Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over platinum catalysts

Article abstract of DOI:10.1016/j.cattod.2015.09.055

The total hydrodeoxygenation of furfural-acetone condensation adduct (F₂A) for obtaining tridecane is studied in this work. Three different Pt catalysts (using alumina, activated carbon, and graphite-MgZr oxide composite as supports) were tested using acetone as solvent (4.5 mmol/L of adduct) in a stirred batch reactor at 493 K and 5.5 MPa. Best results were obtained with Pt/Al₂O₃, yielding 21.5% of n-tridecane after 24 h reaction time, with carbon balances close to 96%. The performance of the carbon supported catalysts was poorer (both in terms of conversion, tridecane selectivity and carbon mass balance closure) mainly because of the strong adsorption of reactants and reaction intermediates, whereas the MgZr-HSAG also present activity for the undesired cleavage of C-C bonds of the condensation adducts. A kinetic model, considering serial-parallel reaction steps and first order dependence on the organic reactant has been successfully applied for modelling the results obtained with the three catalysts. The dependence of the kinetic constants on the catalyst properties suggest that metal dispersion and the concentration of weak acid sites are the main parameters affecting catalyst performance.

Full text of DOI:10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
Catalysis Today xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Hydrodeoxygenation of furfural-acetone condensation adducts to
tridecane over platinum catalysts
∗
Laura Faba, Eva Díaz, Aurelio Vega, Salvador Ordó n˜ ez
Dep. of Chemical, Environmental Engineering, Faculty of Chemistry; University of Oviedo, Julián Clavería 8, Oviedo 33006, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 28 June 2015
Received in revised form 8 September 2015
Accepted 27 September 2015
Available online xxx
The total hydrodeoxygenation of furfural-acetone condensation adduct (F A) for obtaining tridecane is
studied in this work. Three different Pt catalysts (using alumina, activated carbon, and graphite-MgZr
2
oxide composite as supports) were tested using acetone as solvent (4.5 mmol/L of adduct) in a stirred
batch reactor at 493 K and 5.5 MPa. Best results were obtained with Pt/Al2O3, yielding 21.5% of n-tridecane
after 24 h reaction time, with carbon balances close to 96%. The performance of the carbon supported
catalysts was poorer (both in terms of conversion, tridecane selectivity and carbon mass balance closure)
mainly because of the strong adsorption of reactants and reaction intermediates, whereas the MgZr-
HSAG also present activity for the undesired cleavage of C C bonds of the condensation adducts. A kinetic
model, considering serial-parallel reaction steps and first order dependence on the organic reactant has
been successfully applied for modelling the results obtained with the three catalysts. The dependence of
the kinetic constants on the catalyst properties suggest that metal dispersion and the concentration of
weak acid sites are the main parameters affecting catalyst performance.
Keywords:
Biomass-to-fuels
Hydrogenation
Diesel fuels
Biomass upgrading
©
2015 Elsevier B.V. All rights reserved.
1
. Introduction
activity of different mixed oxides and other basic catalysts in the
aldol condensation of furfural and acetone is the topic of many
papers, concluding that Mg-Zr mixed oxide is one of the most
promising materials, reaching condensation yields higher than 80%
at mild conditions [7,8]. Besides, this material yields higher selec-
tivities to C13, increasing the quality of the resulting drop-in fuel
upon hydrogenation of the resulting adducts.
Production of renewable liquid fuels from biomass resources is
nowadays of key interest, especially the promising aqueous-phase
routes [1]. Among them, the process suggested by Dumesic and co-
workers, proposing a catalytic process that allows obtaining diesel
from biomass, is specially promising, since it does not require high
temperatures, improving the global economy of the process [2,3].
The process sequence is summarized in Scheme 1. As it can be
observed, it is a sequence of four steps with different catalysts at
different reaction conditions: the hydrolysis of the sugar polymers;
followed by the dehydration of sugars yielding aldehydes; the con-
densation of these compounds using a linking molecule (usually
acetone); and the total hydrodeoxygenation of these condensation
adducts to the final lineal hydrocarbons, with similar properties as
the mineral diesel.
Considering the promising results of these first studies, the
efforts to improve this process have grown exponentially. In good
agreement, the mechanism of sugar hydrolysis and dehydration
has been deep studied, developing active and selective hetero-
geneous catalysts that allow optimizing the reaction conditions
under the green-chemistry principles [4–6]. In the same way, the
The combustion enthalpy of these condensation adducts is
very low because of the high oxygen content, which markedly
decreases its quality as fuel and promotes corrosion problems [9].
Hydrodeoxygenation (HDO) is the most effective technology for
increasing this combustion enthalpy by hydrogenation of double
bonds and removing the oxygen as water at medium pressure
and temperature [10]. This reaction was previously applied to oil-
derivatives, and the references of applying this process to biomass
upgrading have increased in the last ten years [11]. However, the
hydrodeoxygenation of the furfural-acetone condensation adducts
is much more difficult because of the complexity of the parent
molecules, with different functional groups. Different successive
steps are required for removing oxygen atoms (ketonic and cyclic
ones), to hydrogenate the aliphatic chain and the furan-ring unsat-
urations as well as to open these rings. In addition, these steps
must take place without breaking the carbon chain. The complexity
of these individual reactions is even higher due to steric interac-
tions among the different functional groups. As consequence, a very
limited number of studies about the total HDO of these compounds
∗ Corresponding author. Tel.: +34 985 103 437; fax: +34 985 103434.
E-mail address: sordonez@uniovi.es (S. Ordó n˜ ez).
http://dx.doi.org/10.1016/j.cattod.2015.09.055
0
920-5861/© 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
2
Scheme 1. General mechanism for biodiesel synthesis from hemicellulosic biomass by dehydration-aldolization-hydrogenation/deoxygenation [3].
have been published [12–14]. Despite of the few references, noble
metals (mainly Pt, Pd and Ru) seem to be the optimum catalysts for
these reactions. However, each individual step is controlled by dif-
ferent parameters, so the role of acidity and metal dispersion in the
global process is not negligible [15]. The reaction conditions must
be well defined when these noble metals are used, in order to pre-
vent side-reactions such as decarboxylation and decarbonylation
and active carbons (AC, GF 40) were stocked by Timcal S.A. and
Norit, respectively. In order to study the possibility of implement
the condensation and hydrodeoxygenation as one-step process,
Pt/MgZr/HSAG500 was also prepared, by impregnation of Pt over
the MgZr/HSAG prepared by co-precipitation, as it is detailed in
previous work [8]. The possibility of carry out the condensation and
hydrodeoxygenation steps as on-pot process was previously stud-
ied, leading to important economic and technical improvements
[18].
[
16] that would decrease the carbon chain length, decreasing the
diesel quality.
Previous works of our research group are focused in the
hydrodeoxygenation of the first condensation adduct (C8) obtained
by the reaction between furfural and acetone. Pt was identified as
the most active noble metal, obtaining more than 25% of n-octane
after 24 h of reaction [17]. The reaction mechanism, the kinetic and
the stability were also studied, identified the role of different mor-
phological and surface catalytic properties in each reaction step.
Consequently, different supports were tested concluding that the
metal dispersion is the key parameter of this reaction, obtaining
good results with supports as different as alumina, active carbon
and high surface area graphites [15].
Pt-catalysts were prepared by ion exchange, using metal chlo-
rides as precursor (Pt(NH ) Cl , supplied by Aldrich). The solutions
3
4
2
were prepared considering a final metal loading of 1%. The ion
exchange was carried out in a rotavapor at 343 K for 24 h, after
which the solvent was removed using a vacuum system. Resulting
solids were washed, filtered, dried at 373 K and heating in He flow at
a temperature rate of 5 K/min until 873 K, holding this temperature
during 2 h. All the materials were activated before any character-
ization or activity test at 473 K for 2 h under flowing hydrogen
gas.
The aim of this paper is to study the hydrodeoxygenation of
the second adduct: the C13-condensated (1,5-bis(2-furanyl)-1,4-
pentadien-3-one), which presents the highest practical interest.
Reaction conditions were chosen considering the best results
obtained in the HDO of the first adduct (493 K; 2.0 MPa of H2
with a total pressure of 5.5 MPa, 24 h) using Pt catalyst. Despite
the higher complexity of this molecule, there are same functional
groups in both cases, so it could be predicted that main effect
of this complexity should be in the reaction time but not in the
reaction temperatures. Besides, good results obtained at these con-
ditions could be useful to consider the hydrodeoxygenation of both
adducts (real mixture obtained after the condensation step) with-
out any purification. The differences and similarities between the
HDO of C8 and C13 condensated adducts are also discussed. Results
obtained are used to corroborate the mechanism and the kinetic
model, relating the results to the catalytic properties.
2.2. Catalysts characterisation
The surface, morphological and physical-chemical properties
of all the materials have been determined by N2 physisorption
(surface area, pore volume and diameter), XRD and TEM (metal dis-
persion), ICP-OES (metal loading) and NH -TPD (acid sites strength
3
and concentration). Details about each procedure and instruments
have been reported in our previous works [15,17]. In order to better
understand the catalytic activity of each catalyst, main characteri-
sation results are summarised in Section 3.
2.3. Reaction studies
Reactions were carried out in a 0.5 L stirred batch autoclave
reactor (Autoclave Engineers EZE Seal) equipped with a PID tem-
perature controller and a back pressure regulator. The reactor was
loaded with 0.25 Lof a acetone solution of 0.45 g of 1,5-bis(2-
furanyl)-1,4-pentadien-3-one (labelled as “A” to better identified
it) (Alfa Aesar, 98%). Considering the low aqueous solubility of this
compound, acetone was used as solvent. This solvent was cho-
sen taking into account economic reasons as well as the fact that
acetone is already presented in the condensation reaction. Prelim-
inary experiments using other solvents, including water and linear
alkenes, show very low solubility of the C13 condensation adducts.
2
. Experimental methodology
2.1. Catalytic preparation
Three different catalysts have been tested in this work: Pt/Al O ;
2
3
Pt/AC and Pt/MgZr/HSAG. 0.5% Pt/Al O3 catalyst was a commercial
2
sample supplied by BASF. High surface area graphites (HSAG-500)
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
3
O
A
O
B
O
C
O
O
O
O
O
O
(
1,4)-1,5-di(furan-2-yl)penta-1,4-dien-3-one
1,5-di(furan-2-yl)pentan-3-one
1,5-di(tetrahydrofuran-2-yl)pentan-3-one
D
E
OH
O
O
1,5-di(tetrahydrofuran-2-yl)pentan-3-ol
n-tridecane
Scheme 2. Chemical structures of the compounds identified in the hydrodeoxygenation (HDO) of C13-condensated adduct identified with the capital letter used in the
manuscript.
7
0 mg of the catalyst were added (with an average particle diam-
are calculated as the ratio between the concentration of each
compound and the concentration of C13 adduct converted.
eter of 50–80 ␮m) and air was purged out by adding nitrogen up
to 1.5 MPa for three times before starting reaction. The C13 loading
and the reactant/catalyst ratio were chosen considering the pre-
vious results obtained in the aldolization studies [7]. The reactor
Fig. 1 shows the evolution of the selectivity of different reac-
tion intermediates as function of the C13 conversion for the three
tested catalysts. The catalytic activity of all the supports (before
impregnation with Pt) has been tested, obtaining negligible con-
centration of any reaction product. These results corroborate that
a noble metal is needed at least for the first steps of the reaction.
However, in the case of carbonaceous materials, there is a decrease
in the carbon balance, indicating a relevant influence of reactant
adsorption. Concerning to the metal modified catalysts, same ten-
dencies were obtained with all the materials, despite that there
are important differences in the values. Compounds labelled as “B”
and “C” show a typical profile of primary reaction product, with a
significant selectivity at zero conversion and a decreasing profile
due to the subsequent steps. The decrease is more significant in
the case of the “B” intermediate, indicating that the following HDO
steps start mainly from this intermediate. By contrast, “D” interme-
diate follows the typical pattern of a secondary reactive product,
with negligible selectivity at the initial time and reaching a maxi-
mum after which its selectivity decreases, yielding the final product
(n-tridecane). Considering the fast conversion of the C13 conden-
sation adduct (mainly in the case of Pt/Al O and Pt/AC), the region
was pressurized to 2.5 MPa of H , stirred at 1000 rpm and heated
2
to 493 K, reaching a final pressure of 5.5 MPa. This hydrogen pres-
sure is large enough to consider that reaction is carried out in an
excess of hydrogen in comparison to the stoichiometrically needed.
According to the theoretic calculations, a complete hydrogenation
of all the C13 molecules would imply a consumption of 0.0252 mol
of H2 (0.24 MPa at 293 K).
Samples were withdrawn from the sampling port, filtered and
analysed by capillary GC-FID in a Shimadzu GC-2010, using a CP-
Sil 5 capillary column as a stationary phase. Peak assignment was
performed by GC-MS and responses were determined using stan-
dards and the effective carbon number theory [19]. Each sample
has been analysed at least three times with deviations below 4%
in all the cases. The structure of the different intermediates can be
observed in Scheme 2. Due to the difficult of the nomenclature of
each compound, they were labelled with a capital letter, as it is
summarized in this Scheme.
2
3
of higher conversion is magnified in the right part of each graph,
making easier the analysis of evolution at these values.
3
. Results and discussion
According to these results, C C double bonds are firstly hydro-
genated, but this reaction is so fast that there is not a clear
sequence between the hydrogenation of aliphatic and cyclic unsat-
urations. Consequently, the intermediate in which the furanic
rings are reduced is also observed since the first moments. This
behaviour was already observed in the hydrogenation of other
compounds containing multiple conjugated carbon-carbon dou-
ble bonds [20]. It should be highlighted that the identification of
reaction compounds was carried out using GC-MS without observ-
ing intermediates corresponding to the hydrogenation of only one
of the aliphatic or cyclic C C. Considering the symmetry of this
molecule, there is not any apparent reason that justifies a prefer-
ential attack to one of them. Contrary to the results observed in the
case of the HDO of the C8 condensation adduct, the hydrogenation
of the ketonic group is clearly observed. Next steps between this
intermediate and the final n-tridecane are so fast that the result-
ing intermediates are not detected. Thus, these steps are omitted,
considering directly the transformation of this intermediate (D)
into n-tridecane.
Considering the complex structure of the reactants, a pre-
liminary study to choose the optimum solvent was needed. The
solubility on water is not enough to guarantee high precision in
the analyses. Same drawback is presented when non-polar organic
solvents are used (hexane, diethyl-ether). Higher solubility was
obtained using aprotic polar organic solvents, such as ethyl acetate,
tetrahydrofuran (THF) and acetone, increasing its solubility at
increasing dielectric constant values. Best behaviour was observed
with acetone. Before been chosen as solvent, a blank test was car-
ried out, discarding any acetone hydrogenation product after 24 h.
These analytic results are also supported by other technical and eco-
nomic reasons: acetone is one of the reactants in the previous step
of the biomass to biofuels process (aldol condensation), so there
is not a new compound. Considering all these points, acetone is
chosen as solvent for this process.
Batch experiments at 493 K and 5.5 MPa were performed with
the selected catalysts in order to identify the most active material
and to analyse if reactivity trends are similar in the HDO of C8
and C13 condensation adducts. Reaction conditions are chosen
considering previous results obtained in the HDO of the C8 adduct
Considering the similar tendencies observed with all the mate-
rials, main differences are better analysed as function of the
distributions of all the compounds at 50% and 90% of C13 conver-
sion, considering also the time needed to reach these conversions
[
17]. The most active catalyst was selected considering both, the
selectivity for n-tridecane and the carbon balance. Selectivities
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
4
Fig. 1. Evolution of the selectivity with C13 reactant conversion in the HDO of C13-condensation adduct catalyzed by (a) Pt/Al2O3; (b) Pt/AC; (c) Pt/MgZr/HSAG. Symbols:
) B; ( ) C; ( ) D; ( ) n-tridecane (E).
(
as well as the final distribution obtained after 24 h of reaction.
These selectivities are summarised in Fig. 2. Results at 50% of
conversion indicate a higher activity of Pt/Al O . In addition to dif-
the carbon balance, value obtained with Pt/Al O3 (97.6%) discards
2
any adsorption or side-reaction, whereas the values of the active
carbon and the graphite (55 and 67%, respectively) justify the low
selectivities of the different intermediates. As a consequence, the
selectivity to n-tridecane only reaches values higher than 10% in
the case of Pt/Al O .
2
3
ferences in the selectivities, the differences in the times needed to
reach this conversion must be highlighted. In the case of Pt/Al O3
2
and Pt/AC, this conversion is reached after only 2 h reaction time,
whereas more than 6 h reaction time are needed in the case of
Pt/MgZr/HSAG500. Results of carbonaceous materials are condi-
tioned by the low carbon balances, lower than 50% in the case
of Pt/AC and 72% in the case of Pt/MgZr/HSAG500. As opposite,
2
3
Considering the long time needed to reach the 90% conversion
when Pt/MgZr/HSAG500 is used, the distribution of the final selec-
tivities after 24 h is not significantly different. Despite of this fact,
it must be highlighted the soft increase in the selectivity of the last
intermediate and the n-tridecane, reaching a final value of 11.2%.
These results discard a negative effect of basicity on the whole pro-
cess and suggest that increasing reaction time would have a positive
effect in the n-tridecane yield. Concerning to the Pt/Al O , 21.5%
the carbon balance of Pt/Al O3 is higher than 98%. This mate-
2
rial shows the highest activity, highest yields for “C” and “D”
intermediates and almost 10% of n-tridecane selectivity after only
2
h of reaction. On the contrary, only significant amounts of the
2
3
first intermediate are observed when active carbon is used and
Pt/MgZr/HSAG500 shows an intermediate behaviour, with 3% of
n-tridecane.
selectivity for n-tridecane was reached after 24 h, being this alu-
mina supported catalyst the most active one, with full conversion
and more than 95% of final carbon balance. This results is in good
agreement with the good activity observed in the hydrodeoxygena-
tion of the C8 adduct [15], suggesting that the higher complexity
of the reactant has not a significant effect in the catalytic param-
eters that play a key role in this process. As in the other cases,
results obtained with Pt/AC are strongly conditioned by the low
carbon balance (44.5%). Most significant evolution of the reaction
At 90% of conversion, products distribution obtained with the
three catalysts are very similar. In all the cases, the highest selec-
tivity corresponds to the first intermediate, indicating that the
hydrogenation of aliphatic unsaturations is the first and fastest step.
More than 90% of conversion is reached after 8 h with Pt/Al O , after
2
3
6
h with Pt/AC and after 20 h with the basic materials. Concerning to
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
5
Fig. 3. Evolution of the different furfural hydrogenation products with reaction time
when using Pt/MgZr/HSAG as catalyst: ( ) furfural; ( ) furfuryl alcohol; (
methylfurane and ( ) tetrahydromethylfurane.
)
Fig. 2. Distribution of the different reaction intermediates in the HDO of C13-
condensated adduct catalyzed by Pt/Al2O3 (white); Pt/AC (gray) or Pt/MgZr/HSAG
(
black). Results obtained at (a) 50%; (b) 90% of conversion and after (c) 24 h of
reaction.
is related to the complete conversion of the reactant, despite that
almost negligible evolution of the final product is observed.
Taking into account these data, a deep study to justify the low
values of carbon balances obtained with HSAG and AC supports
is needed for clarifying the reasons of this behaviour. The retro-
condensation was the only side-reaction observed, and its products
were only detected when Pt/MgZr/HSAG500 was used as cata-
lyst, but they were not detected neither for Pt/Al O nor Pt/AC.
Fig. 4. (a) Evolution of the carbon balance as function of the reaction time when
Pt/MgZr/HSAG is used as catalyst: ( ) considering only the main reaction; and (
2
3
)
considering also the C5’s products obtained as side reaction. (b) Evolution of the
The existence of this reaction has been previously demonstrated
21], yielding acetone and furfural. No signals corresponding to
carbon balance as function of the reaction time when Pt/AC is used as catalyst.
[
the first condensation adduct (C8) or its hydrogenated derivatives
were detected. The strong reaction conditions are suggested as
the responsible of this complete retroaldolization. Considering that
acetone is the solvent, only furfural (or its hydrogenated deriva-
tives) would be identified. The basic sites of MgZr mixed oxide [7]
justify the presence of this side reaction, questioning the viability
of this material as good catalyst for the C13 HDO.
its fast hydrogenation to furfuryl alcohol. Contrary to the behaviour
of C8 and C13 condensation adducts, the aromatic cycle is more sta-
ble than the carbonyl group (aldehyde for furfural; ketone for the
adducts), so this functionality suffers the hydrogenation firstly. In
fact, the methyltetrahydrofurane (MTHF) only appears in the last
part of the reaction, with almost negligible selectivity.
In good agreement to this hypothesis, Fig. 3 shows the evolution
of the C5 compounds identified by GC-MS whereas Fig. 4a illus-
trated their influence on the carbon balance values. As it can be
observed in Fig. 3, there are significant selectivities of furfural and
the partial hydrogenated adducts (similar adducts as those iden-
tified in furfural hydrogenation [22]), reaching a maximum global
selectivity of 16%. Furfural appears since the first reaction times
but its selectivity never reaches values higher than 3% because of
Once this side reaction is taken into account, the evolution of the
carbon balance obtained with this material improves, being over
70% during all the reaction (Fig. 4a). However, this value is still much
lower than the value obtained with Pt/Al O3 (constant between
2
100 and 96%), suggesting the presence of additional causes. At this
point, two different zones of carbon balance decrease are identi-
fied for the Pt/MgZr/HSAG, catalyst: the first one in the first two
hours of reaction time, and the other one between the eighth and
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
6
Table 1
Summary of the main morphological al chemical properties of the catalysts used in this work [15].
N2 Physisorption
Metal loading,
ICP-OES (%)
TEM
NH3-TPD
2
3
SBET (m /g)
dp (nm)
v
p (cm /g)
Metal
dispersion
Metal
particle
size (nm)
Weak
acidity
(␮mol/g)
Medium
acidity
(␮mol/g)
Total
acidity
(␮mol/g)
(
%)
Pt/Al2O3
Pt/AC
Pt/MgZr/HSAG
116
1716
234
17
1.9
10.1
0.49
0.59
0.5
0.5
0.59
0.58
24
21.3
13.1
4
4.7
8.4
789
18
243
119
71
230
908
89
473
Scheme 3. Reaction mechanism proposed to the hydrodeoxygenation of the C13-condensated adduct catalyzed by different Pt materials.
the tenth hour. This pattern suggests that the first one is caused by
the reactant itself, whereas the other one is caused by a product.
Considering the chemical structures of reactant and some inter-
mediates, a strong interaction between these compounds and the
catalytic support (adsorption) can be assumed as the reason of the
first carbon balance decrease. In order to corroborate this hypothe-
sis, an adsorption test was carried out, with same concentration of
reactant and catalyst as in the reaction, and analysing the evolution
of C13-condensated adduct in the liquid phase when this solution
is heated to 493 K (reaction temperature) but in an inert atmo-
sphere (in order to prevent the reaction). The C13-condensated
adsorption capacity of Pt/MgZr/HSAG was calculated analysing the
liquid phase with GC-FID, obtaining a value of 1.23 mmol/g. This
value is high enough to justify the initial decrease in the carbon
of the carbon balance in the first hours of reaction. However, the
continuous decreasing tendency suggests that adsorption on active
carbon affects not only to the reactant but also to all the reaction
intermediates. The high surface area of this material (main charac-
terization results are summarized in Table 1 and deeply discussed
in our previous work [15]) and the different functionalities justi-
fies the high relevance of this complex process when this catalyst
is used.
3.1. Kinetic results
Results obtained are better explained considering a kinetic anal-
ysis. These profiles are congruent with the reaction mechanism
previously proposed for the HDO of C8-condensation adducts [17].
According to this mechanism, this process is the result of different
series-parallel steps with a first-order dependence on the reactant
and partially hydrogenated condensation adducts and apparent
zeroth order on H2 concentration. This hypothesis is based on
the large excess of this reactant (observing a maximum decrease
of 0.4 MPa after 24 h of reaction). Considering the experimental
results, the previous mechanism can be directly adapted in order
to analyse the behaviour of Pt/Al O . However, in the case of car-
balance by this adsorption (the value obtained with Pt/Al O3 was
2
0.04 mmol/g). This adsorption can be explained considering the
high number of unsaturations of this molecule and the high affin-
ity between these functionalities and carbonaceous supports [23].
The second decrease in the carbon balance corresponds with the
time at which the concentration of the first intermediate reaches
a maximum with this catalyst. Considering the similar structure, it
is assumed that this second decrease corresponds to the adsorp-
tion of this molecule. TPO analyses were carried out but results are
not conclusive because of the similar oxidation temperature of all
these compounds. These adsorptions cause a partial blockage of
the active sites, making slower the reaction but without causing a
complete deactivation.
In the case of Pt/AC, the presence of side reactions is dis-
carded because no C5 products were detected by the GC-FID.
Observing the evolution of the carbon balance (Fig. 4b), the strong
adsorption of the reactant seems to be not enough to justify
these results and the negligible activity of this catalyst. Same
adsorption analyses were carried out with this material, obtain-
ing an adsorption capacity of 2.36 mmol/g, much higher than the
obtained with the other catalysts. This fact justifies the decrease
2
3
bonaceous materials, two adsorption steps (corresponding to the
reactant and the first intermediate) are also considered, as well as
a side reaction of retroaldolization in the case of Pt/MgZr/HSAG.
Considering that the hydrogenation of furfural is not the aim of
this work, in order to simplify the mechanisms, all the C5 com-
pounds are considered as a single lumped compound. With these
assumptions, the HDO of C13-adduct follows the scheme showed
in Scheme 3. As in our previous works, the presence of mass trans-
fer effects was discarded by ensuring small catalytic particle size
(sieved from 50 to 80 ␮m) and by carrying out the reactions under
high stirring speed (1000 rpm). Consequently, the kinetics of this
process is controlled by the rate of the chemical steps. The tem-
poral evolution of all the species with reaction time was modelled
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
7
Table 2
−1
−1
Kinetic constants k1–k6 (l s molexpPt) and adsorption constants kA1–kA2 (m s ) for the fitting experimental data at 493 K to the proposed kinetic model.
Kinetic and adsorption constants
Catalyst
k1
k2
k3
k4
k-4
k5
k6
kA1
kA2
r2
0.991
Pt/Al2O3
Pt/AC
Pt/MgZr/HSAG
17.5
16.3
15.4
9.79
6.32
7.39
13.4
5.63
9.42
3.17
2.55
3.10
2.99
53.6
18.6
11.1
25.1
21.3
–
–
2.44
0
0
−
−
9
8
−9
−9
9.0 × 10
1.1 × 10
0.98
0.96
◦
2.9 × 10
5.8 × 10
using the SCIENTIST® software, applying the Burlich-Stoer approx-
imation and the following differential equations:
dA
=
=
=
=
=
=
−k · A − k · A − k7 · A − k · A
(1)
(2)
(3)
(4)
(5)
(6)
1
2
AL
dt
dB
dt
−k · A − k · B − k · B + k · C − kA2 · C
1
3
4
−4
dC
dt
k · A − k · B − k · C
2
4
−4
dD
dt
k · B − k · D
4
S
dE
dt
kS · D
dF
dt
2 · k · A
6
Values of the kinetic constants obtained are reported in Table 2.
In order to better comparison of the results, values are normalized
considering that kinetic constants are function of metal loading and
metal dispersion whereas adsorptions are surface processes:
−
1
3
kD(s ) · V(m )
ꢀ
k
=
(7)
(8)
A
2
−1
)
Mcat(g) · SBET (m · s
−1
k(s ) · V(L)
kꢀ =
Mcat(g) · (%metal/100) · (1/APt) · d
Where “V” is the reaction volume; “Mcatꢀꢀ is the mass of catalyst
used in each reaction; “SBETꢀꢀ is the specific surface area measured
by nitrogen physisorption; “APtꢀꢀ is the atomic mass of Pt; and “d”
is the metal dispersion calculated by TEM. All these morphologi-
cal parameters have been measured in our previous work [15] and
main results are summarized in Table 1.
As it can be observed in Table 2, good fits were obtained with
the three materials, with a correlation coefficient higher than 0.95
in all the cases. In good agreement, there is a good fit between
experimental and theoretical values, as it is illustrated in Fig. 5. In
general, kinetic constants values are lower than those previously
proposed for the C8-adduct HDO, but they follow the same trend,
mainly in the first steps of the process, with higher values when
Pt/Al O is used as catalyst [15]. These values are congruent with
2
3
the higher complexity of both, the reactant molecule and the pro-
cess. As to the adsorption constants, values obtained in the C13 HDO
are two orders of magnitude higher than those obtained during the
C8 HDO, in good agreement with the influence of the molecule size
in adsorption processes. Values of kinetic constants have been ana-
lysed in terms of the main characterization parameters of the three
catalysts, observing that each step of the process is conditioned by
different parameters, in such a way that a good balance among dif-
ferent catalytic properties is needed to maximize the n-tridecane
yield. Most significant trends are shown in Fig. 6. According to
these graphs, the hydrogenation of the aliphatic C C (controlled by
Fig. 5. Comparison between evolutions of experimental concentrations (symbols)
and results predicted by the kinetic model proposed (broken lines). Results corre-
sponding to (a) Pt/Al2O3; (b) Pt/MgZr/HSAG; and (c) Pt/AC. Symbols: (×) A; ( ) B;
(
) C; ( ) D; ( ) E.
this acid sites on the kinetic constant is observed in almost all the
direct steps. Acidity is needed to favour the complete hydrogena-
tion of all C C double bonds (conditioned by the value of k ) and
2
the hydrogenation of ketonic group (related to the k3 constant).
Concerning to the equilibrium between the first and the second
intermediate, the global effect of weak acidity is also favourable.
The acidity has a negative effect in the evolution of this constant,
so it can be concluded that higher weak acidity prevents the forma-
tion of intermediate “C” (this product is not directly related to the
n-tridecane formation). The weak acidity is not good to promote
the cyclic breakups (k5). However, analysing the final values, it can
be concluded that this step does not control the whole process and
the k1 and k ) is slightly structure-sensitive, increasing its value at
2
decreasing metal particle size (Fig. 6a). This behaviour has been pre-
viously observed in the hydrogenation of other complex molecules
[
24]. The role of acidity is not clear, observing positive and nega-
tive effects as function of the step analysed. If only the weak acidity
is considered (Fig. 6b), a positive influence of the concentration of
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055

G Model
CATTOD-9869; No. of Pages8
ARTICLE IN PRESS
L. Faba et al. / Catalysis Today xxx (2015) xxx–xxx
Acknowledgements
8
20
15
10
5
0
(
a)
This work was supported by the Spanish Ministry of Econ-
omy and Competitiveness (contracts CTQ2011-29272-C04-02 and
CTQ2014-52956-C3-1-R).
References
[
[
1] E. Furimsky, Hydroprocessing challenges in biofuels production, Catal. Today
17 (2013) 13.
2] G.W. Huber, J.N. Chheda, G.J. Barret, J.A. Dumesic, Production fo liquid alkanes
by aqueous-phase processing of biomass-derived carbohydrates, Science 308
2
0
200
400
600
800
1000
14
(
2005) 1446.
SBET(m2)
[3] J.N. Chheda, J.A. Dumesic, An overview of dehydration, aldol-condensation
and hydrogenation processes for production of liquid alkanes from
biomass-derived carbohydrates, Catal. Today 123 (2007) 59.
[4] A.C. Doiseau, F. Rataboul, L. Burel, N. Essayem, Synergy effect between solid
acid catalysts and concentrated carboxylic acids solutions for efficient furfural
production from xylose, Catal. Today 226 (2014) 176.
5] E.I. Gurbuz, J.M.R. Gallo, D.M. Alonso, S.G. Wettstein, W.Y. Lim, J.A. Dumesic,
Dumesic,Conversion of hemicellulose into furfural using solid acid catalysts in
gamma-valerolactone, Angew. Chem. Int. Edit. 52 (2013) 1270.
60
40
20
0
(
b)
12
10
8
[
[
6] B.T. Kusema, L. Faba, N. Kumar, P. Maki-Arvela, E. Díaz, S. Ordó n˜ ez, T. Salmi,
D.Y. Murzin, Hydrolytic hydrogenation of hemicellulose over metal modified
mesoporous catalyst, Catal. Today 196 (2012) 26.
6
4
[7] L. Faba, E. Díaz, S. Ordó n˜ ez, Aqueous-phase furfural-acetone aldol
condensation over basic mixed oxides, Appl. Catal. B 113–114 (2012)
201.
2
[
8] L. Faba, E. Díaz, S. Ordó n˜ ez, Improvement on the catalytic performance of
MgZr mixed oxides for furfural acetone aldol condensation by supporting on
mesoporous carbons, ChemSusChem. 6 (2013) 463.
0
200
400
600
800
1000
Weak Acidity (μmol/g)
[
9] K.A. Sorate, P.V. Bhale, Biodiesel properties and automotive system
compatibility issues, Renew. Sust. Energ. Rev. 41 (2015) 777.
Fig. 6. Evolution of different kinetic constants as function of different catalytic prop-
erties: (a) metal particle size; and (b) weak acidity. Symbols: ( ) k1; ( ) k2; (
k3; ( ) k4; ( ) k5; ( ) k-4.
)
[
[
10] M.F. Demirbas, Biorefineries for biofuel upgrading: a critical review, Appl.
Energ. 86 (2009) S151.
11] N. Aurun, R.V. Sharma, A.K. Dalai, Green diesel synthesis by
hydrodeoxygenation of bio-based feedstocks: strategies for catalyst design
and development, Renew. Sust. Energ. Rev. 48 (2015) 240.
[12] S. De, B. Saha, R. Luque, Hydrodeoxygenation processes: advances on catalytic
transformations of biomass-derived platform chemicals into hydrocarbon
fuels, Bioresource Technol. 178 (2015) 108.
13] Z. He, X. Wang, Hydrodeoxygenation of model compounds and catalytic
systems for pyrolysis bio-oils upgrading, Catal. Sust. Energ. Prod. 1 (2012)
this step is fast enough if the previous intermediate is obtained.
In good agreement to this hypothesis, the highest n-tridecane
selectivity is obtained with Pt/Al O , despite of its acidity. If same
2
3
kinetic constants are analysed as function of the medium strength
acidity, this tendencies are not so clear, suggesting that an increase
in the strength of these sites can have an opposite effect.
Considering that same parameters condition the hydrodeoxy-
genation of both condensation adducts (C8 and C13), and that best
[
[
[
28.
14] V.N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Hydrodeoxygenation of guaicol
with CoMo catalysts. Part I: Promoting effect of cobalt on HDO selectivity and
activity, Appl. Catal. B 101 (2011) 239.
15] L. Faba, E. Díaz, S. Ordó n˜ ez, Role of the support on the performance and
activity results were obtained with Pt/Al O in both cases; this
2
3
stability of Pt-based catalysts for furfural-acetone adduct
hydrodeoxygenation, Catal. Sci. Technol. 5 (2015) 1473.
material is proposed as the most promising one for the hydrodeoxy-
genation of real mixtures of condensated adducts in order to obtain
diesel from (hemi)cellulose wastes by aqueous-phase reactions.
[16] Y.K. Lugo-José, J.R. Monnier, C.T. Williams, Gas-phase, catalytic
hydrodeoxygenation of propanoic acid, over supported group VIII noble
metals: metal and support effects, Appl. Catal. A 469 (2014) 410.
[
17] L. Faba, E. Díaz, S. Ordó n˜ ez, Hydrodeoxygenation of acetone-furfural
condensation adducts over alumina-supported noble metal catalysts, Appl.
Catal. B 160–161 (2014) 436.
4
. Conclusions
[
18] L. Faba, E. Díaz, S. Ordó n˜ ez, One-pot aldol condensation and
hydrodeoxygenation of biomass-derived carbonyl compounds for biodiesel
synthesis, ChemSusChem 7 (2014) 2816.
The HDO of the C13-condensation adducts was studied,
analysing the catalytic activity of different materials that have
been successfully tested for HDO of similar compounds. All the
experimental results were used to propose a reaction mechanism
and a kinetic model, identifying the role of catalytic parameters in
each step of the process. This analysis corroborates the complex-
ity of this reaction, with different contributions in each step, but
highlighting the role of the metal dispersion. In good agreement,
best results were obtained with Pt/Al O , yielding more than 22%
[19] E. Furimsky, Chemistry of catalytic hydrodeoxygenation, Catal. Rev. 25 (1983)
21.
4
[
20] D. Loffreda, C. Michel, F. Delbecq, P. Sautet, Tuning catalytic reactivity on
metal surfaces: insights from DFT, J. Catal. 308 (2013) 374.
[21] L. Faba, E. Díaz, S. Ordó n˜ ez, Improvement of the stability of basic mixed oxides
used as catalysts for aldol condensation of bio-derived compounds by
palladium addition, Biomass Bioenerg. 56 (2013) 592.
22] M.M. Villaverde, N.M. Bertero, T.F. Garetto, A.J. Marchi, Selective liquid-phase
hydrogenation of furfural to furfuryl alcohol over Cu-based catalysts, Catal.
Today 213 (2013) 87.
23] J.I. DiCosimo, A. Acosta, C.R. Apesteguía, Allylic alcohol synthesis by gas-phase
hydrogen transfer reduction of unsaturated ketones, J. Mol. Catal. A 234
(2005) 111.
24] J.A. Bennett, R.P. Fishwick, R. Spence, J. Wood, J.M. Winterbottom, S.D.
Jackson, E.H. Stitt, Hydrogenation of 2-pentyne over Pd/Al2O3 catalysts:
effect of operating variables and solvent selection, Appl. Catal. A 364 (2009)
57.
[
[
[
2
3
of n-tridecane after 24 h of reaction at 493 K and 5.5 MPa. The pres-
ence of basic sites on the catalyst leads to undesired C C cleavage
yielding furfural, which can be subsequently hydrogenated. On the
other hand, the presence of strong adsorption and side reactions
limit the activity of catalysts supported on active carbon and high
surface area graphites, discarding these materials as catalysts for
this reaction.
Please cite this article in press as: L. Faba, et al., Hydrodeoxygenation of furfural-acetone condensation adducts to tridecane over
platinum catalysts, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.09.055