

Article abstract of DOI:10.1002/cphc.201601370

Valeric acid is an important renewable platform chemical that can be produced efficiently from lignocellulosic biomass. Upgrading of valeric acid by catalytic pyrolysis has the potential to produce value added biofuels and chemicals on an industrial scale. Understanding the different mechanisms involved in the thermal transformations of valeric acid on the surface of nanometer-sized oxides is important for the development of efficient heterogeneously catalyzed pyrolytic conversion techniques. In this work, the thermal decomposition of valeric acid on the surface of nanoscale SiO₂, γ-Al₂O₃, CeO₂/SiO₂, Al₂O₃/SiO₂ and TiO₂/SiO₂ has been investigated by temperature-programmed desorption mass spectrometry (TPD MS). Fourier transform infrared spectroscopy (FTIR) has also been used to investigate the structure of valeric acid complexes on the oxide surfaces. Two main products of pyrolytic conversion were observed to be formed depending on the nano-catalyst used—dibutylketone and propylketene. Mechanisms of ketene and ketone formation from chemisorbed fragments of valeric acid are proposed and the kinetic parameters of the corresponding reactions were calculated. It was found that the activation energy of ketenization decreases in the order SiO₂>γ-Al₂O₃>TiO₂/SiO₂>Al₂O₃/SiO₂, and the activation energy of ketonization decreases in the order γ-Al₂O₃>CeO₂/SiO₂. Nano-oxide CeO₂/SiO₂ was found to selectively catalyze the ketonization reaction.

Full text of DOI:10.1002/cphc.201601370

Accepted Article
Title: Kinetics of valeric acid ketonization and ketenization in catalytic
pyrolysis on nanosized SiO2, γ-Al2O3, CeO2/SiO2, Al2O3/SiO2
and TiO2/SiO2
Authors: Kostiantyn Kulyk, Borys Palianytsia, John D. Alexander, Liana
Azizova, Mykola Borysenko, Mykola Kartel, Mats Larsson,
and Tetiana Kulik
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemPhysChem 10.1002/cphc.201601370
Link to VoR: http://dx.doi.org/10.1002/cphc.201601370
A Journal of
www.chemphyschem.org

ChemPhysChem
10.1002/cphc.201601370
1
Kinetics of valeric acid ketonization and ketenization in catalytic pyrolysis on nanosized
SiO , γ-Al O , CeO /SiO , Al O /SiO and TiO /SiO
2
2
2
3
2
2
2
3
2
2
a,*
b
a
b
Kostiantyn Kulyk , Borys Palianytsia , John D. Alexander , Liana Azizova , Mykola
b
b
a
b
Borysenko , Mykola Kartel , Mats Larsson , Tetiana Kulik
a
Stockholm University, AlbaNova University Center, Department of Physics, SE-106 91
Stockholm, Sweden
b
Chuiko Institute of Surface Chemistry, National Academy of Sciences of Ukraine, 17
General Naumov Street, 03164 Kyiv, Ukraine
ABSTRACT
Valeric acid is an important renewable platform chemical that can efficiently be produced
from lignocellulosic biomass. Upgrading of valeric acid by catalytic pyrolysis has the
potential to produce value added biofuels and chemicals on an industrial scale. Understanding
the different mechanisms involved in the thermal transformations of valeric acid on the
surface of nanosized oxides is important for the development of efficient heterogeneously-
catalyzed pyrolytic conversion techniques. In this work, the thermal decomposition of valeric
acid on the surface of nanoscale SiO , γ-Al O , CeO /SiO , Al O /SiO and TiO /SiO has
2
2
3
2
2
2
3
2
2
2
been investigated by Temperature-Programmed Desorption Mass Spectrometry (TPD MS).
Fourier transform infrared spectroscopy (FTIR) has also been used to investigate the structure
of valeric acid complexes on the oxide surfaces. Two main products of pyrolytic conversion
were observed to be formed depending on the nano-catalyst used – dibutylketone and
propylketene. Mechanisms of ketene and ketone formation from chemisorbed fragments of
valeric acid are proposed and the kinetic parameters of the corresponding reactions were
calculated. It was found that the activation energy of ketenization decreases in the order
SiO >γ-Al O >TiO /SiO >Al O /SiO , and the activation energy of ketonization decreases in
2
2
3
2
2
2
3
2
the order γ-Al O >CeO /SiO . Nano-oxide CeO /SiO was found to selectively catalyze the
2
3
2
2
2
2
ketonization reaction.
*
Corresponding author.
Email address: kostiantyn.kulyk@fysik.su.se (K. Kulyk)
Keywords: catalysis; pyrolysis; ketenization; ketonization; kinetics.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
1
. Introduction
At present, mainly due to low cost, high abundance and fast renewability, lignocellulosic
biomass is considered to be one of the most promising feedstocks for the production of bio-
based fuels and chemicals [1–3]. The efficient separation needed for further processing of the
prime components of lignocellulose (LC), such as lignin, hemicellulose and cellulose, has
been a complex task for a long time due to complexity of the chemical linkages between
them. However, numerous research efforts conducted over the past decade have led to the
development of various technologies that can now be used for efficient separation and further
conversion of LC components to useful platform chemicals [4–9].
Valeric acid (VA) is an example of an important renewable (lignocellulose-derived) platform
compound that has recently attracted significant attention [10–13]. The process of
manufacturing VA from lignocellulose-derived levulinic acid and γ-valerolactone has already
been demonstrated to be efficient, simple and robust [10–13]. VA is the key intermediate in
the conversion of LC to valeric acid esters, which are considered to be a new class of
cellulosic biofuels [10–12].
As an alternative to the esterification process, VA can be upgraded by using catalytic
pyrolysis. Techniques based on catalytic pyrolysis promise to advance the area of biomass
conversion due to their numerous potential socio-economic advantages [14–19]. For example,
a pyrolitic conversion of VA to yield 5-nonanone along with CO and water has recently been
2
demonstrated [20]. The reaction above is known as ketonization, or ketonic decarboxylation,
and is considered to be particularly useful for modern biorefineries. Along with the formation
of a product with an extended carbon chain, this reaction includes simultaneous
decarboxylation and dehydroxylation – two highly desired routes to decrease oxygen content
in the biomass feedstocks by releasing non-polluting by-products, such as CO and H O.
2
2
Formed 5-nonanone can be further upgraded to liquid alkanes through catalytic
hydrogenation/dehydration and used as a fuel [20].
However, the pyrolytic biomass conversion technology is not yet fully developed with respect
to the activity and selectivity of the heterogeneous catalyst used. A full understanding of the
mechanisms of surface-assisted catalytic transformations has proven to be difficult even for
such well-studied biomass components as carboxylic acids. Ultrafine and nano particles of
metal oxides have been reported to be particularly useful heterogeneous catalysts for the
pyrolytic conversion of aliphatic and cinnamic carboxylic acids [21–40]. A number of useful
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
3
thermal transformations for biomass conversion have been reported to proceed on the oxide
surfaces, for example – decarboxylation [37–40], ketenization [21–26] and ketonization
[
16,27–36]. In this sense, the identification of reaction products and the understanding of the
mechanisms of thermal decomposition of VA on the surface of potential catalysts are essential
for the development of alternative routes to utilize the derivatives of cellulosic biomass.
Studies devoted to the ketonization of carboxylic acids on the surface of different
heterogeneous catalysts have attracted much attention during recent years
[
16,28,29,33,34,36]. The advances in determining the ketonization reaction mechanisms
occurring on the surfaces of various heterogeneous catalysts have been recently summarized
in two reviews by Pacchioni [36] and Pham et al. [16]. At present, the mechanism involving a
β-keto acid as an intermediate seems to be the most plausible [16,29,36]. However, the
alternative concerted mechanism cannot be completely disregarded [16,36].
Ketenization of carboxylic acids while receiving much less attention in the literature is also an
important process as ketenes can be intermediates in the production of esters, anhydrides, and
amides. In particular, the interaction of propylketene with aliphatic alcohols can yield esters
of VA, which are known to be high-energy-density valeric biofuels.
Temperature-programmed desorption mass spectrometry (TPD-MS) can help to elucidate the
kinetics and mechanisms of heterogeneous thermocatalytic reactions [26,41,42]. In this work,
a comparative study of the thermal transformation of VA on the surfaces of nanosized SiO2,
γ-Al O , CeO /SiO , Al O /SiO and TiO /SiO has been carried out by using TPD-MS. The
2
3
2
2
2
3
2
2
2
oxides have been selected based on the various types of active surface sites they posses giving
them the ability to catalyze different kinds of reactions.
Data on the pyrolysis kinetics, obtainable from TPD MS, is important for finding optimized
pyrolytic conditions leading to the synthesis of a particular, useful product. It is known that
the conditions of pyrolytic conversion of biomass are sometimes chosen based on purely
empirical observations. For example, it is believed that flash pyrolysis should be conducted in
the range at 450-600 °C as it yields the highest amount of bio-oil. However, according to our
TPD MS data, catalytic reactions can proceed at significantly lower temperatures, hence, the
desired chemicals may already be decomposed. As an advanced alternative, pyrolytic
conversion can be performed at the temperature of the maximum formation rate of the desired
product if the corresponding kinetic parameters are known.
2
. Experimental methods
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
4
2
.1. Materials
2
-1
-1
Powdery fumed silica (with a surface area S =300 m g , average size of the primary particles
a
2
d≈9.6 nm), powdery fumed Al O /SiO (S =207 m g , d≈11.4 nm 1.2 % Al O , amount of
2
3
2
a
2
3
2
-1
aluminum atoms in the surface layer: 6.2 %) and powdery fumed TiO /SiO (S =156 m g ,
2
2
a
d≈20.6 nm 14% TiO , amount of titanium atoms in the surface layer: 7.8 %) were supplied by
2
the Research and Experimental plant of the Chuiko Institute of Surface Chemistry in Kalush,
Ukraine. All the fumed oxides consist of nonporous primary particles. The weight content of
titania in the fumed TiO /SiO and alumina in the fumed Al O /SiO was 14 wt. % and 1.3 wt.
2
2
2
3
2
%
, respectively [43,44]. The surface content of titanium (7.8 wt. %) in the TiO /SiO and
2 2
aluminum (6.2 wt.%) in the Al O /SiO was determined previously using Auger electron
2
3
2
spectroscopy [43]. The physicochemical properties of these materials are described elsewhere
2
-1
[
43–47]. Powdery fumed alumina γ-Al O (S = 111 m g , d ≈ 40 nm) was supplied by
2 3 a
Degussa AG (Germany). All of the pyrogenic oxides were synthesized by the hydrolysis of
2
-1
their chlorides in a hydrogen-oxygen flame. Nanocomposite CeO /SiO (S = 189 m g , 23.3
2
2
a
%
CeO , average size of cerium dioxide crystallites d ≈ 4 nm) was prepared using powdery
2
fumed silica and cerium (III) acetylacetonate (Aldrich). The modification of silica with
cerium acetylacetonate was performed as described previously [48]. Prior to the adsorption of
VA the powders were dried at 400 ºC in air for 2 hours in order to remove adsorbed organic
matter.
VA (99.5% pure) was obtained from Merck (Germany). Samples of the nano-oxides were
-1
impregnated by VA with loadings of 0.3 mmol g , which corresponds to surface coverage
1
3
2
13
2
values of 6.0 x 10 molecules/cm for silica, 16.2 x 10 molecules/cm for γ-Al O , 8.7 x
2
3
1
3
2
13
2
13
1
0
molecules/cm for Al O /SiO , 5.5 x 10 molecules/cm for CeO /SiO , 11.5 x 10
2 3 2 2 2
2
molecules/cm for TiO /SiO . In a typical sample, in a preparation session 1 g of fumed oxide
2
2
was gradually mixed with 25 ml of an aqueous solution of the acid. The samples were then
aged at room temperature for 24 hours, dried at room temperature in air and stored in a
desiccator until its use in the TPD-MS experiment.
2
.2. Fourier transform infrared spectroscopy (FTIR)
FTIR spectra were recorded using a FT IR NEXUS spectrometer (Thermo Nicolet) at room
-1
temperature in the frequency range of 4000 to 400 сm in diffuse reflection mode at a
-1
resolution of 8 сm , a scan rate of 0.5 сm/s and number of scans of 50. The powdered
samples of SiO , γ-Al O , CeO /SiO Al O /SiO and TiO /SiO and the samples of SiO , γ-
2
2
3
2
2
2
3
2
2
2
2
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
5
Al O , CeO /SiO Al O /SiO and TiO /SiO impregnated with VA were mixed with freshly
2
3
2
2
2
3
2
2
2
calcined and milled KBr (10:100). VA was mixed in a 1:10 ratio (mass of VA to mass of
KBr).
2
.3. Temperature-programmed desorption mass spectrometry (TPD MS)
Temperature programmed desorption experiments were carried out using an electron
ionization MKh-7304A monopole mass spectrometer (Sumy, Ukraine) which has been
adapted for TPD measurements [26,49]. For all the experiments, the mass samples were
approximately 20 mg. Before the TPD measurements, the samples were degassed to
-5
~
5·10 Pa at ~20 °C. All TPD measurements were carried out by heating the samples up to
-1
~
750 °C at a constant rate of 0.17 ºC s .
.4. P–T pressure-temperature curves
2
Mass spectra and the P–T pressure-temperature curves (P – the pressure of volatile pyrolysis
products, T – temperature of samples) were recorded and analyzed using a computer-based
data acquisition and processing setup. The pressure of the volatile pyrolysis products was
measured using a VMB-14 pressure gauge. A VMB-14 gauge uses cold-cathode ionization to
determine the pressure, therefore, quantitative estimation of the actual pressure of the
desorbed molecules is only possible if the gas sensitivity factors specific for each molecule
are taken into account. In the present work we only used the P-T curves for qualitative
comparison of pyrolytic processes on different surfaces without consideration of individual
gas sensitivity factors.
2
.5. Non-isothermal parameters
All non-isothermal parameters of the chemical reactions on the oxides surfaces (such as the
≠
temperature of the maximum desorption rate Tmax, reaction order n, activation energy E , pre-
≠
exponential factor ν and activation entropy dS ) were calculated from the TPD-MS data
0
using an in-house computer program [26,49]. The procedure for obtaining the non-isothermal
parameters from TPD-MS data has been described in a number of studies and reviews [40–
4
2]. Here, we employed a method that uses the linear form of the Arrhenius equation. The
procedure used was described in detail earlier [26,49,51,52]. More details can also be found in
the Supplemental Information.
2
.6. The linear free-energy relationship (LFER) method and the reaction constant ρ*
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
6
Kinetic studies, along with the kinetic isotope effect and isotope labelling investigations, are
among the few direct experimental tools available for direct exploration of the mechanisms of
chemical reactions. Another effective tool for investigating the mechanisms of chemical
reactions is the linear free-energy relationship (LFER) method [41, 53, 54]. This principle
means that the free energy values ΔG associated with the structure of the reactants for a given
reaction series linearly depend on the free energy values ΔG of another reaction series with
0
the same variable parameter.
This principle was earlier applied by some of us to establish the mechanism of the
ketenization reaction on the silica surface [26]. In that work, the thermal behavior of a series
of C -C carboxylic acids was studied on the silica surface [26]. The kinetic parameters were
1
10
extracted directly from the experimental data and then used for further calculations.
Specifically, the reaction constant ρ* was calculated by using the modified Taft equation [26]:
lgk/k =lg(B/2.3lgB)·ΔT /T
=ρ∑σ
(1)
o
max o max
where
n-1
B=(nν0TmaxCmax )b
(2)
n-1
In this formula n is the reaction order, ν is the pre-exponential factor, C
is the
max
0
concentration of the adsorbate at Tmax and b is the value of the sample heating rate.
The proposed modified Taft equation can be considered a universal equation for studying the
kinetics and mechanisms of thermal catalytic reactions [26]. In this work we used this
modified Taft equation (1) to compare the reaction rate of ketenization on the surfaces of
SiO , γ-Al O , TiO /SiO , Al O /SiO . In addition, the previously obtained reaction constant
2
2
3
2
2
2
3
2
ρ* was used to help understand the ketenization mechanism on the surfaces these oxides [26].
In order to fulfill the Taft equation within any reaction series, it is necessary that one of the
following conditions is fulfilled: the constancy of the activation entropy, i.e. ΔS = const
(isoentropic series); the constancy of the enthalpy, i.e. ΔH = const (isoenthalpic series); the
proportionality of ΔH and ΔS (isokinetic series). The surfaces of SiO , γ-Al O , TiO /SiO ,
2
2
3
2
2
Al O /SiO have been considered as an isoentropic reaction series in the current work.
2
3
2
Taft equation can be used to study the mechanisms of reactions primarily by comparing the
signs and the absolute values of the reaction constant ρ* [54]. The physical meaning of the
constant ρ* is that it characterizes the relative sensitivity of the reaction to a change in the
structure of the reacting compounds relative to the standard reaction series (in the case of the
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
7
reaction series involving the dissociation of aliphatic acids ρ* = 1). The change in the reaction
constant ρ* during the transition from one reaction series to another is dictated by the reaction
mechanism, the electronic effects of the substituents on the reaction center, and the reaction
conditions.
For example, the type of the reactive transformation in the system can be established by
determining the sign of ρ*. A negative sign means a nucleophilic (by reagent) and an
electrophilic (by substrate) reaction. Such a reaction is accelerated by the introduction of
electron-donating substituents, which contribute to an increase in the electron density at the
reaction center. A positive sign of ρ* means an electrophilic (by reagent) and a nucleophilic
(by substrate) reaction, which is slowed down by electron-donor substituents and accelerated
by electron-withdrawing substituents. Since the kinetics of complex reactions are often
governed by the rate-determining step (the step passing through the transition state with the
highest energy), calculation of the sign of the constant ρ* enables identification of the rate-
determining step in the reaction.
The absolute value of the reaction constant ρ* (which varies from 0 to 11-12) is also of great
importance. The highest interaction of the substituent with the reaction center and the largest
influence on the kinetics appears in ionic reactions. Correspondingly, the absolute value of ρ*
characterizes the degree of ionicity of the reaction (the degree of polarity of the transition
state). We can assume that the higher the absolute value of ρ*, the more polar the transition
state is. The highest values of the reaction constant ρ* are characteristic for purely ionic
reactions, while the low values are typical for reactions that proceed via the concerted
mechanisms (for example, the Diels-Alder reaction, etc.).
3
. Results and discussions
In this study, we used TPD MS to investigate the pyrolysis of VA on various oxide surfaces.
The relative efficiency of the pyrolytic process was estimated by analyzing the pressure of
released pyrolysis gases as a function of temperature. Further analysis of the mass spectra at
different temperatures was performed to identify the specific products formed. Comparison of
the shape of the TPD curves for different fragment and parent ions made it possible to
determine whether the ions originate from a common parent reaction. Integrated intensities of
the respective TPD curves were used to calculate the kinetic parameters of the observed
ketonization and ketenization reactions. Possible reaction mechanisms have been suggested
on this basis.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
8
3
.1. An analysis of the (P–T) curves and monitoring of water desorption
An analysis of the (P–T) curves indicates that VA desorbs from the surfaces of nanosized
oxides in several stages (Figure 1a). Comparison of the peaks on the (P-T) curves, see Fig. 1a,
with the mass spectra and TPD curves of the main pyrolysis products at respective
temperatures, see Figs. 2-5, revealed that peaks at a temperature below < 200 °C are primarily
caused by the processes of desorption of physically adsorbed and hydrogen-bonded molecular
VA and water. In the case of the γ-Al O surface, VA desorbs in a molecular form at a
2
3
relatively high temperature Tmax = 155 °C (γ-Al O ), while for all the other studied oxides the
2
3
correspondent desorption peak is observed at lower temperatures: Tmax = 80 °C (CeO /SiO ),
2
2
Tmax = 95 °C (SiO ), T = 90 °C (TiO /SiO ) and T = 100 °C (Al O /SiO ) (Fig. 1a).
2
max
2
2
max
2
3
2
Peaks observed at higher temperatures, Tmax > 200 °C, correspond to the desorption of
products formed as a result of pyrolytic decomposition reactions that proceed on the oxide
surfaces. Thermal decomposition of VA adsorbed on the CeO /SiO occurs in the most
2
2
narrow temperature range (200-400 °С) as compared with the other oxides (200-600 °С).
Also, the temperature of the maximum decomposition rate is the lowest for the CeO /SiO
2
2
sample – Tmax = 255 ° (CeO /SiO ), T = 360 °С (SiO ), T = 300 °С (TiO /SiO ), T
max
=
2
2
max
2
max
2
2
3
55 °С (γ-Al O ) and T = 330 °С (Al O /SiO ).
2 3 max 2 3 2
We monitored water desorption from the surfaces of the studied nanosized oxides as this
process to a large extent contribute to the shape of the (P–T) curves. The results can be seen in
Fig. 1b. The removal of water from the surface of nanosized oxides by heating under vacuum
is a complex process that has been extensively studied earlier [55-57]. All peaks
corresponding to water desorption at a temperature below < 200 °C, are primarily caused by
desorption of physically adsorbed and hydrogen-bonded molecular water (Fig.1b). The peaks
at temperatures higher than 200 °C are caused by dehydration and partial dehydroxylation of
surface hydroxyl groups. In addition, desorption of water can occur as a result of the catalytic
dehydration of valeric acid.
3
.2. Ketenization
The typical products of pyrolysis of aliphatic carboxylic acids adsorbed on silica surfaces are
the corresponding ketenes [21-26]. The interaction of the carboxylic acid reaction series
(acetic, propionic, butyric, isobutyric, valeric, nonanoic, decanoic and pivalic) with the
surface of nanosized SiO has been investigated previously by TPD MS [26].
2
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
9
In this study the reaction of ketene formation was observed on the surfaces of all the
investigated oxides except for CeO /SiO (Fig. 2-4, Scheme 1). Figures 2-4 show the TPD
2
2
profiles for the samples of VA adsorbed on the surfaces of γ-Al O , Al O /SiO and
2
3
2
3
2
TiO /SiO correspondingly. The molecular ion of propylketene (m/z = 84) accompanied by its
2
2
fragment ion with m/z = 55 was observed for all these samples (Figs. 2-4). Mass spectra of
the samples recorded at the temperature giving the maximum propylketene desorption rate are
shown in Figures 2a-4a. TPD curves corresponding to the molecular propylketene (m/z = 84)
and its fragment with m/z = 55 are shown in Figures 2b-4b. The desorption of the intact
+
molecular form of VA has been monitored using the ion at m/z = 60 (CH =C(OH) ) as the
2
2
+
abundance of the molecular ion at m/z = 102 is too low [58,59]. The ion CH =C(OH) is a
2
2
characteristic of valeric acid, and is formed by a McLafferty rearrangement from the
molecular ion of VA [58,59].
One point of particular note from these results was that the ketenization reaction observed on
the surfaces of γ-Al O and other mixed oxides was found to proceed at considerably lower
2
3
temperatures than in the case of pure silica [26]. Specifically, relative to the pure silica, the
temperature corresponding to the maximum desorption rate of propylketene is shifted by 96
°
C for the Al O /SiO sample, by 77 °C for γ-Al O , and by 87 °C for TiO /SiO (Fig. 6).
2 3 2 2 3 2 2
The Arrhenius plot method was used to determine the values of the kinetic parameters of the
reaction of propylketene formation on the surface of the studied oxides (Table 1). Obtained
2
2
values of R for the first-order reaction model (R =0.951-0.955) were found to be significantly
2
higher than for the second-order reaction model (R =0.907-0.930) (Table 1), which indicates
that the first-order model is more plausible. The pre-exponential factors and the activation
entropies for this reaction have close values in the studied series of oxide samples and so
presumably use the same mechanism of ketene formation (Scheme 1).
The average value (ν0average) of the calculated pre-exponential factors for ketenization of VA
5
-1
adsorbed on Al O /SiO , γ-Al O , and TiO /SiO was found to be 4.6 x 10 s . Whereas the
2
3
2
2
3
2
2
average value of the pre-exponential factors for ketenization of a series of carboxylic acids
6
-1
absorbed on silica surface was previously reported to be 3.7 x 10 s [26]. Therefore, as the
ν0average values differ only by approximately one order of magnitude, the reaction mechanism
is likely to be the same as was previously proposed [26]. This difference may be due to the
large size of the metal atoms present in the structure of the transition state.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
0
Previously, the modified Taft equation and the non-isothermal kinetic parameters were used
to estimate the reaction parameter ρ* for ketenization of a series of carboxylic acids absorbed
on silica (see section 2.6 for details) [26]. The value of the reaction parameter ρ* = +1.1
suggested that the reaction of ketene formation on silica proceed through a low-polar
intramolecular cyclic transition state with increased electron density [26,53,54]. Comparative
* - ρ* analysis was made which allowed the charge on the reaction center in the transition
state, the degree of polarity and even the rate limiting step to be determined [26]. From this, it
was discovered experimentally that the charge in the transition state is negative [26].
This allowed us to determine that the reaction center is an -carbon atom, the reaction is
electrophilic (by the substrate), the rate-determining stage is the cleavage of the C-H bond in
the -position, and the reaction proceeds through a cyclic transition state of low polarity in
which bond are being simultaneously formed and broken. All this information allowed us to
propose the possible reaction mechanism presented in Scheme 1.
The interaction begins with the chemisorption reaction between valeric acid and surface
silanols as shown in Scheme 1a. Carboxylic acids are known to react with silanols during
heating in vacuum to form carboxylates bound to the surface in monodentate fashion [60-62].
The quantum chemical calculations reported recently suggest that chemisorption of caffeic
and ferulic acids on silica is likely to proceed as an intramolecular nucleophilic substitution
[
38,39]. The energy required for the formation of the chemisorbed surface complexes
containing a Si O H cluster was reported to be rather low – 32 kJ∙mol for ferulic acid and
1
2
30 12
2
7.4 kJ∙mol for caffeic. The endothermic nature of this reaction is compensated easily by the
heating of sample up to 100°C in the mass-spectrometer ampule. Further heating drives the
ketenization of surface valerate species (Scheme 1b-d). The carbon atom located next to the
carbonyl group plays the role of the reaction center (Scheme 1c). The scission of the C-H
bond occurs earlier than the formation of a new O-H bond causing the electron density at the
reaction center to increase during the transition state (Scheme 1c). Electron-donating
substituents therefore slow down the reaction of ketene formation while electron-withdrawing
substituents accelerate it. In a similar manner, positively charged Lewis acid sites on the oxide
surfaces can accept electrons, stabilizing the transition state, and thus lower the activation
energy of the reaction (Scheme 1c).
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
1
Scheme 1. Possible mechanism for the formation of propylketene from valeric acid on oxide
surfaces.
The obtained FTIR spectroscopic data do not contradict the proposed scheme. IR absorption
-1
bands in the region of 1710-1670 cm are present in the spectra of all the oxides studied, see
Figure 8a-c. These bands are due to stretching vibrations of the carbonyl group νC=O at
-1
-1
frequencies of 1685 and 1706 cm for SiO (Fig. 8a), 1689 and 1705 cm for Al O /SiO
2
2
2
3
-1
(
Fig. 8c), and 1706 and 1685 cm for TiO /SiO (Fig. 8b). IR absorption in this area is
2 2
attributed to loosely bound forms of acid on the surfaces [63]. In addition, bending vibrations
-1
-1
of the С-Н bonds are observed at δC-H = 1465 cm (Al O /SiO ) and δ = 1462 cm
C-H
2
3
2
-1
(
TiO /SiO ) [64]. There is also a vibration at 1400 cm , which is characteristic of C-O
2 2
stretches for aliphatic acids [64]. Unlike γ-Al O (see Fig. 8e,f), no IR absorptions related to
2
3
-
the asymmetric and symmetric νCOO vibrations of surface carboxylates were observed on the
Al O /SiO (Fig. 8c) and TiO /SiO surfaces (Fig. 8b). On this basis, it can be concluded that
2
3
2
2
2
ketenization is not due to the thermal transformation of strongly bound forms of VA
carboxylates. The ketenization rather involves the loosely bound forms of VA. Thus, the IR
spectroscopic data support the reaction mechanism proposed in Scheme 1.
It was found that the activation energy for the propylketene formation reaction from VA
decreases when performed on the surfaces of Al O /SiO , TiO /SiO and γ-Al O . All these
2
3
2
2
2
2
3
oxides contain Lewis acid sites on their surfaces, which is in contrast to a silica surface that
has only Brønsted ones [55,65,66]. Positively charged Lewis acid sites act as electron
acceptors, decreasing the electron density of the reaction center in the transition state, and
hence lower the activation energy (Scheme 1c,d; Table 1).
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
2
The Taft equation can be used for correlation analysis of not only reaction rates and
equilibria, but also of any properties connected directly or indirectly with changes in free
energies. In our case, it is possible to consider a series of nano-oxides of different acidity as
an isoentropic reaction series and to evaluate their reactivity in the ketonization reaction of
valeric acid. For example, by using the modified Taft equation it is possible to approximately
estimate how many times the reaction rate on the surface of the investigated oxides increases
in comparison with the reaction rate on pure silica. Calculation showed an 11-fold increase in
the rate of ketenization on the surface of γ-Al O , a 16-fold increase on the TiO /SiO , and a
2
3
2
2
2
1-fold increase on the Al O /SiO as compared to surface of pure silica.
2 3 2
3
.3. Ketonization
Each of the studied oxides were found to be more promising catalysts for ketenization of
carboxylic acids than pure silica as this reaction was detected to proceed at lower
temperatures on their surfaces. However, products formed from a concurrent catalytic reaction
were also detected in two cases. Specifically, the reaction resulting in the ketonization of VA
was observed to occur on the γ-Al O and CeO /SiO surfaces (Fig. 2, 5).
2
3
2
2
Figures 5a and 5b show the mass spectrum and TPD curves obtained during the pyrolysis of
VA adsorbed on CeO /SiO . The molecular ion of the ketonization product – the symmetric
2
2
ketone, dibutylketone (5-nonanone), has an m/z value of 142. Its main fragment ions are
+
+
+
known to have m/z values of 57 (C H ), 58 (CH =C(CH )-OH ) and 85 (C H C≡O ) [58,59].
4
9
2
3
4
9
The ion at m/z = 58 is formed from the molecular ion of the dibutylketone as a result of
McLafferty rearrangement. This fragment ion is known to be one of the most abundant in
mass spectra of ketones. All these characteristic ions are present on Figure 5a in the correct
relative proportions – m/z = 57 (100 %), 85 (85 %), 58 (70 %), 142 (15 %). The fact that the
TPD curves of the ions with m/z = 142, 85 and 58 reflect the shape of each other makes it
possible to conclude that these fragment ions originate from the release of the same product
(
Fig. 5b). This way the Tmax of the dibutylketone formation on the surface of CeO /SiO was
2 2
found to be at 254 ºC (Fig. 5b). The ketonization reaction, also known as ketonic
decarboxylation, is always accompanied by the expulsion of water and carbon dioxide. The
intensive release of CO accompanying formation of dibutylketone is clearly visible on Figure
2
5
b.
Of particular note, the CeO /SiO sample was found to promote the ketonization reaction
2
2
selectively since no other products apart from dibutylketone were observed upon pyrolysis of
VA on its surface. This is in contrast to the γ-Al O sample, which was found to
2
3
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
3
simultaneously catalyze both reactions – ketonization and ketenization (Fig. 2). The reason
for the distinct output of pyrolytic conversion on various oxide samples is probably the
existence of diverse types of surface complexes. For example, it has been shown earlier [28]
that surface carboxylates on TiO can decompose at high temperatures via two parallel
2
pathways, depending on the structure of the surface complex. TiO forms two types of surface
2
complexes with acids and thus catalyzes two decomposition pathways – ketene formation via
monomolecular dehydration and ketone formation via bimolecular ketonization.
The analysis of the data by the Arrhenius Plot method yielded higher squared correlation
2
coefficients for the first-order reaction model (R = 0.971-0.994) than for the second-order
2
reaction model (R = 0.930-0.941) (Table 1). Consequently the first-order model corresponds
to the most feasible reaction path. Also it indicates that a mechanism involving homolytic
dissociation-recombination is unlikely to proceed as it belongs to the second-order model.
-1
Indeed, the latter mechanism requires a higher activation energy (>200 kJ mol ) and a
positive value of the activation entropy [67].
In addition, the reaction is confirmed to be first order from the results of the investigation into
the effect of the degree of the surface coverage on the reaction rate. It is known that the
reaction rate of second order reactions increases with increasing surface coverage, while the
reaction rate of the first order does not depend on the surface coverage [68]. Additional
experimental TPD curves for the fragment ion of the ketone with m/z = 85 at different surface
concentrations from 0.1 mmol/g to 1.0 mmol/g have been obtained, Fig. 9. The fact that the
shape of the curves remains unchanged and, more importantly, the shift of Tmax is not
nd
observed is a direct confirmation of the first-order reaction [68]. In the case of a 2 order
reaction the TPD curves would have been shifted to lower temperatures with the increase in
the surface coverage [68]. Along with the kinetic parameters obtained using the Arrhenius
equation, this data strongly supports the first order reaction model proposed.
Therefore, ketonization is likely to proceed as a first-order reaction with negative values for
the activation entropy. Presumably the transformation progresses as an intramolecular
reaction with a highly ordered transition state involving a 6-membered cyclic structure. Such
a highly ordered transitional configuration is likely to be acquired only when two
monodentate valerates are coordinated to the single Ce cation on the surface (Scheme 2). The
presence of such double monodentate carboxylates coordinated to the same Zr [33] or Ti
surface cations [16,32] has earlier been suggested to play the key mechanistic role in the
ketonization of carboxylic acids on the surfaces of ZrO and TiO . Kinetic parameters
2
2
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
4
obtained in our experiments support this idea and indicate that the ketonization of valeric acid
on the CeO /SiO catalyst may start from the coordination of two valerate species to the same
2
2
surface cation of cerium, see Scheme 2.
Scheme 2. Possible mechanism for the formation of dibutylketone from two monodentate
valerates coordinated to the same cerium atom on the surface of ceria nanoparticle.
Historically, several reaction mechanisms for ketonization reaction have been considered: a
radical mechanism, a mechanism involving a β-keto acid intermediate, and a concerted
mechanism [16,29,34].
The β-keto acid pathway involves the abstraction of the alpha-hydrogen of the carboxylic acid
and the formation of an anion. It then reacts with the neighboring carboxylate to produce a β-
keto acid intermediate, which converts to a ketone by releasing CO [16]. The concerted
2
mechanism on the other hand involves the simultaneous single step formation of a carbon–
carbon bond and carbon dioxide [29,34]. The latter mechanism was first proposed by Rand et
al. [30] and further developed by Renz [34], Pulido et al. [29], and Kwart and King [31].
While the majority of current experimental results support the β-keto acid mechanism, we
think that it is still too early to completely disregard the concerted mechanism in favor of the
β-keto acid route as we believe the current results still do not conclusively confirm it either. In
particular, a recently published isotope labelling study failed to confirm the beta-keto acid
mechanism [69, 70]. Specifically, the article reported by Gumidyala et al. dealt with the
ketonization of acetic acid over HZSM-5 [69]. Presented isotope labeling experiments
revealed that the rate-determining step in ketonization reaction is the formation of the C–C
bond rather than the activation of the second acid. Moreover, the lack of a primary Kinetic
Isotope Effect (KIE) upon ketonization of the deuterated acetic acid indicated that extraction
of the alpha hydrogen is not the rate determining step. Because of this it is possible to
conclude that there is no KIE confirmation of the beta-keto acid mechanism. If a primary KIE
had been detected it would have completely eliminated the possibility of the concerted
mechanism occurring. Correspondingly, the absence of a primary KIE leaves open the
possibility of the concerted mechanism occurring.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
5
In addition, previous studies on the reaction mechanism don’t include values of the pre-
exponential factor and entropy of activation for this process. By using direct experimental
data we obtained both the activation energy and activation entropy of the ketonization
reaction. The obtained kinetic parameters favor a concerted reaction mechanism via a cyclic
transition state. The beta-keto acid mechanism would have given us pre-exponential factor of
1
3
-1
9 -1
1
0 s , instead of the 10 s we obtained.
One more fact supporting the concerted mechanism is that the salts of C -C carboxylic acids,
3
6
e.g. propionate, butyrate, valerate and hexanoate, and also the yttrium carboxylates, all
decompose by preferentially forming a symmetric ketone and CO [71-73]. All these salts are
2
characterized by the monodentate type of bonding of carboxylate group to a single metal
cation atom.
3
.4. Fourier transform infrared spectroscopy of VA surface complexes
We performed FTIR analysis to shed light on the structure of surface complexes of VA. The
spectra contain IR absorption bands corresponding to two forms of bound VA – a weakly
bound and a strongly bound. In the IR spectra of VA on the surface of CeO /SiO and γ-
2
2
Al O , the absorption bands present correspond to stretching vibrations of the ionized
2
3
-
-1
carboxyl group COO (see Figure 8d-f). The absorptions at 1527 and 1423 cm correspond to
as
COO
-
s
-
asymmetric (ν
) and symmetric (ν
) vibrations of VA on the surface of CeO /SiO
COO 2 2
-1
(
Fig. 8d). The IR absorptions at 1586-1562 and 1469 cm correspond to asymmetric and
-
symmetric νCOO vibrations of VA on the surface of γ-Al O (Fig. 8e,f) [64,74].
2
3
Weakly bound forms of the acid are also present on the surface of CeO /SiO and γ-Al O
3
2
2
2
-1
(
Fig. 8d-f). Their IR absorptions lie in the region around νC=O = 1680-1690 cm . The presence
of these bands in the spectra of acetic acid on the surface of CeO and Al O is confirmed by
2
2
3
the work of other authors [63]. A recent report has also confirmed the existence of the weakly
bound acids on the surface of zirconium and hafnium oxides by using NMR techniques [75].
-1
We can hypothesize that this weakly bound form of the acid (1680-1690 cm ) exists on the
surface in the form of a complex in which two acid residues are attached to one metal atom.
However, the presence of such a complex on the surface cannot be unambiguously confirmed
by IR spectroscopy. Taking into account the high surface stability of carboxylate species (for
as
COO
-
s
-
acetates, ν
=1590-1532, ν
=1470-1420), which remain present in IR spectra even at
COO
4
00 ◦C [63], we assume that it is the weakly bound acid form which undergoes a
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
6
transformation on the surface to yield a ketone. In fact, the participation of bidentate
carboxylates in the formation of ketones was refuted in the recent work by Wang et al. [70].
Previously, the structure of carboxylate complexes has been studied by comparing the data
from IR spectroscopy and X-ray diffraction analysis [74,76-79]. These studies have shown
that the interaction between a metal atom and a carboxylate ligand can be classified under
four main types: bidentate chelating coordination, bidentate bridging coordination,
monodentate coordination and ionic interaction. More importantly, it has also been
demonstrated that the magnitude of the separation between the asymmetric and symmetric
as
COO
-
s
-
carboxylate stretches ( = ν
– ν
) or, in case of monodentate coordination, between
COO
the C=O and C-O stretches ( = νC=O – νC-O) can be used to determine the type of the
carboxylate (and non-dissociated acid) binding [80]. In terms of the magnitude of the  value
the carboxylic acid binding types can be arranged in the following order: bidentate chelating <
bidentate bridging < ionic < monodentate < non-dissociated acid [80]. The smallest values of
-1

are observed for bidentate chelating coordination ( = <110 cm ), the medium  values are
-1
typical for bidentate bridging coordination ( = 140-190 cm ) and the largest  values are
-1
distinctive of monodentate coordination ( = 200-320 cm ), which are also a typical values
for a non-dissociated acid [74,76-81,].
The more the charge is delocalized and the more symmetric the carboxyl group is, the smaller
the  magnitude becomes. For example, monodentate carboxylates are characterized by the
lowest symmetry of the carboxyl group. Correspondingly, their spectra are the most similar to
the molecular acid spectra and the  magnitude is the highest.
-1
We can assume that the IR absorption band at 1680-1690 cm is attributed to ν
vibration
C=O
-1
of the monodentately bonded acid and the band at 1413-1400 cm to its ν vibration (Fig.
C-O
-1
8
a-f). The  value in this case is equal to 267-290 cm , which completely fits into the range
-1
characteristic for monodentate type of binding ( = 200-320 cm ) [74,76-81]. Besides,
recently conducted theoretical calculations indirectly support the proposed assignment of the
IR absorption bands [70]. In that theoretical work, the IR absorption frequencies of the acetate
moieties on TiO surfaces have been calculated as a function of the degree of surface
2
coverage [70]. The value of the DFT-derived IR absorption frequency of the carbonyl group
-1
was found to be 1641-1659 cm for the monodentate form of acetic acid at a high degree of
the surface coverage. The calculated value of the carbonyl IR absorption frequency is close to
the value we observed in our spectra.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
7
It is known that utilization of the above described relationships between the carbon-oxygen
stretching frequencies and the type of carboxylate coordination has some limitations and
experimental difficulties [76]. In particular, when using the potassium bromide pellet method,
there is a risk of forming potassium salts due to ion exchange as a result of the pressures used.
For this reason we haven’t applied pressure when preparing of our FTIR samples, all the
studied materials were just mechanically mixed with KBr. To further verify the absence of
VA potassium salts we obtained an FTIR spectrum of pure VA prepared under the same
conditions, i.e. mixing VA with KBr without using the press. In the obtained spectrum in the
-1
range of 2100-1300 cm there are no absorption bands corresponding to potassium valerate.
s
-1
s
Instead, there are IR absorption bands of the С-Н bonds vibrations (δ
= 1463 cm , δ
=
CH3
CH2
-1
1
381 cm ) and the bands corresponding to the carboxyl groups of VA dimers (ν
= 1713
C=O
-1
-1
cm ; ν = 1415 cm ).
C-O
Carboxylic acids form dimers due to the formation of one of the strongest donor-acceptor
hydrogen bonds [64]. This can be confirmed by a shift in the IR absorption frequency (νC=O

-1
-1
1
713 cm ) of the dimer relative to the monomer (ν
 1760 cm , solution in CCl ) [64].
C=O
4
-1
The calculated value of  for the dimer of VA was found to be 298 cm .
Based on the values of  given in the Table 2, it can be assumed that monodentately bound
complexes of VA are present on the surfaces of all the oxides studied (Scheme 3a, Figure 8a-
f). Also, it is clear that bidentate chelate complexes are present on the surfaces of γ-Al O and
2
3
CeO /SiO (Table 2, Scheme 3c, Figure 8d-f). With increasing VA concentration on the
2
2
surface of γ-Al O , the amount of bidentately bridged VA complexes increases (Scheme 3b,
2
3
Figure 8f).
Scheme 3. The possible structures of VA complexes on the oxide surfaces, as revealed by
FTIR spectroscopic analysis.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
8
Observed weakly-bound surface complexes of VA may play an essential role in allowing the
ketonization and ketenization reactions to occur. These complexes, in contrast to bidentately
chelated and bidentately bridged ones, are more labile and not so strongly bound to the
surface. Correspondingly, less energy is needed for them to acquire the transition state
conformation. This idea can be rationalized by employing the Sabatier principle, specifically,
by assuming that the ketonization and ketenization reactions proceed via transformation of the
precursor surface complexes that are not coordinated too weakly, nor too strongly. In such a
case, the ketonization reaction may occur from the thermal transformation of monodentate
complexes in which two carboxylic acids are attached to the single metal atom, whereas
ketenization may be associated with transformations of a monodentate complex in which one
acid residue is attached to the metal atom. Following this logic, the fact that the CeO2-
containing catalyst yields predominantly ketone may be explained by the relatively high
coordination number of this metal.
However, further spectroscopic evidence will be necessary to unambiguously confirm the role
of different weakly-bound forms of VA in ketenization and ketonization reactions on the
oxide surfaces. In particular, it will be imperative to conduct FTIR spectroscopic studies of
the thermal evolution of carboxylic acid complexes as a function of the degree of surface
coverage. Also, it will be of crucial importance to use quantum-chemical calculations taking
into account the geometry and the distances between atoms in the reaction center on the
catalytic surfaces.
Conclusions
In the present study, pyrolytic decomposition of VA adsorbed on the surfaces of SiO , γ-
2
Al O , CeO /SiO , Al O /SiO and TiO /SiO was studied by means of TPD MS. Two
2
3
2
2
2
3
2
2
2
reaction pathways of VA conversion were found to be dominant – ketenization to form
propylketene and ketonization to form dibutylketone.
Propylketene formation was observed on the surfaces of all the oxides investigated except for
CeO /SiO . The kinetic parameters of the ketene formation reaction were calculated. The
2
2
activation energy of the ketenization reaction on the surface of the studied oxides decreases in
the order: SiO > γ-Al O > TiO /SiO > Al O /SiO .
2
2
3
2
2
2
3
2
The reaction resulting in the formation of the symmetrical ketone, dibutylketone, formation
was observed to proceed on the surfaces of γ-Al O and CeO /SiO . CeO /SiO oxide was
2
3
2
2
2
2
found to catalyze the formation of dibutylketone selectively as no other products, including
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
1
9
ketene, were registered upon thermal decomposition of VA on its surface. The ketonization
reaction on the surface of CeO /SiO was also found to proceed at a significantly lower
2
2
temperature (Tmax = 254 °C) in comparison to the γ-Al O surface (T = 315 °C).
2
3
max
The kinetic parameters of the dibutylketone formation reaction were calculated. The first
order of ketonization reaction was confirmed by using the Arrhenius Plot method and by
investigating the effect of the degree of the surface coverage on the reaction rate. The
calculated kinetics parameters of the ketonization reaction support the concerted mechanism
as a possible pathway for ketone formation. The concerted mechanism is one of the currently
debated mechanisms for the ketonic decarboxylation reaction, even though the majority of
modern experimental results support an alternative route – the β-keto acid mechanism.
The structure of VA complexes on the oxide surfaces was investigated by using FTIR
spectroscopy. Weakly bound forms of VA were found to be present on all the studied oxide
surfaces. These weakly bound surface complexes may be responsible for the ketonization and
ketenization reactions proceeding.
Acknowledgements
Acknowledgment: This work was supported by the Swedish Research Council (VR) [contract
3
48-2014-4250]. T.K. acknowledges the Swedish Institute for research scholarship within
Visby Programme [03804/2016].
References
[
1] M.J. Climent, A. Corma, S. Iborra, Conversion of biomass platform molecules into fuel
additives and liquid hydrocarbon fuels, Green Chem. 16 (2014) 516–547.
[
2] A. Corma, S. Iborra, A. Velty, Chemical Routes for the Transformation of Biomass into
Chemicals, Chem. Rev. 107 (2007) 2411–2502.
[
3] G.W. Huber, S. Iborra, A. Corma, Synthesis of transportation fuels from biomass:
chemistry, catalysts, and engineering, Chem. Rev. 106 (2006) 4044–4048.
[
4] R.A. Sheldon, Green chemistry, catalysis and valorization of waste biomass, J. Mol. Catal.
A: Chem. 422 (2016) 3–12.
[
5] M. Besson, P. Gallezot, C. Pinel, Conversion of biomass into chemicals over metal
catalysts, Chem. Rev. 114 (2014) 1827−1870.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
0
[
(
6] P. Gallezot, Conversion of biomass to selected chemical products, Chem. Soc. Rev. 41
2012) 1538–1558
[
7] D.A. Bulushev, J.R.H. Ross, Catalysis for conversion of biomass to fuels via pyrolysis and
gasification: A review, Catal. Today 171 (2011) 1–13.
[
8] D.W. Rackemann, W.O.S. Doherty, The conversion of lignocellulosics to levulinic acid,
Biofuels, Bioprod. Biorefin. 5 (2011) 198–214.
[
9] J.P. Lange, Lignocellulose conversion: an introduction to chemistry, process and
economics, Biofuels, Bioprod. Biorefin. 1 (2007) 39–48.
[
10] J.P. Lange, R. Price, P. M. Ayoub, J. Louis, L. Petrus, L. Clarke, H. Gosselink, Valeric
Biofuels: A Platform of Cellulosic Transportation Fuels, Angew. Chem., Int. Ed. 49 (2010)
479–4483.
4
[
11] K. Kon, W. Onodera, K. Shimizu, Selective hydrogenation of levulinic acid to valeric
acid and valeric biofuels by a Pt/HMFI catalyst, Catal. Sci. Technol. 4 (2014) 3227–3234.
[
12] L.L. Dong, L. He, G.-H. Tao, C. Hu, High yield of ethyl valerate from the esterification
of renewable valeric acid catalyzed by amino acid ionic liquids, RSC Advances 3 (2013)
806–4813.
4
[
13] L. Xin, Z. Zhang, J. Qi, D.J. Chadderdon, Y. Qiu, K.M. Warsko, W. Li, Electricity
Storage in Biofuels: Selective Electrocatalytic Reduction of Levulinic Acid to Valeric Acid or
γ-Valerolactone, ChemSusChem 6 (2013) 674–686.
[
14] D. Mohan, C.U. Pittman, P.H. Steele, Pyrolysis of wood/biomass for bio-oil:ꢀ a critical
review, Energy Fuels 20 (2006) 848–889.
[
15] G. Varhegyi, M.J. Antal, E. Jakab, P. Szabó, Kinetic modeling of biomass pyrolysis, J.
Anal. Appl. Pyrolysis 42 (1997) 73-87.
[
16] T.N. Pham, T. Sooknoi, S.P. Crossley, D.E. Resasco, Ketonization of carboxylic acids:
mechanisms, catalysts, and implications for biomass conversion, ACS Catalysis 3 (2013)
456–2473.
2
[
17] A. Kubátová, J. Št’ávová, W.S. Seames, Y. Luo, S.M. Sadrameli, M.J. Linnen, G.V.
Baglayeva, I.P. Smoliakova, E.I. Kozliak, Triacylglyceride thermal cracking: pathways to
cyclic hydrocarbons, Energy Fuels 26 (2011) 672–685.
[
18] A.V. Bridgwater, Principles and practice of biomass fast pyrolysis processes for liquids,
J. Anal. Appl. Pyrolysis 51 (1999) 3–22.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
1
[
19] Y. Wan, P. Chen, B. Zhang, C. Yang, Y. Liu, X. Lin, R. Ruan, Microwave–assisted
pyrolysis of biomass: catalysts to improve product selectivity, J. Anal. Appl. Pyrolysis 86
(2009) 161–167.
[
20] J.C. Serrano-Ruiz, D.J. Braden, R.M. West, J.A. Dumesic, Conversion of cellulose to
hydrocarbon fuels by progressive removal of oxygen, Appl. Catal. B 100 (2010) 184–189.
[
21] V.V. Brei, V.M. Gun'ko, V.V. Dudnik, A.A. Chuiko, Study of kinetics and mechanisms
of some unimolecular reactions on silica surfaces, Langmuir 8 (1992) 1968–1973.
[
22] M.C. Libby, P.C. Watson, M.A. Barteau, Synthesis of ketenes with oxide catalysts, Ind.
Eng. Chem. Res. 33 (1994) 2904–2912.
[
23] A.J. Becerra, R. Martinez, M.C. Huff, M.A. Barteau, Catalytic synthesis of butylketene
and hexylketene from hexanoic acid and octanoic acid on functionalized silica monoliths at
short contact times, Catal. Today 107–108 (2005) 244–249.
[
24] R. Martinez, M.C. Huff, M.A. Barteau, Synthesis of ketenes from carboxylic acids on
functionalized silica monoliths at short contact times, Appl. Catal. A 200 (2000) 79–88.
[
25] L.R. Azizova, T.V. Kulik, B.B. Palianytsia, N.A. Lipkovska, Thermal and hydrolytic
stability of grafted ester groups of carboxylic acids on the silica surface, J. Therm. Anal.
Calorim. 122 (2015) 517–523.
[
26] T.V. Kulik, Use of TPD–MS and Linear Free Energy Relationships for assessing the
reactivity of aliphatic carboxylic acids on a silica surface, J. Phys. Chem. C 116 (2011) 570–
80.
5
[
27] M.A. Barteau, Organic reactions at well–defined oxide surfaces, Chem. Rev. 96 (1996)
413–1430.
1
[
28] R. Martinez, M.C. Huff, M.A. Barteau, Ketonization of acetic acid on titania–
functionalized silica monoliths, J. Catal. 222 (2004) 404–409.
[
29] A. Pulido, B. Oliver-Tomas, M. Renz, M. Boronat, A. Corma, Ketonic decarboxylation
reaction mechanism: a combined experimental and DFT study, ChemSusChem 6 (2013) 141–
51.
1
[
30] L. Rand, W. Wagner, P.O. Warner, L.R. Kovac, Reactions catalyzed by potassium
fluoride. II. The conversion of adipic acid to cyclopentanone, J. Org. Chem. 27 (1962) 1034–
035.
1
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
2
[
31] H. Kwart, K. King, Rearrangement and cyclization reactions of carboxylic acids and
esters, in: S. Patai (Ed.), The Chemistry of Carboxylic Acids and Esters, Interscience,
London, 1969, pp. 341–373.
[
32] K.S. Kim, M.A. Barteau, Structure and composition requirements for deoxygenation,
dehydration, and ketonization reactions of carboxylic acids on TiO (001) single–crystal
2
surfaces, J. Catal. 125 (1990) 353–375.
[
33] V.N. Panchenko, Y.A. Zaytseva, M.N. Simonov, I.L. Simakova, E.A. Paukshtis,
DRIFTS and UV–vis DRS study of valeric acid ketonization mechanism over ZrO in
2
hydrogen atmosphere, J. Mol. Catal. A: Chem. 388–389 (2014) 133–140.
[
34] M. Renz, Ketonization of carboxylic acids by decarboxylation: mechanism and scope,
Eur. J. Org. Chem. 2005 (2005) 979–988.
[
35] O. Nagashima, S. Sato, R. Takahashi, T. Sodesawa, Ketonization of carboxylic acids
over CeO –based composite oxides, J. Mol. Catal. A: Chem. 227 (2005) 231–239.
2
[
36] G. Pacchioni, Ketonization of Carboxylic Acids in Biomass Conversion over TiO and
2
ZrO Surfaces: A DFT Perspective, ACS Catalysis, 4 (2014) 2874–2888.
2
[
37] T.V. Kulik, V.N. Barvinchenko, B.B. Palyanitsa, O.V. Smirnova, V.K. Pogorelyi, A.A.
Chuiko, A desorption mass spectrometry study of the interaction of cinnamic acid with a
silica surface, Russ. J. Phys. Chem. 81 (2007) 83–90.
[
38] T.V. Kulik, N.A. Lipkovska, V.N. Barvinchenko, B.B. Palyanytsya, O.A. Kazakova,
O.A. Dovbiy, V.K. Pogorelyi, Interactions between bioactive ferulic acid and fumed silica by
UV–vis spectroscopy, FT–IR, TPD MS investigation and quantum chemical methods, J.
Colloid Interface Sci. 339 (2009) 60–68.
[
39] T.V. Kulik, N.O. Lipkovska, V.M. Barvinchenko, B.B. Palyanytsya, O.A. Kazakova,
O.O. Dudik, A. Menyhárd, K. László, Thermal transformation of bioactive caffeic acid on
fumed silica seen by UV–Vis spectroscopy, thermogravimetric analysis, temperature
programmed desorption mass spectrometry and quantum chemical methods, J. Colloid
Interface Sci. 470 (2016) 132–141.
[
40] T.V. Kulik, V.N. Barvinchenko, B.B. Palyanytsya, N.A. Lipkovska, O.O. Dudik,
Thermal transformations of biologically active derivatives of cinnamic acid by TPD MS
investigation, J. Anal. Appl. Pyrolysis 90 (2011) 219–223.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
3
[
41] M. Kraus, Linear Correlations of Substrate Reactivity in Heterogeneous Catalytic
Reactions, in: H.P. D.D. Eley, B.W. Paul (Eds.), Advances in Catalysis, Academic Press, New
York, 1967, pp. 75–102.
[
[
42] P.A. Redhead, Thermal desorption of gases, Vacuum 12 (1962) 203–211.
43] V.M. Gun'ko, J.P. Blitz, K. Gude, V.I. Zarko, E.V. Goncharuk, Y.M. Nychiporuk, R.
Leboda, J. Skubiszewska–Zięba, V.D. Osovskii, Y.G. Ptushinskii, O.A. Mishchuk, S.V.
Pakhovchishin, P.P. Gorbik, Surface structure and properties of mixed fumed oxides, J.
Colloid Interface Sci. 314 (2007) 119–130.
[
44] V.M. Gun’ko, Y.M. Nychiporuk, V.I. Zarko, E.V. Goncharuk, O.A. Mishchuk, R.
Leboda, J. Skubiszewska–Zięba, E. Skwarek, W. Janusz, G.R. Yurchenko, V.D. Osovskii,
Y.G. Ptushinskii, V.V. Turov, P.P. Gorbik, J.P. Blitz, K. Gude, Relationships between surface
compositions and properties of surfaces of mixed fumed oxides, Appl. Surf. Sci. 253 (2007)
3
215–3230.
[
45] V.M. Gun'ko, V.I. Zarko, V.V. Turov, R. Leboda, E. Chibowski, E.M. Pakhlov, E.V.
Goncharuk, M. Marciniak, E.F. Voronin, A.A. Chuiko, Characterization of fumed
alumina/silica/titania in the gas phase and in aqueous suspension, J. Colloid Interface Sci. 220
(1999) 302–323.
[
46] V.I. Zarko, V.M. Gun'ko, E. Chibowski, V.V. Dudnik, R. Lèboda, Study of surfaces
properties of fumed alumina/silica materials, Colloids Surf. A 127 (1997) 11–18.
47] V.M. Gun'ko, V.I. Zarko, E. Chibowski, V.V. Dudnik, R. Leboda, V.A. Zaets, Structure
[
of fumed titania and silica/titania and influence of the nature of surface sites on interaction
with water, J. Colloid Interface Sci. 188 (1997) 39–57.
[
48] M.V. Borysenko, K.S. Kulyk, M.V. Ignatovych, E.N. Poddenezhny, A.A. Boiko, A.O.
Dobrodey, Sol–Gel Synthesis of Silica Glasses, Doped with Nanoparticles of Cerium Oxide,
in: A.P. Shpak, P.P. Gorbyk (Eds.), Nanomaterials and Supramolecular Structures, Springer
Dordrecht, 2009, pp. 227–233.
[
49] K. Kulyk, V. Ishchenko, B. Palyanytsya, V. Khylya, A TPD-MS study of the interaction
of coumarins and their heterocyclic derivatives with a surface of fumed silica and nanosized
oxides CeO /SiO , TiO /SiO , Al O /SiO , J. Mass Spectrom. 45 (2010) 750–761.
2
2
2
2
2
3
2
[
50] K. Kulyk, M. Borysenko, T. Kulik, L. Mikhalovska, J.D. Alexander, B. Palianytsia,
Chemisorption and thermally induced transformations of polydimethylsiloxane on the surface
of nanoscale silica and ceria/silica, Polym. Degrad. Stab. 120 (2015) 203–211.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
4
[
51] S.I. Nicholl, J.W. Talley, Development of thermal programmed desorption mass
spectrometry methods for environmental applications, Chemosphere 63 (2006) 132–141.
[
52] W. Rudziński, T. Borowiecki, A. Dominko, T. Pańczyk, A New Quantitative
interpretation of temperature–programmed desorption spectra from heterogeneous solid
Surfaces, based on statistical rate theory of interfacial transport:ꢀ the effects of simultaneous
readsorption, Langmuir 15 (1999) 6386–6394.
[
53] R.W. Taft Jr, Polar and steric substituent constants for aliphatic and o-Benzoate groups
from rates of esterification and hydrolysis of esters1. J. Am. Chem. Soc. 74 (1952) 3120-
128.
3
[
54] P. Sykes, A Guidebook to Mechanism in Organic Chemistry, sixth ed., Longman, New
York, 1991.
[
55] L.T. Zhuravlev, The surface chemistry of amorphous silica. Zhuravlev model, Colloids
Surf. A 173 (2000) 1–38.
[
56] V.M. Gun'ko, V.I. Zarko, B.A. Chuikov, V.V. Dudnik, Yu.G. Ptushinskii, E.F. Voronin,
E.M. Pakhlov, A.A. Chuiko, Temperature-programmed desorption of water from fumed
silica, silica/titania, and silica/alumina, Int. J. Mass Spectrom. 172 (1998) 161-179.
[
57] K.Kulyk, M. Borysenko, T. Kulik, L. Mikhalovska, J.D. Alexander, B. Palyanytsya,
Chemisorption and thermally induced transformations of polydimethylsiloxane on the surface
of nanoscale silica and ceria/silica, Polym. Degrad.Stabil. 120 (2015) 203-211.
[
[
58] A.T. Lebedev, Mass spectrometry in organic chemistry, Binom, Moscow, 2003.
59] E. Stein in NIST ChemistryWebBook. IR and Mass Spectra, Database Number 69 (Eds.:
W.G.Mallard, P.J. Linstrom), National Institute of Standards and Technology, Gaithersburg,
MD, 2000, http://webbook.nist.gov.
[
60] R.P. Young, Infrared spectroscopic studies of adsorption and catalysis. Part 3.
Carboxyllic acids and their derivatives adsorbed on silica, Can. J. Chem. 47 (1969) 2237–
247.
2
[
61] K. Marshall, C.H. Rochester, Infrared study of the adsorption of oleic and linolenic acids
onto the surface of silica immersed in carbon tetrachloride, J. Chem. Soc. Faraday Trans. 1 71
(1975) 1754–1761.
[
62] V.A. Basyuk, Infrared spectra of carboxylic compounds on silica surfaces at 1500–1800
–
1
cm , J. Appl. Spectrosc. 60 (1994) 29–33.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
5
[
63] M.A. Hasan, M.I. Zaki, L. Pasupulety, Oxide-catalyzed conversion of acetic acid into
acetone: an FTIR spectroscopic investigation, Appl. Catal. A 243 (2003) 81–92.
[
64] L.J. Bellamy, The Infrared Spectra of Complex Molecules, vol. 1, Chapman & Hall,
London, 1975.
[
[
65] K. Tanabe, Solid acids and bases: their catalytic properties, Kodansha, Tokyo, 1970.
66] L.F. Sharanda, A.P. Shimansky, T.V. Kulik, A.A. Chuiko, Study of acid–base surface
properties of pyrogenic γ–aluminium oxide, Colloids Surf. A 105 (1995) 167–172.
[
67] F.A. Carey, R.J. Sundberg, Advanced Organic Chemistry, Plenum Press, New York,
977.
1
[
68] D.P. Woodruff, T.A. Delchar, Modern Techniques of Surface Science, Cambridge
University Press, London, 1986.
[
69] A. Gumidyala, T. Sooknoi and S. Crossley, Selective ketonization of acetic acid over
HZSM-5: The importance of acyl species and the influence of water, J. Catal. 340 (2016) 76-
4.
8
[
70] S.Wang, E. Iglesia, Experimental and theoretical assessment of the mechanism and site
requirements for ketonization of carboxylic acids on oxides, J. Catal. 345 (2017) 183–206.
[
71] J.-C. Grivel, Y. Zhao, X. Tang, P.G.P.A. Pallewatta, A. Watenphul, M.V. Zimmermann,
Thermal decomposition of yttrium (III) valerate in argon, J. Anal. Appl. Pyrolysis 106 (2014)
25–131.
1
[
72] J.-C. Grivel, M.J.S. Guevara, F. Attique, Y. Zhao, X. Tang, P.G.P.A. Pallewatta, A.
Watenphul, M.V. Zimmermann, Thermal decomposition of yttrium (III) hexanoate in argon,
J. Anal. Appl. Pyrolysis 112 (2015) 237–243.
[
73] J.-C. Grivel, Thermal decomposition of yttrium (III) propionate and butyrate, J. Anal.
Appl. Pyrolysis 101 (2013) 185–192.
[
74] K. Nakamoto, Infrared and Raman spectra of inorganic and coordination compounds,
John Wiley & Sons, New York, 1997.
[
75] J. De Roo, E. A. Baquero, Y. Coppel, K. De Keukeleere, I. Van Driessche, C. Nayral,
Insights into the Ligand Shell, Coordination Mode, and Reactivity of Carboxylic Acid Capped
Metal Oxide Nanocrystals, ChemPlusChem 81 (2016) 1216-1223.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
6
[
76] G.B. Deacon, R.J. Philips, Relationships between the carbon-oxygen stretching
frequencies of carboxylato complexes and the type of carboxylate coordination, Coord. Chem.
Rev. 33 (1980) 227-250.
[
77] B.S. Manhas, A.K. Trikha, Relationships between the direction of shifts in the carbon-
oxygen stretching frequencies of carboxylato complexes and the type of carboxylate
coordination, J. Ind. Chem. Soc. 59 (1982) 315-319.
[
78] R.C. Mehrotra, and B.R. Bohra. Metal carboxylates, Academic Press, 1983.
[
79] V. Zeleňák, Z. Vargová, K. Györyová, Correlation of infrared spectra of zinc (II)
carboxylates with their structures, Spectrochim. Acta Part A: Mol. Biomol. Struct. 66 (2007)
62-272.
2
[
80] E.G. Palacios, G. Juarez-Lopez, A.J. Monhemius, Infrared spectroscopy of metal
carboxylates, II. Analysis of Fe(III), Ni and Zn carboxylate solutions, Hydrometallurgy 72
(2004) 139–148.
[
81] N. Wu, L. Fu, M. Su, M. Aslam, K.Ch. Wong, and V.P. Dravid, Interaction of Fatty Acid
Monolayers with Cobalt Nanoparticles, Nano Lett. 4 (2004) 383-386.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
7
Table 1. Kinetic parameters (temperature of the maximum desorption rate T , reaction order
max
≠
≠
n, activation energy E , pre-exponential factor ν , and activation entropy dS ) of the pyrolytic
0
reactions of valeric acid on the nanosized γ-Al O , CeO /SiO , Al O /SiO and TiO /SiO .
2
3
2
-1
2
2
3
2
2
2
≠
≠
2[a]
Reaction Tmax
m/z
n
E
ν [s ], (n=1)
dS
R
0
-1
-1 -1
-1
-1
Scheme [°C]
[kJ∙mol ]
ν [l∙mol ∙s ], (n=2) [cal∙К ∙mol ]
0
TiO /SiO (propylketene formation)
2
2
5
1
275
84
1
85
5.1101
1.810
-34
6
0.955
0.907
4
2
3
2
9
-
2
169
Al O /SiO (propylketene formation)
2
3
2
5
1
266
84
1
80
2.1101
-35
-3
0.951
0.920
-
2
145
2.010
γ-Al O (propylketene formation)
2
3
5
1
285
84
1
87
6.6101
-33
3
0.953
0.927
-
2
166
5.110
CeO /SiO (dibutylketone formation)
2
2
9
2
254
142
1
122
6.4 102
8.410
-15
45
0.959
0.921
-
2
250
γ-Al O (dibutylketone formation)
2
3
9
2
315
142
1
131
2.8101
-12
28
0.994
0.938
-
2
231
1.510
[
a]
Squared coefficient of determination.
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
8
as
s
Table 2. Wavenumber of the carboxyl group stretches, the values of  = ν COO- – ν COO- and

= νC=O-νC-O.
as
-
COO ,
-1
s
-
COO ,
-1
Oxide
νC=O
νC-O
ν
ν
 = νC=O-ν
;
Type of
interaction
C-O
s
-
1
-1
as
(cm )
(cm ) (cm )
(cm )

= ν
-ν
cm )
COO
COO
-
1
(
SiO2
1685
1685
1689
1400
1400
1400
-
-
-
-
-
-
285
285
289
monodentate
monodentate
monodentate
monodentate
bidentate
TiO /SiO2
2
Al O /SiO
2
3
2
γ-Al O3
1680-1690 1413
267-277
116
2
1
1
586
562
1470
1470
bridging
bidentate
92
chelating
CeO /SiO
1685
1713
1408
1415
277
85
monodentate
bidentate
chelating
2
2
1
526
-
1441
-
VA dimer
298
H-bonding
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.201601370
2
9
TABLE, SCHEME AND FIGURE CAPTIONS
Table 1. Kinetic parameters (temperature of the maximum desorption rate Tmax, reaction order
≠
≠
n, activation energy E , pre-exponential factor ν , and activation entropy dS ) of the pyrolytic
0
reactions of valeric acid on the nanosized γ-Al O , CeO /SiO , Al O /SiO and TiO /SiO .
2
3
2
2
2
3
2
2
2
as
s
Table 2. Wavenumber of the carboxyl group stretches, the values of  = ν COO- – ν COO- and
= νC=O-νC-O

.
Figure 1. (a) Vapor pressure measured as a function of temperature (P-T) for the samples of
-1
nano-oxides impregnated with valeric acid (0.3 mmol∙g ); (b) TPD MS curves of water (m/z
-1
=
18) desorption for the samples of nano-oxides impregnated with valeric acid (0.3 mmol∙g ).
Figure 2. Mass spectrum at 300 °C (a) and TPD MS curves (b) for a sample of γ-Al O
2
3
-1
impregnated with valeric acid (0.3 mmol∙g ). Numbers above the curves correspond to the
m/z values, and drawings to the supposed structures of the ions.
Figure 3. Mass spectrum at 260 °C (a) and TPD MS curves (b) for a sample of Al O /SiO
2
2
3
-1
impregnated with valeric acid (0.3 mmol∙g ). Numbers above the curves correspond to the
m/z values, and drawings to the supposed structures of the ions.
Figure 4. Mass spectrum at 273 °C (a) and TPD MS curves (b) for a sample of TiO /SiO
2
2
-1
impregnated with valeric acid (0.3 mmol∙g ). Numbers above the curves correspond to the
m/z values, and drawings to the supposed structures of the ions.
Figure 5. Mass spectrum at 255 °C (a) and TPD MS curves (b) for a sample of CeO /SiO
2
2
-1
impregnated with valeric acid (0.3 mmol∙g ). Numbers above the curves correspond to the
m/z values, and drawings to the supposed structures of the ions.
Figure 6. TPD MS curves for molecular ions with m/z = 84 (propylketene) obtained during
the pyrolysis of valeric acid impregnated on the nanosized SiO , γ-Al O , Al O /SiO and
2
2
3
2
3
2
-1
TiO /SiO (0.3 mmol∙g ).
2
2
Figure 7. TPD MS curves for molecular ions with m/z = 142 (dibutylketone) obtained during
the pyrolysis of valeric acid impregnated on the nanosized γ-Al O and CeO /SiO (0.3
2
3
2
2
-1
mmol∙g ).
This article is protected by copyright. All rights reserved.