Catalytic coupling of carboxylic acids by ketonization as a processing step in biomass conversion

Article abstract of DOI:10.1016/j.jcat.2009.05.015

Carboxylic acids, common intermediate products in biomass conversion processes, can be converted into ketones via ketonization reactions over a ceria-zirconia catalyst. Reaction kinetics studies were carried out using hexanoic acid, as a representative ca

Full text of DOI:10.1016/j.jcat.2009.05.015

Journal of Catalysis 266 (2009) 71–78
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Catalytic coupling of carboxylic acids by ketonization as a processing step
in biomass conversion
Christian A. Gaertner, Juan Carlos Serrano-Ruiz, Drew J. Braden, James A. Dumesic *
University of Wisconsin-Madison, Department of Chemical and Biological Engineering, Madison, WI 53706, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Carboxylic acids, common intermediate products in biomass conversion processes, can be converted into
ketones via ketonization reactions over a ceria–zirconia catalyst. Reaction kinetics studies were carried
out using hexanoic acid, as a representative carboxylic acid, in the presence of 1-pentanol and 2-buta-
none, as representative biomass-derived alcohol and ketone species. Studies were carried out at temper-
atures from 448 to 623 K, and employing a range of hexanoic acid partial pressures from 0.05 to 0.3 atm.
Two different reactions were observed to take place at these reaction conditions: esterification and keto-
nization, both consuming hexanoic acid. Product inhibition by water and carbon dioxide was observed
and studied by co-feeding these components to the reactor. Hexanoic acid adsorption on the catalyst sur-
face is an important step in the reaction, and the rate of ketonization shifts from second order to zero
order as the partial pressure of acid increases. The measured activation energy for the ketonization of
hexanoic acid (132 kJ/mol) is higher than the esterification reaction (40 kJ/mol), such that the irreversible
ketonization reaction is favored at higher temperatures (>573 K) compared to the reversible esterification
reaction. Direct ketonization of esters does not take place in the presence of acids, and instead takes place
by hydrolysis with water followed by ketonization of the corresponding acid. The results of this study can
be described by a simple kinetic model including site blocking by adsorbed hexanoic acid, carbon dioxide,
and water.
Received 25 February 2009
Revised 20 April 2009
Accepted 15 May 2009
Available online 10 June 2009
Keywords:
Biomass conversion
Ketonization
Esterification
Hexanoic acid
Ceria–zirconia
Kinetic modeling
Ó 2009 Elsevier Inc. All rights reserved.
1
. Introduction
4 6
carboxylic acids in the C to the C range that spontaneously
separate from water. In a subsequent step, these functional inter-
mediates are upgraded by C–C coupling reactions into larger com-
pounds. In this respect, the carboxylic acids and esters can be
converted into heavier ketones via ketonization reactions, involv-
ing the condensation of two molecules of acid (or ester) to pro-
The utilization of renewable resources for the substitution of
fossil fuels is a key issue in moving our economy toward a more
sustainable future [1]. In this respect, biomass is currently the
only sustainable source for organic carbon [2]; however, the pro-
duction of renewable fuels from biomass necessitates the devel-
opment of new efficient processes to convert the biomass-
derived feedstocks into liquid fuels [3,4]. The high functionality
and corresponding reactivity of renewable carbohydrate feed-
stocks (i.e., high oxygen to carbon stoichiometry) compared to
petroleum feedstocks requires alternative catalytic processing
techniques to convert these resources into fuels for use in the
transportation sector (e.g., gasoline, diesel, and jet fuels). We have
recently reported a strategy to transform two important biomass-
derived carbohydrates (glucose and sorbitol) into hydrocarbons of
different classes to be used as transportation fuels [5]. This ap-
proach involves a two-step process in which the oxygen content
of the feedstock is first reduced over a Pt–Re/C catalyst to gener-
ate monofunctional intermediates such as alcohols, ketones, and
duce a linear ketone with 2n ꢀ 1 carbon atoms, CO
2
, and water
[6–9]. Use of this reaction as a means of reacting organic acid is
of particular importance because biomass-derived organic efflu-
ents can possess a high content of carboxylic acids (e.g., as high
as 30 wt% when glucose is used as reactant [5]).
In the present study, kinetic factors have been investigated for
ketonization upgrading processes over a Ce0.5Zr0.5O catalyst. This
2
material showed desirable catalytic properties for ketonization of
carbohydrate-derived carboxylic acids in the presence of other
monofunctional oxygenated species [5] and thus was selected for
further study. As a model for monofunctional organic effluents ob-
tained in our previous work [5], we have studied the ketonization
of hexanoic acid in mixtures with 1-pentanol and 2-butanone. Un-
der the condition of this study, two primary reactions can proceed
in parallel: (i) the reversible esterification of hexanoic acid and 1-
pentanol to generate 1-pentylhexanoate (reaction (1)), and (ii) the
ketonization of hexanoic acid to produce 6-undecanone, carbon
dioxide, and water (reaction (2)).
*
Corresponding author.
E-mail address: dumesic@engr.wisc.edu (J.A. Dumesic).
0
021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.05.015

7
2
C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
Hexanoic acid þ 1-Pentanol $ 1-Pentylhexanoate þ Water
Hexanoic acid þ Hexanoic acid ! 6-Undecanone
þ Carbon Dioxide þ Water
ð1Þ
ð2Þ
a close-fitting aluminum block heated externally by a well-insu-
lated furnace (Applied Test Systems Inc.). Temperature was mea-
sured using a K-type thermocouple (Omega) attached to the
outside of the reactor and controlled with a 1600 series type tem-
perature controller (Love controls Series 16A). A mass-flow con-
troller (5850 Brooks Instruments) was used to control the helium
or carbon dioxide flowrates. A HPLC pump (Alltech 301 HPLC
pump) was used to introduce the liquid feed solution into the reac-
tor. The effluent liquid was collected at room temperature in a gas–
liquid separator and drained for gas chromatography (GC) analysis
(Shimadzu GC-2010 with a FID detector and Rtx-5 column) and
identification (Shimadzu GC-2010 with a mass spectrometer and
Rtx-5 column). The effluent gas stream passed through a back-
pressure regulator (GO Regulator, Model BP-60) which controlled
the system pressure. Gas phase samples were analyzed with two
different GCs (Varian Star 3400 CX with a FID detector and a GS-
Q capillary column for organic components and Shimadzu GC8A
with a TCD detector and a HAYESFP DB100/120 column for the car-
bon dioxide measurements). All the feeds were prepared from high
purity chemicals: hexanoic acid (99%, Aldrich), 2-butanone (99%,
Aldrich), propyl hexanoate (Aldrich, P98%), hexane (95%, Aldrich),
and 1-Pentanol (P99%, Aldrich).
We have studied the effects of solvents, partial pressures of the
reactants and products, and reaction temperature on the rates of
ketonization and esterification reactions. Additionally, we have
investigated the inhibiting effects of carbon dioxide and water (pro-
duced by ketonization) on the kinetics of esterification and ketoni-
zation reactions. We have also studied the possibility for direct
2
ketonization of esters to form the ketone adduct and CO , plus the
stoichiometric alcohol and olefin. Finally, a simple kinetic model
is presented that incorporates these effects to give a satisfactory
description of the experimental results.
2
. Experimental
A Ce0.5Zr0.5
O
2
catalyst was prepared by a co-precipitation meth-
od described in detail elsewhere [10] using aqueous solutions of
Ce(NO O (Aldrich, 99%) and ZrO(NO (Aldrich, 99%). Pure
ꢁ 6H
ZrO oxide was prepared using the same method. Temperature
programmed desorption of NH and CO was used to probe the
3
)
3
2
3 2
)
2
3
2
acid and base sites, respectively, on these two catalysts, using an
apparatus described elsewhere [11].
3. Results
The reactor used for the reaction kinetic studies (Fig. 1) was a
fixed-bed, upflow reactor consisting of a 6.35 mm outer-diameter
tubular stainless steel tube (wall thickness 0.71 mm) containing
the catalyst bed between two plugs of quartz wool (Alltech) and
Experiments were carried out to determine the suitability of
using 2-butanone as solvent for the kinetic studies of the ketoniza-
tion of hexanoic acid. In this respect, 2-butanone was selected as
the preferred solvent used to simulate the monofuctional organic
effluent obtained in our original work [5] and these results were
compared with those obtained using hexane as a solvent. Fig. 2
fused SiO
2
granules (Sigma–Aldrich). The reactor was heated with
shows the ketonization rates (in terms of 6-undecanone and CO
2
GC
formation) for a mixture of 7.5 mol% of hexanoic acid in both sol-
vents at three different temperatures at a total pressure of 1 atm.
It can be seen that the ketonization rates have similar values for
both solvents at each of the three temperatures. Additionally, no
Gas effluent
Back pressure
regulator
p
2
reaction products aside from undecanone, CO , and water from
the ketonization process were detected when 2-butanone was
used as a solvent, and this ketone can thus be considered to be
essentially inert over the ceria–zirconia catalyst at the conditions
of the present study.
Separator
Sweep gas
Liquid drain
4.0
3.5
3.0
2.5
2.0
1.5
1.0
Thermocouple
Heating
Quartz wool
Catalyst bed
p
Gas feed
-3
-3
-3
-3
-3
1
.6x10
1.7x10
1.7x10
1.8x10
1.8x10
-
1
-1
Liquid feed
T
(K )
HPLC pump
Fig. 2. Hexanoic acid ketonization reaction rates at various temperatures for a
7.5 mol% hexanoic acid feed using (.) 2-butanone and (h) hexane as solvent.
Fig. 1. Reactor setup used for reaction kinetics experimental studies.

C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
73
1
00
ate, was fed to the reactor and reacted over ceria–zirconia. This
experiment was performed by co-feeding helium to the reactor in-
stead of using 2-butanone, since this organic solvent may contain
traces of water which could lead to hydrolysis of the ester to an
acid, followed by ketonization of the acid. As a comparison, we also
studied the ketonization of pure hexanoic acid at the same condi-
9
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
0
3
ꢀ1
ꢀ1
tions (623 K, 6.8 cm (STP) min helium flow, 0.18 ml min hexa-
noic acid or ester flow) leading to an acid/ester partial pressure of
0
.026 atm at a total pressure of 1 atm. The ketonization rate for
ꢀ1
ꢀ1
hexanoic acid (71 lmol min g cat ), measured as the rate of
undecanone formation, was then compared to the ketonization of
ꢀ1
ꢀ1
propyl hexanoate (31
lmol min g cat ). Interestingly, in both
cases CO was detected in the gas effluent while, when propyl
2
hexanoate was used as a feed, 1-propanol in the liquid phase and
propylene in the gas effluent were also found.
2
Since CO and water are both products from the ketonization
0
.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
reaction, experiments were carried out to investigate the potential
inhibiting effects of these compounds on the rates of esterification
and ketonization. Thus, in the case of carbon dioxide, experiments
similar to the those described above were performed with a gas co-
feed through the reactor bed at a temperature of 598 K at a total
pressure of 1 atm with a liquid feed composed of 7.5 mol% hexa-
p Hex acid (atm)
Fig. 3. Ketonization reaction rates for varying partial pressures of hexanoic acid
using 2-butanone as solvent at (j) 547 K, (.) 572 K, (ꢀ) 597 K, and (d) 623 K.
ꢀ
1
noic acid in 2-butanone at a flow of 0.1 ml min and a gas co-feed
Fig. 3 shows the ketonization rate at various temperatures
547–623 K) as a function of the partial pressure of hexanoic acid
in the reactor at a total pressure of 1 atm. It can be seen that the
ketonization rate reaches an asymptotic value with an increase
in the hexanoic acid partial pressure in the reactor. The partial
pressure where this limit is approached begins at approximately
3
ꢀ1
with varied composition at 42 cm (STP) min . The total molar
flow of the gas feed through the reactor was kept constant, and
(
the molar ratio between helium and CO
in Table 2, the ketonization rate decreased to approximately 29% of
the initial value with a co-feed of CO (runs K2-C and K3-C in Table
). In order to check the reversibility of the CO inhibition effect, a
2
was varied. As can be seen
2
2
2
0
.1 atm. In order to study the effect of the presence of carboxylic
second run (K4-C) was carried out at the same conditions of the
initial experiment (K1-C). It can be seen that the ketonization rate
acids and alcohols on the kinetics of simultaneously occurring
esterification and ketonization reactions, experiments were per-
formed at varying temperatures (448–623 K) at a total pressure
of 1 atm and partial pressures of hexanoic acid and 1-pentanol,
using different mixtures of these compounds in 2-butanone. The
weight hourly space velocity (WHSV) for each experiment was var-
ied to maintain the conversion below 60%. For all feeds listed in Ta-
ble 1, a temperature study was performed, and the individual
esterification and ketonization reaction rates were determined. Re-
sults are presented in Figs. 6–9, where the results of the kinetic
model, discussed below, are also shown.
ꢀ1
ꢀ1
increased (4.8 versus 3.5
lmol min g cat ) compared to the CO
2
co-feed experiments (K2-C and K3-C), but the initial ketonization
ꢀ
1
ꢀ1
catalytic activity (12.6
To investigate the CO
reaction, experiments were carried out by co-feeding CO
perature of 548 K at a total pressure of 1 atm with a feed of
0 mol% hexanoic acid and 15 mol% 1-pentanol with 2-butanone
as solvent (E1-C, E2-C, and E3-C shown in Table 2) at a liquid flow
l
mol min g cat ) was not regained.
-inhibiting effect for the esterification
at a tem-
2
2
1
ꢀ1
rate of 0.15 ml min with a gas feed of varied composition at
3
ꢀ1
4
5.6 cm (STP) min . In contrast to the ketonization reaction, the
esterification reaction rate was not significantly affected by the
presence of CO in the reactor.
To check for the possible formation of the condensed ketone by
the direct ketonization of the esters, a pure ester, propyl hexano-
2
The inhibiting effect of water on the ketonization reaction was
investigated by carrying out experiments at varying concentrations
of water in the feed (up to 5 wt%) while the concentration of the
hexanoic acid was kept constant at 7.5 mol% (Table 3). A significant
inhibiting effect of water on the hexanoic acid ketonization rate was
Table 1
Feed mixtures for reaction kinetics studies.
WHSV range h 1
ꢀ
Feed No.
mol%
Hexanoic acid
1-Pentanol
2-Butanone
Lower
Upper
ꢀ1
ꢀ1
observed (17.7 versus 6.7 lmol min g cat ) when co-feeding
1
2
3
4
5
6
7
8
9
0.10
0.05
0.10
0.15
0.10
0.20
0.10
0.30
0.10
0.10
0.10
0.05
0.10
0.15
0.10
0.20
0.10
0.30
0.80
0.85
0.85
0.75
0.75
0.70
0.70
0.60
0.60
3.5
3.5
3.5
3.5
3.5
3.4
3.5
3.6
3.5
19.7
19.9
19.7
20.1
19.7
19.5
19.7
20.4
19.8
only 5 mol% of water in the hexanoic acid and 2-butanone mixture.
The results presented above show that esterification and
ketonization reactions take place simultaneously over ceria–
zirconia. To assess the functionality of this catalyst, we carried
out TPD experiments, using NH
sites, respectively. After integration of the desorption peak areas,
showed higher number of basic sites
3 2
and CO to probe the acid and base
Ce0.5Zr0.5
O
2
a
Table 2
Effect of carbon dioxide on reaction rates (E = esterification at 548 K, K = ketonization at 598 K).
mol min 1 g cat
ꢀ
ꢀ1
)
Run
mol% CO
gas feed
Rate esterification (lmol min g cat )
ꢀ1
ꢀ1
Run
mol% CO
2
gas feed
Rate ketonization (
l
2
K1-C
K2-C
K3-C
K4-C
0.0%
12.6
3.5
3.7
E1-C
E2-C
E3-C
0.0%
30.8%
54.2%
6.2
7.0
8.9
28.3%
55.5%
0.0%
4.8

7
4
C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
Table 3
network used to describe our experimental data. The absence of
Effect of water on ketonization rate (at 598 K).
the direct ketonization of esters in our experiments may result
from the stronger adsorption of acids on the catalyst, thereby
inhibiting the direct ketonization of esters in the presence of acids.
Rate ketonization (l )
mol min 1 g cat
ꢀ
ꢀ1
Run
mol% water in solvent
K1-W
K2-W
K3-W
0.0%
2.5%
5.0%
17.7
10.2
6.7
2
The TPD results show that Ce0.5Zr0.5O possesses a mixture of
basic and acid sites, with the former being more abundant on its
surface, in agreement with previous work [15]. Whereas basic sites
are needed for ketonization to take place [6,16–18], esterification
reactions are commonly carried out over acidic catalysts [19–21].
Consequently, we assume that ketonization (basic sites) and ester-
ification (acid sites) are catalyzed by different types of sites over
ceria–zirconia. This assumption is justified by the fact that the
esterification rates increase when zirconia oxide, which has more
acidic sites, is used as a catalyst.
(
380
l
mol g catꢀ1) than acidic sites (56
lmol g cat ). Conversely,
ꢀ1
pure ZrO
2
showed
a
considerable number of acidic
ꢀ1
ꢀ1
(
212
l
mol g cat ) and basic sites (296
l
mol g cat ). This in-
creased number of acidic sites for zirconia compared to those for
ceria–zirconia is consistent with the literature [12]. To address
the role of acid and base sites for the esterification and ketonization
reactions, we then carried out selected reaction kinetics studies
over the more acidic zirconium oxide catalyst. Thus, feed 1 (Table
Possible product inhibition by CO has to be taken into consid-
2
eration for the results of the present study, because CO2 adsorbs
1) was reacted over ZrO
2
at temperatures between 548 and 623
strongly on basic sites of ceria–zirconia oxide [15]. The results of
this study, in fact, show strong inhibition of the rate of ketonization
by adsorbed CO2 (Table 2). In contrast, we have observed that
K. A noticeable increase in the rate of esterification (about 2 times
higher than ceria–zirconia) was observed for this catalyst, while
the ketonization rate was not significantly affected.
esterification is not inhibited by CO , which supports our assump-
2
tion that esterification does not take place on basic sites of ceria–
4
. Discussion
zirconia, but rather that it takes place over acid sites.
Experiments carried out by co-feeding water to the reactor
showed an inhibiting effect of water on the ketonization rate
(Table 3). The strong adsorption of water on ceria–zirconia is re-
ported elsewhere [22,23]. For the kinetic model it is assumed that
water blocks the basic sites that are required for the ketonization,
whereas water inhibition is not important for the esterification
reaction. However, inhibition of the ketonization rate by adsorbed
water does not appear to be as significant compared to the inhib-
The results shown in Fig. 2 suggest that 2-butanone does not in-
hibit the catalytic activity for the ketonization and esterification
reactions. Accordingly, surface sites do not appear to be highly cov-
ered by species derived from 2-butanone. A second-order depen-
dence of the ketonization reaction is observed with respect to
the partial pressure of hexanoic acid, for partial pressures between
0
and 0.1 atm. For partial pressures greater than 0.1 atm, the reac-
tion rate follows a zero-order dependence, suggesting that the cat-
alyst sites are highly covered by species derived from hexanoic
acid.
The experiments using propyl hexanoate as a feed showed the
formation of 1-propanol and propylene. This result suggests that
there is a direct mechanism for the conversion of esters to ketones,
producing carbon dioxide, alcohol, and propylene. Different mech-
anisms have been proposed for this direct conversion of esters
2
iting effect of CO adsorption.
The rate of ketonization exhibits a strong temperature depen-
dence. Activation energies and pre-exponential factors for ketoni-
zation and esterification reactions can be determined with
Arrhenius-plots, using experiments with the same feed composi-
tion. Fig. 5 shows the Arrhenius plot of the rate data obtained using
Feed 1 (Table 1). The activation energy values determined were 28
and 134 kJ/mol for the esterification ketonization reactions,
respectively. These values are used as initial guesses in our kinetic
modeling described below.
A simple kinetic model was developed to describe the reaction
kinetics for the combined ketonization and esterification reactions
occurring on the ceria–zirconia catalyst. Previous studies [6,13]
have investigated the ketonization mechanism using various tech-
niques, and agreement about the detailed mechanism has not been
found yet [24]. Thus, in the present study we use simple rate
expressions to capture the essential aspects of the reaction kinet-
ics. The ketonization reaction is modeled as an irreversible reaction
[
13,14], involving various intermediates and products, such as al-
kenes. Importantly, the signature for this direct conversion is the
formation of olefins. In our experiments using 1-pentanol and hex-
anoic acid as the feed, the conversion of the pentyl hexanoate
formed by esterification could follow this direct ketonization route,
as shown in Fig. 4, thereby forming pentene. However, because
measurable amounts of pentene were not observed under our reac-
tion conditions, it appears that the direct ketonization of esters
does not take place in our system. Thus, we only consider two reac-
tions (esterification and ketonization) as part of the reaction
O
O
OH
O
H2O
OH
O
O
O
O
CO2
H2O
OH
Fig. 4. Reaction network for esterification, ketonization, and ester-ketone transformation.

C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
75
4
4
3
3
2
2
1
1
0
0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
150
5
4
125
100
75
Ea = 28.3 kJ/mol
3
2
5
0
5
0
1
2
-
-
-
0.5
1.0
1.5
Ea = 133.9 kJ/mol
450
475
500
525
550
575
600
625
0.0016 0.0017 0.0018 0.0019 0.0020 0.0021 0.0022 0.0023
-
1
-1
T (K)
T
(K )
Fig. 6. Experimental (exp) data and simulation (sim) results for ester formation,
partial pressure of hexanoic acid (hex ac) varied, partial pressure of 1-pentanol
constant at 0.1 atm: (d) 0.05 atm hex ac exp, (1) 0.05 atm hex ac sim, (h) 0.1 atm
hex ac exp, (2) 0.1 atm hex ac sim, (ꢀ) 0.15 atm hex ac exp, (3) 0.15 atm hex ac sim,
Fig. 5. Arrhenius plot for conversion of Feed Mix 1 for (.) esterification and (h)
ketonization reactions.
(O) 0.2 atm hex ac exp, (4) 0.2 atm hex ac sim, (w) 0.3 atm hex ac exp, (5) 0.3 atm
in which two adsorbed hexanoic acid molecules react to form 6-
undecanone, adsorbed carbon dioxide, water, and a vacant catalyst
site. Because of the release of carbon dioxide, the equilibrium con-
stant for this process is high and the reaction can be assumed to be
irreversible. The rates of adsorption and desorption for all gas
phase species are assumed to be quasi-equilibrated. Therefore,
hex ac sim.
nia [26] supports the idea that none of the reactants or products of
esterification is strongly adsorbed on the catalyst surface for the
reaction conditions studied. Accordingly, the esterification reaction
is modeled as a reversible reaction taking place on acidic sites in
which adsorbed hexanoic acid and 1-pentanol react to form 1-pen-
tylhexanoate and water, with negligible blocking of surface sites by
adsorbed species. Thus, the rate expression for the esterification
reaction can be written as Eq. (5).
the fractional surface coverage for species j, h
Eq. (3), where Kads,j is the adsorption equilibrium constant for spe-
cies j, and P is the partial pressure of species j. The fraction of va-
j
, can be written as
j
cant catalyst sites is expressed through the sum of species covering
the catalyst surface subtracted from one. This site balance is used
for the basic sites of the ceria–zirconia catalyst.
ꢀ
ꢁ
P
ester water
P
!
rateesterification ¼ kf;ester ꢁ Phex-ac
P
1 ꢀ Phex-acP
1ꢀpentanol
X
K
eq;ester
1ꢀpentanol
h
j
¼ Kads;j
P
j
1 ꢀ
h
j
ð3Þ
ð5Þ
The model is parameterized by the forward activation energy barri-
j
The experimental results (Fig. 3 and Tables 2 and 3) suggest that
hexanoic acid, CO , and water have a detrimental effect on the keto-
ers, E , and pre-exponential factors, A , for the ketonization and
esterification reactions and the binding energies, BE , for hexanoic
i
i
2
j
nization rate. In this respect, carbon dioxide and water are known to
adsorb strongly on ceria–zirconia [12,15,22,23], and the ketoniza-
tion rate has previously been found to be dependent on the acid
coverage on the surface of the catalyst [25]. Consequently, it is as-
sumed that the surface coverages by these species may be signifi-
cant and thus need to be included in the site-blocking term of the
ketonization rate expression. In contrast, the amount of adsorbed
acid, water, and carbon dioxide. The standard state enthalpies and
entropies for hexanoic acid, 1-pentanol, water, and carbon dioxide
in the gas phase at 448 K were determined from tabulated values
[
p
27]. The heat capacity at constant pressure, C , was calculated at
4
48 K and was assumed to be constant. The heat capacity for each
species was used to calculate the gas phase thermodynamics at
other temperatures. Enthalpy and entropy values for pentylesters
derived from other carboxylic acids (pentyl formate, pentyl ethano-
ate, pentyl propionate, and pentyl butanoate) were extrapolated to
obtain the data for 1-pentylhexanoate. In particular, enthalpy, en-
tropy, and heat capacities for other pentylesters were calculated
at 448 K. The values were plotted versus the carbon number of
the acid part of the ester. The slope and intercept of the plot were
used to extrapolate the value for the pentyl ester of hexanoic acid
6
-undecanone is assumed to be negligible, because we found that
a similar ketone (2-butanone) did not have an effect on the rate
of reaction. The rate expression for the ketonization reaction can
then be written as Eq. (4), where kf,i is the forward rate constant
for reaction i.
2
k
f;keton ꢁ P
hexꢀac
rateketonization
¼
2
ð1 þ Kads;hexꢀac
P
hexꢀac þ Kads;water
P
water þ Kads;CO2
P
CO2
Þ
(
4
6 carbon atoms). All thermochemical values are shown in Table
. The thermodynamic properties for adsorbed species were deter-
ð4Þ
Esterification is reported to be catalyzed by acid catalysts [19–21].
In our case, esterification can be considered as a side reaction which
occurs on the acidic sites of ceria–zirconia. Even though site-block-
ing effects have been reported elsewhere for esterification reactions
mined by adjusting the gas phase enthalpies and entropies with the
binding energies and translational entropies, respectively. The local
entropy values (i.e., rotational and vibrational entropies) of the ad-
sorbed species are calculated by subtracting the three dimensional
translational entropy, calculated using Eq. (6), from the gas phase
entropy. In Eq. (6) m is the mass of the molecule of interest, h is
Plank’s constant, k is Boltzman’s constant and T is the temperature
B
in Kelvin. For simplicity, it is assumed that all of the local entropy is
retained by the adsorbed species. Any errors introduced by entropy
[
19], our observations suggested that site-blocking effects are neg-
ligible for this reaction at the conditions of the present study, be-
cause the rate of esterification increased linearly with the partial
pressure of both the reactants (Figs. 6 and 8). Furthermore, the
weak nature of the acid sites present on the surface of ceria–zirco-

7
6
C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
Table 4
Thermochemistry for esterification.
Enthalpy of formation at 448 K (kJ/mol)
Entropy of formation at 448 K (J/mol/K)
Heat capacity at const. pressure at 448 K (J/mol/K)
1
-Pentanol
ꢀ312
ꢀ520
ꢀ243
ꢀ634
–
274
290
165
768
223
183
218
35
371
43
Hexanoic acid
Water
1
-Pentylhexanoate
Carbon dioxide
assumption would be compensated by adjustments to the values of
the enthalpies and activation energies fit in the kinetic model.
mined from the optimization procedure are shown in Table 5.
Fig. 6–9 show the whole range of the experimental data and the
model predictions obtained with the optimized parameters for
the esterification and ketonization reactions.
"
!
#
ꢂk Tꢃ
3
=2
ð2
p
mk
B
3
TÞ
B
5
2
ꢂ
S
¼ R ln
þ ln
þ
ð6Þ
trans;3D
h
h
Examining the confidence intervals for the predicted parame-
ters in Table 5, it can be seen that the proposed kinetic model de-
scribes the experimental data well. The calculated confidence
intervals are relatively narrow for the activation energies, suggest-
ing that the determined values are appropriate for the conditions
tested and modeling assumptions used. This same statement can
be made for the binding energies of carbon dioxide and hexanoic
acid. To further investigate the validity of the assumption that
inhibiting effects of adsorbed species are negligible for the esterifi-
cation reaction, a site-blocking term (involving adsorbed hexanoic
acid) was included in the esterification rate expression. The good-
ness of the fit did not improve upon addition of this term, thus sug-
gesting that the acid concentrations are too low or the species
binding energies are not significantly strong to necessitate the
inclusion of site blocking for the esterification reaction in the mod-
The enthalpies of the adsorbed species are calculated by adding the
binding energy to the gas phase enthalpy for a given molecule. The
equilibrium constant, Ki,eq, can be calculated from the enthalpy and
entropy of reaction using Eq. (7). The forward rate constants are cal-
culated using the Arrhenius expression in Eq. (8). The reverse rate
constants can be calculated from the equilibrium constant and for-
ward rate constants using Eq. (9).
ꢀ
ꢂꢁ
ꢂ
ꢀ
D
RT
H
D i
R
S
i
K
i;eq ¼ exp
þ
ð7Þ
ꢄ
ꢅ
ꢀ
E
i
RT
k
i
¼ A
i
exp
ð8Þ
ð9Þ
k
i;for
k
i;rev
¼
;
K
i;eq
The kinetic model was implemented in MATLAB using the material
balances for a plug flow reactor. The differential equations that de-
scribe the changing gas phase compositions down the length of the
reactor are solved using a built-in differential equation solver in
Matlab. The activation energies, pre-exponential factors, and bind-
ing energies are optimized to fit the experimental data using the
nonlinear parameter estimation function ‘nlinfit’. For this optimiza-
tion, the Talwar method for a robust optimization implemented in
the MATLAB toolbox was used. The responses used in the parameter
estimation algorithm are the ester and ketone flow rates out of the
reactor. Confidence intervals (95%) are determined for the opti-
mized parameters using the ‘nlparci’ function in MATLAB that uses
a statistical method based on the asymptotic normal distribution
for the parameters estimates.
Based on the experiments shown in Tables 2 and 3, the model
was run without an optimization function and the parameters
were adjusted manually. These adjusted values for the binding
energies of hexanoic acid, water, and carbon dioxide were used
as initial guesses for the optimization runs, including all the exper-
iments shown in Table 1.
8
8
7
7
6
6
5
50
45
40
5
0
5
0
5
0
5
5
4
3
35
30
25
20
15
10
5
0
2
1
550
575
600
625
T (K)
In the optimization runs, the binding energy of water on the
ceria/zirconia surface turned out to be less sensitive than the other
parameters for calculating the overall rate. Thus, it was set to the
physically reasonable value ꢀ96 kJ/mol for the final optimization
runs. The model parameters and their confidence intervals deter-
Fig. 7. Experimental data (exp) and simulation (sim) results for ketonization,
partial pressure of hexanoic varied (hex ac), partial pressure of 1-pentanol constant
at 0.1 atm: (d) 0.05 atm hex ac exp, (1) 0.05 atm hex ac sim, (h) 0.1 atm hex ac exp,
(
2) 0.1 atm hex ac sim, (ꢀ) 0.15 atm hex ac exp, (3) 0.15 atm hex ac sim, (O) 0.2 atm
hex ac exp, (4) 0.2 atm hex ac sim, (w) 0.3 atm hex ac exp, (5) 0.3 atm hex ac sim.
Table 5
Optimized values of kinetic parameters.
Optimized value
Lower confidence interval
Upper confidence interval
Unit
Activation energy
Pre-exponential factor
Binding energy
Esterification
Ketonization
Esterification
Ketonization
Hexanoic acid
Carbon dioxide
Water
40.4
132.2
39.8
41.0
kJ/mol
kJ/mol
–
129.4
5.7E+05
7.6E+14
ꢀ84.1
ꢀ139.7
–
135.1
7.4E+05
8.0E+14
ꢀ78.0
ꢀ136.1
–
6.6E+05
7.8E+14
ꢀ81.1
ꢀ137.9
ꢀ96.0
–
kJ/mol
kJ/mol
kJ/mol

C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
77
1
1
1
1
75
50
25
00
ported elsewhere [28], showing strong adsorption of CO
particular, a fraction of the adsorbed CO
high temperature (860 K). Thus, the high binding energy for CO
predicted from the model (Table 5) is consistent with previous
studies reported.
The kinetics of the esterification and ketonization reactions are
described well with our simple model in a concentration range
typical of biomass-upgrading intermediates. Accordingly, this
model can then be used to predict the reaction conditions required
to upgrade various kinds of biomass-derived feeds over a Ce0.5Z-
. In
2
2
desorbs at a relatively
2
5
4
3
7
5
2
5
0
5
0
2
0.5 2
r O catalyst, as carboxylic acids are present in various bio-
1
mass-derived feeds [29]. In this respect, we have employed this
ceria–zirconia catalyst to ketonize fully the carboxylic acids and
esters present in liquid organic streams derived from the process-
ing of glucose over a Pt–Re/C catalyst [5].
The ketonization of esters, which are an essential intermediate
product in many biomass-conversion processes [30], is a more
difficult problem compared to the conversion of carboxylic acids,
and the literature does not agree about the details of the reaction
mechanism [13,14]. One of the suggestions is the intermediate
formation of an acid by hydrolysis, which we assume is the pre-
dominant pathway in our system. Thus, one approach to convert
a large amount of esters to ketones is to provide water with the
reaction mixture fed to the reactor. Water, however, decreases
450
475
500
525
550
575
600
625
T (K)
Fig. 8. Experimental data (exp) and simulation (sim) results for ester formation,
partial pressure of 1-pentanol (pent) varied, partial pressure of hexanoic acid
constant at 0.1 atm: (d) 0.05 atm pent exp, (1) 0.05 atm pent sim, (h) 0.1 atm pent
exp, (2) 0.1 atm pent sim, (ꢀ) 0.15 atm pent exp, (3) 0.15 atm pent sim, (O) 0.2 atm
pent exp, (4) 0.2 atm pent sim, (w) 0.3 atm pent exp, (5) 0.3 atm pent sim.
2
the activity of the Ce0.5Zr0.5O catalyst, so this approach is not
7
6
6
5
5
4
4
3
3
2
2
1
1
0
5
0
5
0
5
0
5
0
5
0
5
0
5
0
the optimal solution. However, it is possible that biomass-derived
intermediate feeds will contain acids and esters, as we have ob-
served in the conversion of glucose over Pt–Re/C. Because of the
stronger adsorption of acids on the catalyst surface compared to
esters, direct ketonization of esters will not take place as long
as acids are present. Thus, ketonization of acids takes place pref-
erentially compared to ketonization of esters. Accordingly, our
model does not include the direct ketonization route from esters,
which is a limitation. Above 623 K and without presence of acids
this reaction becomes essential. Thus, expanding the model for
this reaction and for temperatures above 623 K is a goal for future
study. Importantly, the water formed by the ketonization of car-
boxylic acids leads to the subsequent hydrolysis of esters to form
carboxylic acids and alcohols, which is followed by the ketoniza-
tion of these newly formed carboxylic acids. In this way, the cou-
pling between ketonization and hydrolysis reactions (the latter
being the reverse of esterification) provides an efficient pathway
for the condensation of esters to larger ketones in the tempera-
ture range of the present study.
1
3
2
4
5
550
575
600
625
T (K)
Fig. 9. Experimental data (exp) and simulation (sim) results ketonization, partial
pressure of 1-pentanol (pent) varied, partial pressure of hexanoic acid constant at
0
0
.1 atm: (d) 0.05 atm pent exp, (1) 0.05 atm pent sim, (h) 0.1 atm pent exp, (2)
.1 atm pent sim, (ꢀ) 0.15 atm pent exp, (3) 0.15 atm pent sim, (O) 0.2 atm pent
5. Conclusions
exp, (4) 0.2 atm pent sim, (w) 0.3 atm pent exp, (5) 0.3 atm pent sim.
The ketonization of carboxylic acids to form ketones was carried
out using a feed containing primary alcohols over a ceria–zirconia
catalyst. The esterification reaction proceeds more rapidly than the
ketonization at lower temperatures, while ketonization becomes
the dominant process at the higher temperatures, in good agree-
ment with the higher predicted activation energy for ketonization
(132 kJ/mol) compared to the esterification reaction 40 kJ/mol.
This difference in activation energy indicates that temperatures
higher than approximately 548 K are required to make the ketoni-
zation reaction rate competitive with the rate of esterification. Be-
cause ketonization is an irreversible reaction, a depletion of
hexanoic acid occurs when the rate of ketonization becomes signif-
icant. At high ketonization rates, the reverse esterification reaction
becomes significant as the equilibrium shifts to replace hexanoic
acid that is consumed by ketonization.
el. As seen from the results of the kinetic model, the esterification
reaction has a lower activation energy barrier than the ketoniza-
tion reaction and thus occurs at lower temperatures. The rate of
ketonization becomes significant at temperatures higher than
approximately 548 K. The value of the ketonization activation en-
ergy determined here (132 kJ/mol) is similar to the value reported
elsewhere (159 kJ/mol) [25], even though a different catalyst and a
different carboxylic acid were used. It has been suggested [25] that
for temperatures below 673 K, adsorbed carboxylic acid species ap-
pear to have the most significant effect on the rate ketonization,
followed by water and then CO
2
. From the present study, it was
determined that adsorbed CO and water have significant effects
2
on the rate of ketonization, followed by the effect of adsorbed hex-
anoic acid. However, our study used a different catalyst that is
Hexanoic acid adsorption on the catalyst surface is an important
step in the reaction, and the rate of ketonization shifts from second
order to zero order as the partial pressure of hexanoic acid in-
known to adsorb CO
2
strongly. In this respect, results from temper-
ature-programmed desorption studies of ceria oxide have been re-

7
8
C.A. Gaertner et al. / Journal of Catalysis 266 (2009) 71–78
creases. Product inhibition takes place through strong binding of
CO , and to a lesser extent water, on the basic sites.
[10] J.C. Serrano-Ruiz, J. Luettich, A. Sepulveda-Escribano, F. Rodriguez-Reinoso, J.
Catal. 241 (2006) 45.
2
[
[
11] D.A. Simonetti, E.L. Kunkes, J.A. Dumesic, J. Catal. 247 (2007) 298.
12] M. Daturi, C. Binet, J.-C. Lavalley, A. Galtayries, R. Sporken, Phys. Chem. Chem.
Phys. 1 (1999) 5717.
Acknowledgments
[
13] R. Klimkiewicz, H. Grabowska, L. Syper, Kinet. Catal. (Translation of Kinetika i
Kataliz) 44 (2003) 283.
This work was supported by the U.S. Department of Energy Of-
fice of Basic Energy Sciences and the National Science Foundation
Chemical and Transport Systems Division of the Directorate for
Engineering. C.A. Gaertner thanks the German Academic Exchange
Service (DAAD) for a scholarship. J.C. Serrano-Ruiz thanks the
Spanish Ministry of Science and Innovation for postdoctoral sup-
port. We thank E.L. Kunkes for the temperature-programmed
desorption experiment results and for valuable advice. We also
thank M. Mavrikakis, R.M. West, D. Wang and E.I. Gürbüz for valu-
able discussions and technical assistance.
[14] M. Glinski, W. Szymanski, D. Lomot, Appl. Catal. A 281 (2005) 107.
[
[
[
[
[
[
15] V. Solinas, E. Rombi, I. Ferino, M.G. Cutrufello, G. Colon, J.A. Navio, J. Mol. Catal.
A: Chem. 204–205 (2003) 629.
16] Y. Kamimura, S. Sato, R. Takahashi, T. Sodesawa, T. Akashi, Appl. Catal. A 252
(2003) 399.
17] G.A.H. Mekhemer, S.A. Halawy, M.A. Mohamed, M.I. Zaki, J. Catal. 230 (2005)
109.
18] O. Nagashima, S. Sato, R. Takahashi, T. Sodesawa, J. Mol. Catal. A: Chem. 227
(2005) 231.
19] D.E. Lopez, K. Suwannakarn, J.G. Goodwin Jr., D.A. Bruce, Ind. Eng. Chem. Res.
47 (2008) 2221.
20] B. Schmid, M. Doeker, J. Gmehling, Ind. Eng. Chem. Res. 47 (2008) 698.
[21] T.J. Schildhauer, I. Hoek, F. Kapteijn, J.A. Moulijn, Appl. Catal. A 358 (2009) 141.
[
[
22] L. Deng, Y. Fu, Q.-X. Guo, Energy Fuels 23 (2008) 564.
23] J.I. Gutierrez-Ortiz, B. de Rivas, R. Lopez-Fonseca, J.R. Gonzalez-Velasco, J.
Therm. Anal. Calorim. 80 (2005) 225.
References
[
24] R. Pestman, R.M. Koster, A. van Duijne, J.A.Z. Pieterese, V. Ponec, J. Catal. 168
(1997) 265.
[
1] C.H. Christensen, J. Rass-Hansen, C.C. Marsden, E. Taarning, K. Egeblad, Chem.
Sus. Chem. 1 (2008) 283.
[25] S. Rajadurai, Catal. Rev.-Sci. Eng. 36 (1994) 385.
[
[
[
[
2] G.W. Huber, A. Corma, Angew. Chem., Int. Ed. 46 (2007) 7184.
3] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044.
4] D.A. Simonetti, J.A. Dumesic, Chem. Sus. Chem. 1 (2008) 725.
5] E.L. Kunkes, D.A. Simonetti, R.M. West, J.C. Serrano-Ruiz, C.A. Gaertner, J.A.
Dumesic, Science 322 (2008) 417.
[26] M.G. Cutrufello, I. Ferino, V. Solinas, A. Primavera, A. Trovarelli, A. Auroux, C.
Picciau, Phys. Chem. Chem. Phys. 1 (1999) 3369.
[27] C.L. Yaws, Knovel, Yaws’ handbook of thermodynamic and physical properties
of chemical compounds: Physical, thermodynamic and transport properties
for 5000 organic chemical compounds, 2003.
[
[
[
[
6] M. Renz, Eur. J. Org. Chem. (2005) 979.
[28] J. Stubenrauch, E. Brosha, J.M. Vohs, Catal. Today 28 (1996) 431.
[29] B.J. Nikolau, M.A.D.N. Perera, L. Brachova, B. Shanks, Plant J. 54 (2008) 536.
[30] D.C. Rennard, P.J. Dauenhauer, S.A. Tupy, L.D. Schmidt, Energy Fuels 22 (2008)
1318.
7] A. Corma, M. Renz, C. Schaverien, Chem. Sus. Chem. 1 (2008) 739.
8] K.M. Dooley, A.K. Bhat, C.P. Plaisance, A.D. Roy, Appl. Catal. A 320 (2007) 122.
9] R. Martinez, M.C. Huff, M.A. Barteau, J. Catal. 222 (2004) 404.