Vapour-phase C-C coupling reactions of biomass-derived oxygenates over Pd/CeZrOx catalysts

Article abstract of DOI:10.1016/j.jcat.2009.06.014

Studies of aldol condensation/hydrogenation reactions of 2-hexanone were carried out over Pd/CeZrO_x and CeZrO_x catalysts at temperatures between 573 and 673 K, and pressures of 5-26 bar. These studies were formulated to address the catalytic upgrading to transportation fuels of the mono-functional oxygenated compounds (consisting primarily of C₄-C₆ ketones, alcohols, carboxylic acids and heterocyclics) formed by the catalytic conversion of polyols over a Pt-Re/C catalyst. Characterization by XRD, TPR and NH₃/CO₂-TPD showed that Pd/CeZrO_x catalyst consists of a partially reducible solid solution of cerium and zirconium oxides, and possesses both acidic and basic functionalities. Reaction kinetics studies show that in addition to the expected C₁₂ condensation product (7-methyl-5-undecanone), the CeZrO_x-based catalysts produce C₁₈ and C₉ secondary species, along with light alkanes (≤C₇). Low loadings of Pd (e.g., 0.25 wt%) lead to optimal activity and selectivity for the production of C₁₂ species. The high apparent activation energy of the formation of C₉ (140 kJ/mol) compared to the formation of C₁₂ and C₁₈ species (15 and 28 kJ/mol, respectively) indicates that these species may be formed as a result of the decomposition of heavier condensation products. The self-coupling of 2-hexanone was found to be positive order in both 2-hexanone and hydrogen. The addition of primary alcohols and carboxylic acids as well as water and CO₂ to the feed was found to reversibly inhibit the self-coupling activity of 2-hexanone. This inhibition is strongest in the presence of CO₂, and TPSR studies indicate that CO₂ is removed from the surface by conversion to CO in the presence of reduced ceria species.

Full text of DOI:10.1016/j.jcat.2009.06.014

Journal of Catalysis 266 (2009) 236–249
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Vapour-phase C–C coupling reactions of biomass-derived oxygenates
over Pd/CeZrOx catalysts
*
Edward L. Kunkes, Elif I. Gürbüz, James A. Dumesic
Department of Chemical and Biological Engineering, University of Wisconsin, 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:
Studies of aldol condensation/hydrogenation reactions of 2-hexanone were carried out over Pd/CeZrOx
and CeZrOx catalysts at temperatures between 573 and 673 K, and pressures of 5–26 bar. These studies
were formulated to address the catalytic upgrading to transportation fuels of the mono-functional oxy-
genated compounds (consisting primarily of C4–C6 ketones, alcohols, carboxylic acids and heterocyclics)
formed by the catalytic conversion of polyols over a Pt–Re/C catalyst. Characterization by XRD, TPR and
NH3/CO2–TPD showed that Pd/CeZrOx catalyst consists of a partially reducible solid solution of cerium
and zirconium oxides, and possesses both acidic and basic functionalities. Reaction kinetics studies show
that in addition to the expected C12 condensation product (7-methyl-5-undecanone), the CeZrOx-based
catalysts produce C18 and C9 secondary species, along with light alkanes (6C7). Low loadings of Pd
(e.g., 0.25 wt%) lead to optimal activity and selectivity for the production of C12 species. The high apparent
activation energy of the formation of C9 (140 kJ/mol) compared to the formation of C12 and C18 species
(15 and 28 kJ/mol, respectively) indicates that these species may be formed as a result of the decompo-
sition of heavier condensation products. The self-coupling of 2-hexanone was found to be positive order
in both 2-hexanone and hydrogen. The addition of primary alcohols and carboxylic acids as well as water
and CO2 to the feed was found to reversibly inhibit the self-coupling activity of 2-hexanone. This inhibi-
tion is strongest in the presence of CO2, and TPSR studies indicate that CO2 is removed from the surface by
conversion to CO in the presence of reduced ceria species.
Received 20 April 2009
Revised 20 May 2009
Accepted 14 June 2009
Available online 17 July 2009
Keywords:
Aldol-condensation
Hydrogenation
Ceria
Zirconia
2-Hexanone
Biofuels
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
In our recent work, we have developed a catalytic process that
can be tuned to convert biomass-derived carbohydrates to a vari-
Petroleum currently supplies more than 95% of the energy
requirements for the US transportation sector [1]. This non-renew-
able fossil fuel resource is in diminishing supply, and its combus-
tion contributes to the accumulation of CO2 in the atmosphere,
leading to global climate change. However, petroleum-derived
fuels possess optimal physical properties with respect to energy
density, volatility and hydrophobicity, which allow for efficient
storage, distribution and combustion. Biomass-derived fuels have
attracted considerable attention as alternatives to petroleum, be-
cause they are derived from abundant, renewable resources, and
because combustion of these fuels is CO2 neutral. However, obtain-
ing biomass-derived fuels with the appropriate physical and com-
bustion properties to integrate into the current transportation
infrastructure has presented numerous challenges. For example,
utilization of the most successful biomass-derived fuel, ethanol,
necessitates modifications to the internal combustion engine of
the vehicle and the use of bio-diesel requires special precautions
in cold climates [2,3].
ety of fuel-grade products, such as light alkenes (C1–C4), benzene,
substituted aromatics, branched alkenes (C6–C12) and linear or sin-
gly branched alkanes (C7–C12). This process consists of an initial
de-oxygenation step that converts concentrated aqueous solutions
of sugars and polyols (40–60%) over a carbon-supported Pt–Re cat-
alyst into hydrophobic mixtures of mono-functional C4–C6 alco-
hols, ketones, carboxylic acids and heterocyclic compounds [4].
The carbon chain length of these monofunctional species is limited
to six by the chain length of the parent carbohydrate, i.e., glucose
and sorbitol. To produce higher molecular weight species required
for gasoline, diesel and jet fuel applications, the mixture of mono-
functional species is subsequently subjected to catalytic C–C cou-
pling processes. These coupling processes include (i)
ketonization, in which two carboxylic acid molecules react to form
a linear ketone, CO2 and water [5], and (ii) aldol condensation/
hydrogenation, in which two ketone or alcohol molecules react
to form a heavier branched ketone [6]. The combination of ketoni-
zation and aldol condensation/hydrogenation yields C7–C12 linear
and branched ketones, which can be converted into fuel-grade al-
kanes via dehydration/hydrogenation over solid acid supported
noble metal catalysts, such as Pt/NbOPO4 [7].
* Corresponding author.
E-mail address: dumesic@engr.wisc.edu (J.A. Dumesic).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.06.014

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
237
Ketonization of carboxylic acids has been investigated in the lit-
erature using a variety of basic, acidic and amphoteric oxide cata-
lysts, and occurs at temperatures near or higher than 623 K [6,8,9].
We have demonstrated that the carboxylic acids found in the
aforementioned carbohydrate-derived mixture of monofunctional
species can be ketonized with nearly 100% yield over a CeZrOx
mixed oxide catalyst at temperatures from 623 to 673 K [4]. The
combined aldol condensation/hydrogenation of alcohols and ke-
tones has been studied in the literature on metals (Cu and Pd) sup-
ported on mixed oxides such as MgAlOx [5,10] and anion-exchange
resins [11] in the temperature range from 423 to 473 K in the pres-
ence of hydrogen. This reaction consists of metal-catalyzed dehy-
drogenation of alcohol groups, followed by base-catalyzed aldol
condensation, acid-catalyzed dehydration and subsequent metal-
monofunctional oxygenates present in the carbohydrate-derived
hydrophobic mixture, such as primary and secondary alcohols, car-
boxylic acids, heterocyclics and 3-ketones on the self-condensation
of 2-hexanone. Furthermore, we study the effects of the ketoniza-
tion by-products, CO2 and water, on the downstream condensation
reaction, and we probe the interaction of CeZrOx with CO2 using
temperature-programmed desorption (TPD), temperature-pro-
grammed reduction (TPR) and temperature-programmed surface
reaction (TPSR) of CO2. We observe that the addition of oxygenates
as well as water and CO2 all lead to reversible inhibition of the self-
condensation activity, and this inhibition is strongest when CO2 is
present in the feed. Through these observations we conclude that it
is beneficial to remove water and CO2 prior to the aldol/condensa-
tion hydrogenation step over Pd/CeZrOx.
catalyzed hydrogenation of the
a–b unsaturated aldol adduct.
Hydrogenation of the C@C bond in the dehydrated aldol adduct im-
proves the overall thermodynamics for the process, allowing a
greater extent of C–C coupling than aldol condensation/dehydra-
tion alone [5,10,12]. Several investigators report that the presence
of intermediate strength acidic and basic sites in proximity results
in optimal catalytic performance [10,13–15], and the presence of
strong acid or basic sites leads to undesirable side reactions that
may also lead to coking [16].
Most studies of aldol condensation/hydrogenation have focused
on self-condensation of acetone or isopropanol to form MIBK
[5,12,17–21], and few reports have been published regarding
self-condensation of higher ketones [22] (such as the ones pro-
duced from the conversion of carbohydrates on Pt–Re/C). In con-
trast to acetone, C4+ methyl ketones possess a single set of
2. Experimental
2.1. Catalyst preparation
The CeZrOx support with a 1:1 Ce:Zr molar ratio was prepared
via co-precipitation of Ce(NO3)3 and ZrO(NO3)2 with NH4OH (Al-
drich) according to Serrano-Ruiz et al. [27]. The Pd/CeZrOx catalyst
was prepared via incipient wetness impregnation of CeZrOx with
an amount of aqueous solution of Pd(NO3)2 (Aldrich) necessary
to obtain nominal metal loadings of 0.05, 0.25 and 1.45 wt%. The
catalyst was dried in air at 373 K overnight and calcined in air at
623 K for 2 h. A 0.5 wt% Pd/SiO2 catalyst was prepared via the same
procedure, using a Cab-O-Sil silica gel support.
reactive
a-hydrogen atoms, resulting in a substantial decline in
2.2. Reaction kinetics studies
reactivity when compared to acetone. As a result, higher reaction
temperatures are required to achieve appreciable conversion, and
appropriate catalysts must be used that are resistant to coking
and sintering at these conditions. In this respect, we have reported
the self-condensation/hydrogenation of 2-hexanone on a CuM-
g10Al7Ox catalyst at 573 K and have shown that even small
amounts of carboxylic acids are detrimental to the self-condensa-
tion reaction [23]. For this reason, the ketonization reaction pro-
vides an efficient means of removing carboxylic acids from the
feed prior to aldol condensation/hydrogenation, and we have dem-
onstrated that a catalyst consisting of Pd-supported on CeZrOx is
effective at 623 K for the condensation of ketones present in the
ketonized mixture of monofunctional species derived from glucose
[4]. The CeZrOx support was selected because it possesses high lat-
tice oxygen mobility and the ability to interact strongly with sup-
2-Hexanone, 1-butanol, tetrahydrofuran, heptane, 3-pentanone
and butanoic acid were purchased from Sigma–Aldrich and were
used without further purification. Hydrogen (99%), helium
(99.9%) and 10 mol% CO2 in helium were purchased from Airgas
and used without further purification. The conversions of oxygen-
ated species over Pd/CeZrOx were carried out at temperatures be-
tween 573 and 673 K, pressures of 5–26 bar, and liquid flow
rates between 0.02 and 0.08 mL/min. The molar ratio of the gas
flow rate to the liquid flow rate was maintained at 5.5, with the
exception of reaction order studies. A fixed bed, down-flow reactor
consisting of a half-inch stainless steel tube, shown schematically
in Fig. 1, was used for all experiments. Quartz wool was used in
the lower end of the reactor to keep the catalyst bed in place.
The catalyst was mixed with crushed fused quartz granules (Al-
drich) in a 2:1 volumetric ratio to maintain bed height. The reactor
was heated with an aluminium block that was heated externally by
a well-insulated furnace (Applied Test Systems). Type-K thermo-
couples (Omega) were used to measure the reaction temperature,
which was controlled by a PID controller (Love controls) connected
to a variable transformer (Tesco). Mass flow controllers (Brooks
5850E) were used to regulate the flow of H2, He and CO2
(10 mol% in He) during the experiments. The liquid feed was
pumped from a graduated cylinder by an HPLC pump (Lab Alliance
series 1) to a needle located at the entrance of the catalyst bed. A
back-pressure regulator (GO model BP-60) was used to control
the total pressure, which was measured by two gauges at the en-
trance and the exit of the bed. A gas–liquid separator at room tem-
perature was used to collect the liquid phase for analysis. The
ported metals
– properties that have been associated with
resistance to the formation of carbonaceous species that lead to
deactivation [24,25]. Additionally, CeZrOx contains a combination
of acidic and basic functionalities for aldol condensation [26]. Thus,
we suggest that ketonization and aldol condensation/hydrogena-
tion can be carried out at comparable operating conditions, and
can potentially be integrated into a single reactor system, contain-
ing a catalytic bed consisting of CeZrOx to perform ketonization,
followed by a downstream bed of Pd/CeZrOx to perform aldol con-
densation/hydrogenation.
In the current work, we explore the possibility of the aforemen-
tioned integration between ketonization and aldol condensation/
hydrogenation by elucidating the reaction pathways for aldol con-
densation of a representative ketone (2-hexanone) on Pd/CeZrOx
by varying reaction conditions, such as temperature, pressure
and space velocity. We show that in addition to the expected C12
condensation product (7-methyl-5-undecanone), the catalyst pro-
duces C18 and C9 secondary species. We investigate the effect of
metal loading and proximity of metal and support on catalytic
activity, and we find that low metal loadings lead to optimal activ-
ity and selectivity. Additionally, we investigate the effect of other
catalyst was reduced in situ at 623 K (ramp rate of 0.5 K minꢀ1
)
for 2 h in flowing H2 [200 cm3(STP) minꢀ1]. After the reduction
was completed, the temperature and pressure were adjusted and
the feed flow was started with flowing H2 or H2–He or H2–CO2
mixtures. The weight hourly space velocity (WHSV) was calculated
for experiments using the mass flow rate of the liquid flow into the

238
E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
Pressure
Gauge
Furnace
Mass Flow
Controllers
HPLC
Pump
Feed
Reservoir
Quartz
Chips
Catalyst
Bed
He
H2
Temperature
Controller
Quartz
Wool
Thermocouple
Gas
Outlet
Back-Pressure
Regulator
Gas-Liquid
Separator
Fig. 1. Reactor system used in reaction kinetics studies.
ꢀ
ꢁ
R2-hexanone
þ R2-hexanol
reactor and the mass of the catalyst used (2 g, except for reaction
order studies).
out
out
f2-hexanone
¼
1 ꢀ
ꢁ 100%
ð1Þ
ð2Þ
R2-hexanone
in
0
1
The rate of gas production was measured with a bubble flow
meter. An HP GC5890 gas chromatograph, equipped with a Haysep
DB 100/120 column (Alltech) and thermal conductivity detector
(TCD), was used to quantify CO and CO2, and an Agilent GC6890,
equipped with an Rtx column (Agilent) and a flame ionization
detector (FID), was used to quantify gas-phase alkanes. Liquid
phase analysis was performed with a Shimadzu GC 2060, equipped
with a DB-5 column (Restek) and an FID detector. Unknown spe-
cies were identified with a Shimadzu 2060 GC/MS with a NIST li-
brary of spectra. Some species could not be identified definitively
using the library, and their identification was based on the atomic
mass of the molecular ion fragment and analogous species re-
ported for acetone condensation. Liquid and gas analysis points
were collected every 2–3 h, and steady state was usually achieved
after 6 h time-on-stream. At our reaction conditions on Pd-con-
taining catalysts, 2-hexanol was always in equilibrium with 2-
hexanone. Therefore, the conversion of 2-hexanone to 2-hexanol
was not factored into the total conversion calculation, as given
by Eq. 1. The selectivities to reaction products were calculated on
a molar carbon basis, as given by Eq. 2. In Eqs. (1) and (2), R repre-
sents the molar flow rate, f represents reactant conversion, S repre-
sents the carbon selectivity, and N is the carbon number of a given
species. All carbon balances closed to within 10%
Rj outNj
ðRj outNjÞ
B
C
A
P
S ¼
j
ꢁ 100%:
@
j
2.3. Catalyst characterization
2.3.1. CO chemisorption and BET measurements
The adsorption uptakes of carbon monoxide at 300 K were mea-
sured on a standard gas adsorption apparatus described elsewhere
[28]. The number of surface metal sites was taken to be the irre-
versible CO uptake. The BET surface area was measured by nitro-
gen adsorption at liquid nitrogen temperature on the same
apparatus.
2.3.2. XRD
X-ray diffraction (XRD) was used to investigate the phase struc-
ture of the Pd/CeZrOx catalyst. A Scintag PAD V X-ray diffractome-
ter with a monochromated CuK X-ray tube was used in the
diffraction studies. The tube voltage and current were 45 kV and
40 mA, respectively. Diffraction patterns were collected in the
20° to 65° 2h range, with 0.01° intervals and a dwell time of 2 s.
a

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
239
2.3.3. TPD
Carbon dioxide and ammonia temperature-programmed
desorption (TPD) experiments were carried out using an apparatus
consisting of a mass flow controller (Teledyne-Hastings) and a tube
furnace connected to a variable power-supply and PID temperature
controller (Love Controls) with a K-type thermocouple (Omega).
The effluent was monitored by a mass spectrometer system con-
sisting of a quadruple residual gas analyzer (Stanford Instruments
RGA 200) inside a vacuum chamber. Vacuum was provided by a
diffusion pump connected in series to a rotary pump. The effluent
was introduced into the vacuum chamber via a constricted quartz
capillary, resulting in a pressure of 5 ꢁ 10ꢀ5 Torr inside the cham-
ber. Dried, unreduced catalyst samples were loaded into a 12.6 mm
(0.5 in.) outer diameter, fritted quartz tube reactor. Prior to desorp-
tion experiments, 0.3–1 g of catalyst was reduced in flowing
hydrogen (100 cm3(STP) minꢀ1) for 2 h at 623 K. The catalyst was
degassed for 1 h in flowing helium (200 cm3(STP) minꢀ1) at the
reduction temperature, and was then cooled to 300 K for CO2
adsorption and 423 K for NH3 adsorption, respectively. Carbon
dioxide was adsorbed onto the reduced catalysts by exposure to
flowing 10 mol% CO2 in helium (100 cm3(STP) minꢀ1) for 30 min.
Residual CO2 was removed by purging the catalyst with helium
(200 cm3(STP) minꢀ1) at 300 K for 2 h. Ammonia adsorption was
performed in a similar manner as CO2 adsorption, except that both
adsorption and purging steps were performed at 423 K, and a
1 mol% NH3 in He gas mixture was used. Desorption of CO2 or
NH3 was performed by heating the catalyst at a rate of 10 K minꢀ1
under flowing helium (50 cm3(STP) minꢀ1) from room temperature
to 1073 K for NH3 or 1123 K for CO2.
2.3.4. TPR and TPSR measurements
Temperature-programmed surface reaction (TPSR) experiments
followed the same procedure as those involving TPD of CO2, except
that the desorption of CO2 was performed in flowing hydrogen
(50 cm3(STP) minꢀ1). Temperature-programmed reduction (TPR)
measurements were performed by heating 0.2–0.3 g of calcined
Fig. 2. NH3 desorption profiles (A) and CO2 desorption profiles (B) from CeZrOx and
Pd/CeZrOx catalysts.
catalyst at 10 K minꢀ1 under
a 5 mol% H2 in N2 mixture
(50 cm3(STP) minꢀ1) to 973 K, and measuring hydrogen consump-
tion with the MS system described above.
Bell showed that the presence of a hydrogen-dissociating metal
on CeZrOx causes the reduction of Ce4+ into Ce3+ via hydrogen spill-
over, and the presence of Ce3+ is shown to increase the acidity of
the CeZrOx surface [30]. In this respect, the 1.45 wt% Pd/CeZrOx cat-
alyst may contain sufficient concentrations of Ce3+ to influence the
acidic site properties, as measured by NH3 TPD.
3. Results and discussion
3.1. Surface and bulk characterization
The CO2 desorption profiles show the presence of basic sites of
varying CO2 adsorption strength. These profiles can be divided into
three distinct regions representing weak adsorption sites (300–
500 K), medium-strength adsorption sites (500–700 K), and strong
adsorption sites (700–1000 K). Di Cosimo et al. have correlated TPD
and FTIR studies of CO2 adsorbed on MgAlOx and have associated
the CO2 desorption temperature with different types of CO2 bind-
ing sites. The weakly binding sites are associated with surface
OH groups and the formation of bicarbonate, while the medium-
strength sites are associated with Mx+–O2ꢀ pairs and the formation
of bridged and bidentate carbonates. The strongly binding sites are
The desorption profiles of NH3 and CO2 from CeZrOx and several
Pd/CeZrOx catalysts of various loadings are shown in Fig. 2. These
profiles were integrated to obtain the quantities of acidic and basic
surface sites, which are reported in Table 1. The values of the irre-
versible CO uptake and BET areas are also listed out in Table 1. The
NH3 desorption profiles of all catalysts show a single, uniform
desorption peak, centered at 570–600 K, suggesting a narrow dis-
tribution of acidic sites on the surface. These results are consistent
with those obtained by Zhu et al. for NH3 TPD from Pt/CeZrOx [29].
The 1.45 wt% Pd catalyst contains nearly 50% more NH3 adsorption
sites than CeZrOx and the lower loading catalysts. Pokrovski and
Table 1
Quantification of metal (CO), basic (CO2) and acidic sites (NH3), and BET surface areas of selected catalysts.
Catalyst
CO (irreversible) (lmol/g)
Apparent Pd dispersion (%)a
CO2 (TPD) (lmol/g)
NH3 (TPD) (lmol/g)
BET area (m2/g)
CeZrOx
0
8
18
76
9
–
370
390
380
395
–
56
50
54
77
–
130
135
133
139
–
0.05% Pd/CeZrOx
0.25% Pd/CeZrOx
1.45% Pd/CeZrOx
0.5% Pd/SiO2
170
77
57
20
a
The Pd dispersion is over-estimated because of CO chemisorption on reduced sites on the support.

240
E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
associated with low coordination O2ꢀ ions and the formation of
more strongly bound unidentate carbonates [14]. Several of the
aforementioned species have been observed on ZrO2 [31], and it
is probable that the correlation found by Di Cosimo et al. for MgA-
lOx may be extended to other oxides such as CeZrOx.
Di Cosimo et al. and Bianchi et al. observed weak CO2 adsorp-
tion sites on CeO2 and weak- and medium-strength sites on ZrO2,
respectively [21,32]. The presence of stronger CO2 binding sites
on CeZrOx surface likely results from the surface heterogeneity
introduced by substituting Ce ions with Zr4+ ions. The addition of
Zr to Ce is known to increase the mobility of O2ꢀ ions in the lattice
[27,33], and is therefore likely to cause higher concentrations of
low coordination O2ꢀ (associated with surface defects) that
strongly bind CO2. The number and strength of CO2 binding sites
are not influenced significantly by metal loading. This observation
is different from that of Nikolopolous et al. who observed that the
surface basicity decreases with increasing Pd loading on MgAlOx.
Nikolopolous et al. suggest that residual chlorine from the metal
precursor (PdCl2) is responsible for this effect [5]. This detrimental
effect on basicity was not observed in the present investigation,
probably due to the use of a nitrate precursor (Pd(NO3)2) that
decomposed to PdO upon calcination.
Fig. 3. XRD pattern of 0.25 wt% Pd/CeZrOx. Dashed lines represent reflections from
fluorite cubic phase of CeO2.
The BET surface area is not affected by metal loading in the
loading range studied here, suggesting that there is no modifica-
tion to the pore structure upon metal deposition. The BET surface
area (130–140 m2/g) is consistent with measurements obtained
by other investigators on CeZrOx prepared by the co-precipitation
method [27].
Measurements of irreversible CO adsorption show that the CO
uptake increases with metal loading. When a reducible oxide sup-
port is present (such as CeO2), reduced sites on the support can
contribute to the CO uptake, which normally occurs only on metal
sites [34]. This phenomenon is particularly evident in the case of
0.05 wt% Pd/CeZrOx, where the amount of chemisorbed CO appears
to exceed the number of Pd atoms in the catalyst. TPR studies in
the present work and by others show that hydrogen spillover is
responsible for the reduction of surface Ce4+ into Ce3+, and this
reduction takes place at the interface of the metal and the support
[27]. For smaller particles, the ratio of these interfacial sites to me-
tal sites is higher than that for large particles, leading to a greater
normalized CO uptake (apparent dispersion). Even though the dis-
persion measured by CO chemisorption is over-estimated, these
results demonstrate that the Pd particle size decreases with lower
Pd loadings, since the apparent dispersion increases with decreas-
ing loading, as shown in Table 1. Metal particle sizes were not mea-
sured independently, because the objective of the chemisorption
study was to measure the relative change with metal loading in
the number of reducible sites.
The XRD pattern of calcined 0.25 wt% Pd/CeZrOx (the predomi-
nant catalyst studied in this work) shows broad reflections at
2h = 29.2°, 33.8°, 48.6° and 57.5°. This pattern (Fig. 3) is consistent
mainly with the CeO2 fluorite-cubic structure, as observed by other
investigators [27,33,35]. All peaks in the pattern displayed a small
shift to higher values of 2h, as compared to fluorite cubic CeO2
(JCPDS #43-1002), and this shift is associated with lattice contrac-
tion caused by the introduction of smaller Zr4+ ions in the CeO2 lat-
tice. No reflections corresponding to palladium or palladium oxide
phases were detected, indicating the presence of small (<4 nm)
particles.
Fig. 4. TPR profiles of CeZrOx and 0.25 wt% Pd/CeZrOx.
reduction of the metal, and of ceria species in intimate contact
with the metal [27]. This interpretation is supported by hydrogen
consumption measurements, which show that 10 times the
amount of hydrogen is consumed for the 480 K reduction peak,
than is required to reduce all the Pd2+ in the calcined catalyst to
metallic Pd. The broad peak observed at 720 K for 0.25 wt% Pd/CeZ-
rOx has been assigned to the reduction of surface ceria species that
are not in contact with the metal [27], and this reduction is shifted
to lower temperature (as compared to 860 K on the support alone)
due to spillover of hydrogen from the metal. In summary, charac-
terization results show that the Pd/CeZrOx catalysts used in this
work consist of a reducible mixed oxide of Ce and Zr, containing
both acidic and basic surface sites, and that the addition of Pd to
CeZrOx promotes the reduction of ceria surface species in contact
with Pd particles.
3.2. Reaction network
The conversion of 2-hexanone over 0.25 wt% Pd/CeZrOx was
studied to probe the pathways for aldol condensation and related
reactions catalyzed by metal, acidic and basic surface sites. The
conversion of 2-hexanone was subsequently carried out at identi-
cal conditions on CeZrOx to address the effects of removing the me-
tal sites for hydrogen dissociation. For brevity, the reaction
products are grouped by carbon number, and the product distribu-
tions obtained at space velocities of 0.48 and 1.92 hꢀ1 are summa-
rized in Table 2. The temperature dependence of the product
distributions resulting from 2-hexanone conversion over
0.25 wt% PdCeZrOx is shown in Table 3.
The TPR profiles of calcined 0.25 wt% Pd/CeZrOx and CeZrOx are
shown in Fig. 4. The profile of 0.25 wt% Pd/CeZrOx contains two
reduction peaks: a sharp peak at 480 K and a broad peak centred
at 720 K. The reduction profile of CeZrOx only displays a broad peak
at 860 K. The 860 K reduction peak in the TPR of CeZrOx has been
assigned to the reduction of surface Ce4+ [35]. The low temperature
reduction peak in the TPR of Pd/CeZrOx has been assigned to the

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
241
Table 2
Product distributions from the conversion of 2-hexanone over 0.25 wt% PdCeZrOx and CeZrOx, at 623 K and 5 bar total pressure.
Catalyst
WHSV (hꢀ1
)
Conversion (%)
Carbon selectivities
C12 (%)
C18 (%)
C11 (%)
C9 (%)
Alkanes (6C7) (%)
0.25 wt% Pd CeZrOx
0.25 wt% Pd CeZrOx
CeZrOx
1.92
0.48
1.92
0.48
58
77
21
33
83
69
66
46
2
3
25
31
2
3
–
–
5
11
7
8
13
2
CeZrOx
18
5
Table 3
Effect of reaction temperature on the conversion of 2-hexanone over 0.25 wt% Pd/CeZrOx at 5 bar and WHSV = 1.92 hꢀ1
.
T (K)
Conversion (%)
Carbon selectivities
C12 (%)
C18 (%)
C11 (%)
C9 (%)
Alkanes (6C7) (%)
573
598
623
42
49
60
96
93
83
1
1
2
1
1
2
1
2
5
1
2
8
The major organic liquid-phase product formed during the con-
version of 2-hexanone over 0.25 wt% Pd/CeZrOx was the self-aldol
condensation product, a C12 ketone (7-methyl-5-undecanone).
Additionally, C9 ketones (m/z = 142), C18 ketone (m/z = 268) and a
C11 ketone (5-undecanone) were detected. Liquid-phase alkane
products included hexane, with smaller amounts of 2-methyl-hex-
ane and 5-methyl-undecane. Lighter alkanes (C1–C6) were detected
in the gas phase, with methane, butane and hexane being the pre-
dominant gas phase species. As mentioned previously, 2-hexanone
was observed to be in equilibrium with 2-hexanol, and a 9:1 ke-
tone to alcohol ratio was measured experimentally at standard
reaction conditions (5.5:1 H2:2-hexanone ratio, 623 K, 5 bar). The
experimentally measured ketone/alcohol ratio was consistent with
the calculated ratio for the equilibrium conversion of acetone into
2-propanol at the aforementioned conditions.
The activity of CeZrOx towards 2-hexanone conversion was a
factor of 3 lower than that of 0.25 wt% Pd/CeZrOx at 1.92 hꢀ1 (Table
2). The major organic products formed in this conversion were
unsaturated C12 ketones (m/z = 182), C9 ketones, and a doubly
unsaturated C18 ketone (m/z = 264). Additionally, significant
amounts of C18 aromatic species (1,3,5-tributyl benzene and its
isomers) were detected. Conversion to light alkane species was less
than 5% at all reaction conditions.
The selectivities to C9 and C18 species for both catalysts in-
creased with decreasing space velocity, suggesting that these spe-
cies are secondary reaction products resulting from further
reactions of C12 ketones. Similarly, the selectivity towards light al-
kanes (on the Pd containing catalyst) increased by a factor of ꢂ2
with decreasing space velocity. The combined selectivities to C9
and C18 products also decreased with decreasing reaction temper-
ature, and decreased by a factor of ꢂ3 at 0.48 hꢀ1 when adding Pd
to CeZrOx to form the 0.25 wt% PdCeZrOx catalyst.
A reaction network is presented in Fig. 5 for Pd/CeZrOx, based on
the observed product distributions and results from studies of ace-
tone condensation [36,37]. Species marked with an asterisk are
reaction intermediates that were not detected in the reaction efflu-
ent. These intermediates are believed to react rapidly, leading to
undetectable steady-state concentrations. The primary reaction
pathway involves the self-coupling of 2-hexanone (0) to yield 7-
methyl-5-undecanone (3). In this pathway, 2-hexanone undergoes
self-aldol condensation on basic sites to form an aldol alcohol (1),
which then undergoes dehydration on acidic (or basic) sites to
reactions are fast in comparison to aldol condensation. 7-methyl-
5-undecanone undergoes further aldol condensation with 2-hexa-
none (in the pathway described above) to yield a C18 ketone (4).
Similar observations were reported by Di Cosimo for the self-cou-
pling of 2-propanol on Cu/MgAlOx [21].
Thermodynamic analysis shows that C–C coupling steps involv-
ing aldol condensation and dehydration are equilibrium limited
(such as (0) ! (2)). In contrast, the hydrogenation of
a–b unsatu-
rated ketones such as (2) is not equilibrium limited at these condi-
tions [5,23]. This hydrogenation reaction prevents reverse-aldol
reactions, and therefore drives C–C coupling reactions towards
higher conversions. Therefore, the absence of a hydrogenation
functionality on CeZrOx limits the yields of C12 condensation
products.
The equilibrated hydrogenation of the C@O group of 2-hexa-
none yields 2-hexanol, which can undergo dehydration over acidic
sites to yield hexene. Hexene is not observed among the reaction
products, because it is rapidly hydrogenated to form hexane. By
analogy, 7-methyl-5-undecanone undergoes dehydration and
hydrogenation steps to form 5-methyl undecane (7). Alcohols have
been shown to undergo reforming (cleavage of C–C bonds adjacent
to the –COH group) over metal catalysts to yield COx species and
alkanes. In this respect, 2-hexanol undergoes reforming to form
methane, butane and COx [38,39]. As we will show in subsequent
sections, COx species are hydrogenated to methane over 0.25 wt%
Pd/CeZrOx, and only small amounts of COx species (<1% of total car-
bon) are detected in the gas phase. Similarly, 7-methyl-5-undeca-
nol (5) undergoes reforming to yield methane and 2-methyl
hexane (6). These reforming and dehydration steps remove the
C@O functionality, and they are therefore detrimental to C–C cou-
pling processes.
The reaction temperature was varied to determine the apparent
activation energies of reactions forming the C12, C18 and C9 species
resulting from the conversion of 2-hexanone over 0.25 wt% Pd/
CeZrOx. The apparent activation energies were calculated from
Arrhenius plots in Fig. 6. The apparent activation energy for C12 for-
mation was found to be 15 kJ/mol, in agreement with the value ob-
tained by Al-Wadaani et al. [40] for the conversion of acetone to
MIBK over Pd/ZnCrOx, a similar reducible amphoteric oxide. The
apparent activation energy of C18 formation (28 kJ/mol) was com-
parable to that of C12 formation, suggesting that both species form
through similar bond-making processes (aldol condensation). In
contrast, the apparent activation energy for the formation of C9
species is an order of magnitude higher (140 kJ/mol). The C9 ke-
tones could not have been formed by aldol condensation, because
this process would yield oligomers with carbon numbers that are
form an
on metal sites to yield a saturated ketone (3). Neither the aldol
alcohol nor the –b unsaturated ketone was observed in the reac-
tion effluent, suggesting that the dehydration and hydrogenation
a–b unsaturated ketone (2), followed by hydrogenation
a

242
E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
C-O
Hydrogenation
α-Scission (reforming)
(0)
H2
Pd
H2
+2 CH4
Pd
OH
O
A
H2O
H2
Pd
B
O
*
Aldol
Condensation
(1)
OH
*
O
Michael
Addition
(9)
(8)
O
*
*
O
Dehydration
H2O
A
Double
bond shift
B
O
O
Retro-Michael
Reaction
A
B
(2)
*
O
*
+
O
C-C
O
Hydrogenation
(4)
O
H2
A/B/Pd
O
H2
Pd
Pd
H2
(10)
(3)
Dehydration/
hydroganation
O
H2
(7)
O
+H2O
H2
Pd
H2
(5)
(6)
A/ Pd
+
CH4
Pd
OH
Fig. 5. Proposed reaction scheme for the conversion of 2-hexanone on Pd/CeZrOX. The numbers in parentheses are referenced in the text. A, B and Pd refer to acidic, basic and
palladium sites, respectively. * represents reaction intermediates not observed in effluent.
Fig. 6. Arrhenius plots and calculated activation energies for the formation of C9 and C18 species (A) and C12 species (B) from the conversion of 2-hexanone on 0.25 wt% Pd/
CeZrOx.
multiples of six. The high apparent activation energy associated
with C9 formation suggests that a bond-breaking step decreases
the carbon number in the products. Furthermore, the absence of
significant amounts of C3 species suggests that the C9 products
are not formed via degradation of C12 ketones, and that the cleav-
age of C18 products may be involved.
One possible pathway for the formation of C9 ketones proceeds
through C@C double bond migration in the a–b unsaturated aldol
product (2) followed by Michael addition with 2-hexanone on ba-
sic catalytic sites to produce a C18 dione (9). Migration of the C@C
Mussolino et al. over Pd/TiO2 [41]. The conversion of 2-butanone
over CeZrOx in the present study showed the existence of double
bond migration, because a–b unsaturated condensation products
(C8 ketones) with the C@C double bond in different positions were
identified by GC/MS analysis. These products included 5-methyl-4-
hepten-3-one, 5-methyl-5-hepten-3-one and 5-methylene-3-hep-
tanone. The resulting unsaturated ketone (8) contains a less substi-
tuted double bond, which is more susceptible to Michael addition
with 2-hexanone to form a dione (9). The C18 dione then undergoes
retro-Michael reaction [42] to form the two different C9 ketones
(10) identified in the effluent. Finally, small amounts (ꢂ2% selectiv-
double bond in
a–b unsaturated ketones has been observed by

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
243
ity) of a C11 ketone (5-undecanone) may be obtained from C–C
hydrogenolysis of the methyl branch of 7-methyl-5-undecanone,
also yielding methane (not shown in the figure).
surface reaction, the reaction order in the numerator would be 2.
However, under the partial pressures used here, inhibition due to
surface coverage by reactant or product species reduces the appar-
ent reaction order, as observed by both previous investigators, and
it is possible that either first-order or second-order kinetics is ap-
proached in the limit of low 2-hexanone partial pressures. The
apparent reaction orders for 2-hexanone conversion observed in
the present study are similar to ones found by Torres et al. and
Melo et al. for acetone condensation reaction. [10,45] Torres
et al. calculated an apparent order in 2-propanol of 0.35 for the
MIBK formation rate for partial pressures of 2-propanol around
0.4 bar. Melo et al. found that the apparent order with respect to
acetone was 0.3. Also Melo et al. reported an apparent reaction or-
der in hydrogen of 1.3 for partial pressures up to 0.8 bar, which is
in accordance with the values found in the present study for a
higher partial pressure range. The effects of total reactor pressure
on the condensation activity and selectivity are shown in Table 4.
The conversion of 2-hexanone is increased significantly upon
increasing the total pressure, while the selectivity to primary and
secondary products remained approximately constant. This behav-
ior results from a combination of the effects of increasing H2 and 2-
hexanone partial pressures that were documented above.
Reactions involving a–b unsaturated ketones become more sig-
nificant over CeZrOx, because these species are not stabilized
against further reaction by hydrogenation. The selectivity towards
C9 and C18 secondary products is enhanced substantially compared
to that on 0.25 wt% PdCeZrOx. Furthermore, the a–b unsaturated
ketone (2) can undergo cross-aldol condensation with 2-hexanone
followed by a 1,6-aldol addition or aromatization to form 1,3,5-
tributylbenzene, a C18 species. This reaction is analgous to the for-
mation of mesitylene from acetone [36,37]. It is also important to
note that C@O hydrogenation (that produces alcohols), reforming,
and hydrogenolysis reactions are minimized in the absence of Pd.
3.3. Effects of partial and total pressures.
Reaction kinetic studies were conducted to determine the reac-
tion orders with respect to 2-hexanone and H2. These studies were
performed using an inert species (heptane for 2-hexanone, He for
H2) to change the partial pressures of 2-hexanone and hydrogen
independently, while maintaining the total pressure constant.
The space velocity of the reaction system was adjusted to achieve
conversions of <30%. At these conversions, secondary products
were produced in negligible yields. The logarithm of the reaction
3.4. Effect of metal loading and metal-support proximity
rate (
l
mol C12 minꢀ1 g catꢀ1) was plotted against the logarithm
The activity and selectivity data for the conversion of 2-hexa-
none over Pd/CeZrOx catalysts with different metal loadings are gi-
ven in Table 5. The selectivity towards saturated C12 product
decreases with increasing metal loadings. The decline in selectivity
towards the major condensation product is accompanied by an in-
crease in selectivity towards light alkanes. This result demon-
strates that higher metal loadings promote hydrogenolysis
reactions to form hexane and reforming reactions leading to the
formation of butane and methane. These reactions take place in
parallel with the condensation reaction pathway leading to the
of the partial pressure of the species, with the slopes yielding the
reaction orders, as shown in Fig. 7. For both hydrogen and 2-hexa-
none, the reaction orders were fractional and decreased with
increasing partial pressures. For 2-hexanone, the reaction order
was 0.9 for low partial pressures (0.08–0.16 bar), whereas it de-
creased to 0.5 as partial pressure was increased (0.31–0.63 bar).
For hydrogen, the order decreased from 0.8 to 0.6 going from low
to higher partial pressures (0.9–3.5 bar).
Several investigators have studied the aldol condensation of
acetone with citral on basic oxides. Diez et al. showed that the
C12 ketone. It is important to note that hydrogenation of the a–b
rate-determining step in the reaction is the abstraction of an
a
-
unsaturated ketone is complete for all metal loadings, as unsatu-
rated C12 species were not detected in the effluent. This observa-
tion shows that the hydrogenation step of the overall ketone
coupling process is a fast step in the reaction network. Therefore,
the rates of the side reactions, such as hydrogenolysis and reform-
ing, can be minimized by avoiding the use of excess metal
functionality.
hydrogen from acetone over a Li-doped MgO catalyst [43]. How-
ever, Abello et al. showed the slow step to be the surface reaction
in which the enolate ion generated from an acetone molecule at-
tacks a citral molecule over MgAlOx [44]. Either rate-determining
step is possible for 2-hexanone self-condensation. If it is abstrac-
tion of an a-hydrogen atom, then the reaction order with respect
to 2-hexanone in the numerator of a Langmuir–Hinshelwood type
rate expression should be 1, and if the rate-determining step is the
The importance of the proximity of metal sites to acid/base sites
was investigated by comparing the conversion of 2-hexanone over
Fig. 7. Rate versus partial pressure, and calculated reaction orders with respect to 2-hexanone pressure at PH2 = 4.2 bar (A) and with respect to hydrogen pressure at
P2ꢀhexanone = 0.8 bar (B).

244
E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
Table 4
Effect of total pressure on the conversion of 2-hexanone over 0.25 wt% Pd/CeZrOx at 623 K and WHSV = 1.92 hꢀ1
.
Ptotal (bar)
Conversion (%)
Carbon selectivities
C12 (%)
C18 (%)
C11 (%)
C9 (%)
Alkanes (6C7) (%)
5
16
25
58
75
84
83
80
79
2
3
5
2
2
2
5
4
4
8
10
9
Table 5
Product distributions from conversion of 2-hexanone over Pd/CeZrOx with different metal loadings, and a physical mixture of 0.5 wt% Pd/SiO2 and CeZrOx at 623 K and 5 bar total
pressure.
Catalyst
WHSV (hꢀ1
)
Conversion (%)
Carbon selectivities
C12 (%)
C18 (%)
C11 (%)
C9 (%)
Alkanes (6C7) (%)
0.05 wt% Pd/CeZrOx
0.25 wt% Pd/CeZrOx
1.45 wt% Pd/CeZrOx
0.5 wt% Pd/SiO2
1.92
1.92
1.92
2.84
1.92*
68
58
85
6
85
83
57
12
77
3
2
2
–
2
2
1
–
4
4
5
4
–
5
3
8
36
87
3
0.5 wt% Pd/SiO2 + CeZrOx
48
10
a physical mixture of Pd/SiO2 and CeZrOx to that over 0.25 wt% Pd/
CeZrOx, and the results are presented in Table 5. Due to the weak
acid/base properties of Si–OH groups on the surface, the SiO2 sup-
port can be considered to be inert [5]. In this respect, 0.5 wt% Pd/
SiO2 was tested alone, and a very low conversion of 2-hexanone
(6%) was observed, with a low selectivity towards the C12 ketone
(12%). A physical mixture of 0.5 wt% Pd/SiO2 with CeZrOx was then
prepared to give the same amount of acid, base and metal func-
tionality as 0.25 wt% Pd/CeZrOx, and this physical mixture was
studied for 2-hexanone condensation at the same reaction condi-
tions. A lower (ꢂ10%) activity was obtained in the case of the phys-
ical mixture. Even though the selectivity towards the C12 ketone
did not vary considerably from that on 0.25 wt% Pd/CeZrOx, the
selectivity towards higher molecular weight products, such as C18
aromatics was increased at the expense of light alkanes on the
physical mixture. This change in selectivity suggests that a me-
tal-support interaction is responsible for reforming reactions and
study, primary alcohols are readily equilibrated (through dehydro-
genation) with the corresponding aldehydes, in this case butanal
[46]. The aldehydes can undergo self-aldol condensation reactions
or cross-aldol condensation reactions with ketones, where the ke-
tones act as the nucleophilic species. A reaction scheme for the
conversion of primary alcohols is illustrated in Fig. 8A. Condensa-
tion of butanal with 2-hexanone yields 5-decanone, and this prod-
uct accounted for 26% of the product carbon. The self-condensation
product of butanal, 2-ethyl hexanal, was not observed due to the
low concentration of butanal. Butanal can also undergo decarbony-
lation reaction to produce CO and propane [47], which were both
detected in the effluent. A significant decrease in the activity of
2-hexanone self-condensation (70%) was caused by the addition
of the primary alcohol, which was almost completely converted.
This loss of activity was completely reversed upon a 12 h treat-
ment in flowing H2 and removal of the primary alcohol from the
feed. Possible causes for inhibition by primary alcohols include
strongly bound aldehyde species [33] or oligomers of butanal that
can act as coke precursors [48,49] and which can be removed by H2
treatment. Microcalorimetric studies have shown that primary
alcohols and aldehydes bind more strongly to metal and basic sites
compared to secondary alcohols and ketones [50,51]. Therefore, it
is likely that the catalytic sites required for coupling of ketones are
preferentially occupied by primary alcohols or aldehydes under
reaction conditions, leading to inhibition of ketone coupling. Low-
ering the WHSV to 0.48 hꢀ1 increased 2-hexanone conversion to
65% and the C12 selectivity to 60%, showing that the detrimental ef-
fect of primary alcohols on condensation activity can be overcome
if the inhibiting species are reacted to completion.
dehydration/hydrogenation reactions. In addition, because the a–
b unsaturated aldol condensation product cannot be hydrogenated
on the physical mixture unless it desorbs from the CeZrOx surface,
it has a higher probability to undergo further condensation reac-
tions (e.g., Michael or aldol) to produce the higher molecular
weight species using the pathways described earlier.
3.5. Effects of oxygenated additives and water
The influence of oxygenated compounds such as primary alco-
hols, secondary alcohols, 3-ketones, heterocyclics and carboxylic
acids on the aldol condensation of ketones was studied, because
these species represent classes of compounds (together with 2-ke-
tones) present in the hydrophobic organic mixture derived from
the conversion of sugars or polyols over a Pt–Re/C catalyst [4].
One representative species was selected from each group to facili-
tate the analysis, and the results are given in Table 6. An experi-
The effect of secondary alcohols on the self-condensation of 2-
hexanone was studied by feeding a mixture of 10 mol% 2-butanol
in 2-hexanone (entry 4). Secondary alcohols are dehydrogenated
to quasi-equilibrium with the corresponding ketones (2-butanone
in this case), the latter of which can undergo either self- or cross-
condensation, as illustrated in Fig. 8B The observed products are
the self-condensation product of 2-hexanone and the cross-con-
densation product of 2-butanone and 2-hexanone (3-methyl-5-
ment with 10 mol% inert species (heptane) serves as
a
benchmark for the additives study (entry 2), because although
heptane is expected to be inert, it may act as a diluent and there-
fore decrease the overall rate of a positive order reaction. Results
(entry 2) show that heptane has little effect on the activity and
selectivity of 2-hexanone conversion.
The effect of a primary alcohol on the self-condensation of 2-
hexanone was studied by feeding a mixture of 10 mol% 1-butanol
in 2-hexanone (entry 3). At the reaction conditions used in this
nonanone and/or 5-methyl-3-nonanone). If
a-hydrogen atoms of
the methyl group are abstracted from 2-butanone, and the result-
ing carbanion attacks the carbonyl group of 2-hexanone, then 3-
methyl-5-nonanone is formed. In the opposite case, 5-methly-3-
pentanone is formed. The 2-butanone self-condensation product
was not observed due to the low 2-butanone concentration, result-

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
245
Table 6
Effects of monofunctional oxygenates and water on the conversion of 2-hexanone and 2-butanone over 0. 25 wt% Pd/CeZrOx at 623 K, 5 bar and WHSV = 1.92 hꢀ1
.
Entry Feed (mol%)
Conversion
Primary product(s)
Secondary product(s)
Selectivity Species
Ketone Additive Species
Selectivity (%)
(%)
58
(%)
–
(%)
1
2
3
2-Hexanone
7-Methyl-5-undecanone (C12 ketone)
7-Methyl-5-undecanone (C12 ketone)
83
C18 ketones
C9 ketones
Alkanes (6C7)
C11 ketones
2
5
8
2
90% 2-Hexanone and 10% heptane
90% 2-Hexanone and 10% 1-butanol
56
29
0
75
C18 ketones
C9 ketones
Alkanes (6C7)
C11 ketones
1
6
14
4
100
7-Methyl-5-undecanone (C12 ketone)
5-Decanone (C10 ketone)
47
26
C18 ketones
C9 ketones
Alkanes (6C7)
C11 ketones
CO + CO2
1
4
17
3
1
4
5
90% 2-Hexanone and 10% 2-butanol
53
8
65
7-Methyl-5-undecanone (C12 ketone)
3-Methyl-5-nonanone and 5-methyl-3-nonanone 11
(C10 ketones)
77
C18 ketones
C9 ketones
2
4
Alkanes (6C7)
C11 ketones
4
2
90% 2-Hexanone and 10% butanoic acid
100
7-Methyl-5-undecanone (C12 ketone)
4-Heptanone (C7 ketone)
5-Decanone (C10 ketone)
35
28
2
C18 ketones
C9 ketones
Alkanes (6C7)
C11 ketones
CO + CO2
0
6
24
4
1
6
7
90% 2-Hexanone and 10% THF
75% 2-Hexanone and 25% 3-pentanone
2-Butanone
48
50
70
12
27
7-Methyl-5-undecanone (C12 ketone)
72
C18 ketones
C9 ketones
Alkanes (6C7)
C11 ketones
1
7
16
5
7-Methyl-5-undecanone (C12 ketone)
7-Ethyl-5-nonanone (C11 ketone)
73
9
C18 ketones
C9 ketones
Alkanes (6C7)
5-Undecanone
1
5
9
3
8
9
–
–
5-Methyl-3-heptanone
5-Methyl-3-heptanone
91
93
C6 ketones
C12 ketones
Alkanes (6C7)
2
5
2
Water-saturated 2-butanone (12 wt% water) 39
C6 ketones
C12 ketones
Alkanes (6C7)
1
3
3
ing in preferential cross-condensation. The overall ketone conver-
sion was 55%, showing that secondary alcohols do not significantly
affect the condensation activity.
ketones. The production of additional alkanes results from the
more facile reforming of the terminal functionalized carbon in
comparison to internal C–C cleavage.
The effect of carboxylic acids was investigated using a 10 mol%
butyric acid in 2-hexanone feed (entry 5). The butyric acid was
found to ketonize to 4-heptanone and also undergo a coupling
reaction with 2-hexanone to form 5-decanone. The 5-decanone re-
sults from the aldol condensation of 2-hexanone with butanal,
which is formed upon the partial reduction of butanoic acid. The
reactions involved in the conversion of butanoic acid are illustrated
in Fig. 8C. The addition of acid causes a significant loss in activity
for 2-hexanone self-condensation, such that the 2-hexanone con-
version decreased from 58% to 8%. However, complete deactivation
was not observed and the activity could be regained as described
for primary alcohols. Poisoning of the basic catalytic sites by car-
boxylic acids is the most probable cause for the loss of activity
[49,52]. Further poisoning of the basic sites may be caused by
the CO2 released from the ketonization reaction of butyric acid.
The effect of CO2 was investigated separately and will be explained
in the following sections. Decreasing the space velocity to 0.48 hꢀ1
increased the 2-hexanone conversion and C12 selectivity to 48%
and 56%, respectively. It should be noted that selectivities to light
alkanes are substantially higher in the presence of either primary
alcohols or carboxylic acids as compared to secondary alcohols or
A slight decrease of 2-hexanone self-condensation activity re-
sulted with the addition of 10 mol% THF (2-hexanone conversion
declined from 58% to 48%, entry 6). Product analysis revealed that
10% of the THF was converted, and the amount of propane detected
approximately matched the amount of THF converted. This result
suggests a ring opening reaction yielding a linear oxygenated inter-
mediate species, 1-butanol, which undergoes further decarbonyla-
tion to produce propane as observed by Kreuzer et al. [47].
Therefore, the decrease of activity can be attributed to partial con-
version of THF to 1-butanol, for which the deactivating effects of
primary alcohols were explained above.
The effect of 3-ketones was investigated (entry 7) via the reac-
tion of a 25 mol% mixture of 3-pentanone in 2-hexanone. The self-
condensation of 3-pentanone was not observed, and the cross-con-
densation of 3-pentanone with 2-hexanone was slower relative to
the self-condensation of 2-hexanone. The methyl group in 2-ke-
tones contains
more stable carbanion, than the secondary carbanions that form
upon abstraction of -hydrogen atoms from 3-ketones. Further-
a-hydrogen atoms that upon abstraction lead to a
a
more, the methyl carbanion is less sterically hindered than internal
carbanions. The aforementioned properties make the methyl

246
E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
OH
(A)
(B)
Pd
O
H2
O
O
H2
‡
+
A/B/Pd
O
H3C
*
Pd
H2
O
OH
O
O
O
H2
+
‡
H3C
A/B/Pd
*
O
O
O
H2
+
‡
H3C
A/B/Pd
*
OH
(C)
+ CO2 +H2O
A/B
O
O
O
H2
Pd
OH
O
O
O
H2
A/B/Pd
+
‡
O
H3C
*
(D)
O
O
O
H2
A/B/Pd
‡
H3C
+
*
Fig. 8. Reactions of selected oxygenates-primary alcohols (A), secondary alcohols (B), carboxylic acids (C) and 3-ketones (D). For aldol-type reactions, the source of the
carbanion is indicated by à, and the attacked carbonyl group is indicated by ꢃ.
carbanion a better nucleophile than secondary carbanions. Addi-
tionally, the carbonyl group of 3-ketones is more sterically hin-
dered than that of 2-ketones, making it less susceptible to
nucleophilic attack [53]. In this respect, the self-condensation of
2-hexanone is favoured over cross-condensation with 3-penta-
none, which is favoured substantially over the self-condensation
of 3-pentanone. Accordingly, a single cross-condensation product
(where 2-hexanone acts as the nucleophile) would be expected,
as illustrated in Fig. 8D. These results are observed experimentally.
In addition to containing alkanes and oxygenated hydrocarbon
species, organic streams derived from the conversion of sugars
over Pt–Re/C are saturated with water, resulting from contact with
an aqueous phase. Water is also formed as a by-product of aldol
condensation, and associated secondary hydrogenolysis reactions.
Furthermore, water is produced during ketonization. Accordingly,
it is important to study the effects of water on the self-coupling
of ketones, and in this respect we have investigated the effects of
water on the self-aldol condensation of 2-butanone into 5-
methyl-2-heptanone (Entries 8 and 9). In these studies, 2-buta-
none was chosen as the reactant, because it can be saturated with
a larger amount of water (ꢂ12 wt%) than 2-hexanone, and this
water content is representative of the products of ketonization.
The addition of water decreased the condensation activity by
40%, and this activity loss was reversible, suggesting inhibition.
The inhibition of the aldol condensation activity by water may
be responsible for the fractional H2 and 2-hexanone reaction orders
described previously. In this respect, Gammara et al. report that
the presence of relatively large amounts of molecularly adsorbed
water on ceria is responsible for inhibition of CO oxidation on
Cu/CeOx catalysts [54]. Moreover, Gutierrez-Ortiz et al. showed
by H2O–TPD experiments that the hydrophilicity of ceria is in-
creased by addition of ZrO2 [55]. In addition to inhibition by site
blocking, the presence of water may also shift the aldol condensa-
tion equilibrium to the reactants side by inhibiting dehydration of
the aldol-alcohol product.
3.6. Effect of CO2 co-feeding
Carbon dioxide is produced as a by-product of the ketonization
of carboxylic acids. Therefore, it is necessary to understand the ef-
fects of the by-product CO2 on the self-coupling of ketones to
accomplish the integration of ketonization and aldol condensa-
tion/hydrogenation in a single reactor system. To that end, we have
substituted the pure hydrogen gas co-feed used above with H2–CO2

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
247
Table 7
Effects of CO2 on the conversion of 2-hexanone on CeZrOx and Pd/CeZrOx at 623 K and 5 bar total pressure.
Catalyst
% CO2 in H2
WHSV (hꢀ1
)
Conversion
Ketone (%)
Carbon selectivities
CO2 (%)
C12 (%)
C18 (%)
C11 (%)
C9 (%)
Alkanes (C2–C7) (%)
0.25 wt% Pd CeZrOx
0.25 wt% Pd CeZrOx
0.25 wt% Pd CeZrOx
CeZrOx
10
10
5
1.92
0.48
1.92
1.92
5
17
7
63
45
44
0
22
12
26
45
0
–
0
4
5
4
7
–
18
24
26
46
55
60
41
5
10
7
gas mixtures. The results of CO2 co-feeding at various conditions are
shown in Table 7. Co-feeding a 10 mol% CO2 in H2 stream resulted in
almost complete loss of self-condensation activity on 0.25 wt% Pd/
CeZrOx, decreasing the conversion of 2-hexanone from 60% to 5%.
Under these conditions, 63% of the CO2 was converted to CO by
the reverse water–gas shift reaction and to CH4 by methanation,
and 2-hexanone was observed to be in equilibrium with 2-hexanol.
Carbon monoxide adsorbs more strongly than hydrogen on the Pd
surface, and CO poisoning has been reported in hydrogenation reac-
tions [56,57]. However, the observation of quasi-equilibration of 2-
hexanone with 2-hexanol in the presence of CO (derived from
reduction of CO2) suggests that hydrogenation is still fast in the
presence of CO, possibly because of the high reaction temperatures.
A decrease in the concentration of CO2 in the co-feed to 5% did
not improve the self-coupling yield. Similarly, decreasing the space
velocity of the ketone feed was not effective in overcoming the ef-
fects of CO2. The self-coupling of 2-hexanone was also significantly
suppressed on CeZrOx, suggesting that CO2 influences aldol con-
densation sites on the support. For all conditions listed in Table
7, the activity loss associated with CO2 was observed to be com-
pletely reversible upon removal of CO2 from the gas co-feed
stream. Kamimura et al. showed that co-feeding of CO2 suppresses
the aldol condensation activity of CeFeOx catalysts, and they attri-
bute this effect to the blocking of basic catalytic sites by acidic CO2
[58]. The suppression of basic catalytic sites by CO2 causes the
product selectivity to be substantially altered by reactions taking
place on other functionalities such as metal and acidic sites, there-
by leading to higher selectivities to alkanes for all conditions in Ta-
ble 7. The alkanes are formed by dehydration/hydrogenation, and
reforming of 2-hexanol.
3.7. Interaction of CO2 with Pd/CeZrOx and CeZrOx
Fig. 9. TPSR of adsorbed CO2 under H2 for 0.25 wt% Pd/CeZrOx (A) and CeZrOx (B).
The carbon dioxide desorption profiles described in Section 3.1
show that CO2 desorption from Pd/CeZrOx persists at temperatures
up to 1000 K. It is therefore, interesting to note that the loss of
activity resulting from CO2 co-feeding was completely reversible
on 0.25 wt% Pd/CeZrOx and CeZrOx at a temperature of 623 K, be-
cause a fraction of the adsorbed CO2 may still be blocking active
basic sites, even after desorption at this reaction temperature.
Due to the setup of our reactor system we were not able to obtain
a transient reaction profile needed to establish the relationship be-
tween the rate of CO2 desorption at 623 K and the return of aldol
condensation activity; however, activity was returned after 6–
12 h treatment in H2 at reaction temperature.
Strongly binding sites may become irreversibly covered by
tightly bound spectator species that prevent catalytic turnover.
Since these sites do not participate in the overall reaction, their
deactivation by CO2 will not be reflected in catalytic activity mea-
surements. However, the catalyst regeneration studies carried out
in the present work employed flowing H2, whereas CO2-TPD stud-
ies were carried out using an inert carrier gas. Therefore, studies
employing TPSR of CO2 under flowing H2 were carried out. Fig. 9
shows the desorption profiles of CO2 and associated products (CO
and methane) for 0.25 wt% Pd/CeZrOx and CeZrOx. On 0.25 wt%
Pd/CeZrOx, the desorption of CO2 was complete by 620 K, and
was accompanied by the production of CO and methane, which
peaked at 550 and 650 K, respectively. Palladium is a poor catalyst
for the methanation of CO2 and CO [59]. However, the presence of a
reducible support such as CeO2 has been shown to substantially
promote the activity of Pd for CO2 activation. Leitenberg et al. pro-
pose a mechanism in which Ce4+ is reduced to Ce3+ by hydrogen,
which is activated via dissociation over a metal, followed by spill-
over onto the support [60]. The resulting Ce3+ reacts with CO2 to
form CO and regenerates Ce4+. The formation of Ce3+ is required
for the activation of CO2. In agreement with this behavior, we ob-
serve that the onset temperature of CO formation (450 K) in TPSR
corresponds to the reduction temperature (obtained from TPR) of
Ce4+ in intimate contact with Pd. At higher temperatures
(>600 K), CO is hydrogenated to methane. Accordingly, the activa-
tion of CO2 by a hydrogen-reduced ceria permits the removal of
CO2 from the surface at substantially lower temperatures than in
the absence of hydrogen.

248
E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
Fig. 9B shows that even in the absence of metal, desorption of
Wisconsin. We also thank Professor Manos Mavrikakis for valuable
discussions and collaborations throughout this project.
CO2 from CeZrOx is facilitated by the presence of hydrogen, with
CO2 desorbing completely at 850 K. This observation may explain
the fact that even CeZrOx regains catalytic activity after regenera-
tion with hydrogen, and suggests that hydrogen interacts with
the surface in the absence of a metal. In this respect, Kondo et al.
have observed the dissociation of hydrogen on ZrO2 to form Zr–
OH surface species at temperatures above 373 K[61]. These species
may drive the conversion of surface carbonates into more labile
species, thereby facilitating the desorption of CO2. Furthermore,
the desorption of CO2 above ꢂ700 K is accompanied by the produc-
tion of CO, which subsides at a temperature near 900 K. As with the
presence of metal, the production of CO appears to be associated
with the reduction of surface CeO2 species observed in TPR.
References
[1] Annual Energy Outlook 2009, Energy Information Administration of the US
Department of Energy, March 2009.
[2] J.A. Kinast, Production of Biodiesels from Multiple Feedstocks and Properties of
Biodiesels and Biodiesel/Diesel Blends, National Renewable Energy Laboratory,
2003.
[3] B. West, K. Knoll, W. Clark, R. Graves, J. Orban, S. Przesmitzki, T. Theiss, Effects
of Intermediate Ethanol Blends on Legacy Vehicles and Small Non-Road
Engines, National Renewable Energy Laboratory, 2008.
[4] E.L. Kunkes, D.A. Simonetti, R.M. West, J.C. Serrano-Ruiz, C.A. Gärtner, J.A.
Dumesic, Science 322 (2008) 417.
[5] A.A. Nikolopoulos, B.W.L. Jang, J.J. Spivey, Appl. Catal. A 296 (2005) 128.
[6] O. Nagashima, S. Sato, R. Takahashi, T. Sodesawa, J. Mol. Catal. A 227 (2005)
231.
[7] R.M. West, Z.Y. Liu, M. Peter, J.A. Dumesic, ChemSusChem 1 (2008) 417.
[8] K.M. Dooley, A.K. Bhat, C.P. Plaisance, A.D. Roy, Appl. Catal. A 320 (2007) 122.
[9] M. Glinski, J. Kijenski, A. Jakubowski, Appl. Catal. A 128 (1995) 209.
[10] G. Torres, C.R. Apesteguia, J.I. Di Cosimo, Appl. Catal. A 317 (2007) 161.
[11] K.H. Lin, A.N. Ko, Appl. Catal. A 147 (1996) 1259.
[12] R.D. Hetterley, R. Mackey, J.T.A. Jones, Y.Z. Khimyak, A.M. Fogg, I.V.
Kozhevnikov, J. Catal. 258 (2008) 250.
[13] M.J. Climent, A. Corma, S. Iborra, A. Velty, J. Mol. Catal. A Chem. 182–183
(2002) 327.
[14] J.I. DiCosimo, C.R. Apesteguia, M.J.L. Gines, E. Iglesia, J. Catal. 190 (2000) 261.
[15] J.I. DiCosimo, V.K. Diez, M. Xu, E. Iglesia, C.R. Apesteguia, J. Catal. 178 (1998)
499.
[16] V.K. Diez, C.R. Apesteguia, J.I. DiCosimo, Stud. Surf. Sci. Catal. 139 (2001) 303.
[17] Y.Z. Chen, C.M. Hwang, C.W. Liaw, Appl. Catal. A 169 (1998) 207.
[18] Y.Z. Chen, B.J. Liaw, H.R. Tan, K.L. Shen, Appl. Catal. A 205 (2001) 61.
[19] A.C.C. Rodrigues, J.L.F. Monteiro, Stud. Surf. Sci. Catal. 154 (2004) 2395.
[20] G. Waters, O. Richter, B. Kraushaar-Czarnetzki, Ind. Eng. Chem. Res. 45 (2006)
5701.
[21] J.I. DiCosimo, G. Torres, C.R. Apesteguia, J. Catal. 208 (2002) 114.
[22] S.G. Powell, A.T. Nielsen, J. Am. Chem. Soc. (1948) 3627.
[23] R.M. West, E.L. Kunkes, D.A. Simonetti, J.A. Dumesic, Catal. Today 5 (2008).
[24] S.M. de Lima, I.O. da Cruz, G. Jacobs, B.H. Davis, L.V. Mattos, F.B. Noronha, J.
Catal. 257 (2008) 356.
4. Conclusions
The self-coupling of 2-hexanone to form C12 species via aldol
condensation/hydrogenation was studied over Pd/CeZrOx catalysts.
XRD studies show that the CeZrOx support consists of a cerium–zir-
conium solid solution. TPR studies of these catalysts indicate that
the presence of palladium causes reduction of neighbouring ceria
species. Ammonia and CO2 TPD profiles indicate the presence of
both acidic and basic functionalities. TPSR studies in flowing H2
for catalysts containing adsorbed CO2 indicate that CO2 is con-
verted to CO over Ce3+ sites that are formed during reduction of
Ce4+ to Ce3+. Reaction kinetics studies show that the primary prod-
uct of aldol condensation/hydrogenation is C12 ketone, with the
formation of C9 and C18 ketones as secondary products, as well as
the production of light alkanes (C1–C7). The high activation energy
of C9 formation (140 kJ/mol) compared to the formation of C12 and
C18 species (15 and 28 kJ/mol, respectively), and the increase in the
selectivity towards C9 species in the absence of Pd, indicate that
these species may be formed as a result of the decomposition of
unsaturated C18 condensation products. The light alkane species
are formed as a result of dehydration/hydrogenation and reforming
of 2-hexanol. The self-coupling of 2-hexanone was found to be po-
sitive order in both 2-hexanone and hydrogen, and consequently
the yield to C12 products was found to increase with total reaction
pressure. Lower metal loadings resulted in enhanced selectivity to
C12 products and suppressed de-oxygenation side reactions. Metal-
support proximity was not found to substantially influence catalyst
performance, although it had a measureable effect on the forma-
tion of secondary products.
Monofunctional oxygenated additives such as primary alcohols
and carboxylic acids were found to reversibly inhibit the self-cou-
pling activity of 2-hexanone. The detrimental effects of these spe-
cies could be overcome by increasing the overall reaction residence
time. Water (the product of both aldol condensation and ketoniza-
tion) and carbon dioxide (a product of ketonization) were also ob-
served to reversibly inhibit 2-hexanone self-coupling. These
observations lead us to conclude that it may be difficult to inte-
grate efficiently the ketonization of carboxylic acids and the aldol
condensation/hydrogenation of ketones in a single reactor system.
However, because water and CO2 can easily be separated from the
organic reaction mixture, the operation of a cascade system con-
sisting of separate ketonization and aldol condensation reactors
appears to be preferred.
[25] F. Romero-Sarria, J.C. Vargas, A.C. Roger, A. Kiennemann, Catal. Today 133
(2008) 149.
[26] V. Solinos, E. Rombi, I. Ferino, M.G. Cutrufello, G. Colon, J.A. Navio, J. Mol. Catal.
A Chem. 204-205 (2003) 629.
[27] J.C. Serrano-Ruiz, J. Luettich, A. Sepulveda-Escribano, F.R. Reinoso, J. Catal. 241
(2006) 45.
[28] B.E. Spiewak, J. Shen, J.A. Dumesic, J. Phys. Chem. 99 (1995) 17640.
[29] H.O. Zhu, J.R. Kim, S.K. Ihm, Appl. Catal. B 86 (2009) 87.
[30] K.A. Pokrovski, A.T. Bell, J. Catal. 241 (2006) 276.
[31] J. Wambach, A. Baiker, A. Wokaun, Phys. Chem. Chem. Phys. 1 (1999) 5071.
[32] D. Bianchi, T. Chafik, M. Khalfallah, S.J. Teichner, Appl. Catal. A 112 (1994) 219.
[33] S.M. DeLima, A.M. Silva, U.M. Graham, G. Jacobs, B.H. Davis, L.V. Mattos, F.B.
Noronha, Appl. Catal. A 352 (2009) 95.
[34] Y.T. Kim, E.D. Park, H.C. Lee, D. Lee, K.H. Lee, Appl. Catal. B (2009).
[35] K. Otsuka, Y. Wang, M. Nakamura, Appl. Catal. A 183 (1999) 317.
[36] S. Lippert, W. Baumann, K. Thomke, J. Mol. Catal. 69 (1991) 199.
[37] G.S. Salvapati, K.V. Ramanamurty, M. Janardanarao, J. Mol. Catal. 54 (1989) 9.
[38] R. Alcala, M. Mavrikakis, J.A. Dumesic, J. Catal. 218 (2003) 178.
[39] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964.
[40] F. Al-Wadaani, E.F. Kozhevnikova, I.V. Kozhevnikov, J. Catal. 257 (2008) 199.
[41] M.G. Musolino, P. Maio, A. Donato, R. Pietropaolo, J. Mol. Catal. A Chem. 208
(2004) 219.
[42] J. Wang, T. Li, Collect. Czech. Chem. Commun. 64 (1998) 107.
[43] V.K. Diez, C.R. Apesteguia, J.I. DiCosimo, J. Catal. 240 (2006) 235.
[44] S. Abello, S. Dhir, G. Colet, J. Perez-Ramirez, Appl. Catal. A 325 (2007) 121.
[45] L. Melo, F. Chevalier, P. Magnoux, M. Guisnet, in: Proceedings of the XIII
Iberoamerican Symposium on Catalysis, Segovia, Spain, 1992.
[46] A.S. Ndou, N.J. Coville, Appl. Catal. A 275 (2004) 103.
[47] K. Kreuzer, R. Kramer, J. Catal. 167 (1997) 391.
[48] H. Tsuji, F. Yagi, H. Hattori, H. Kita, J. Catal. 148 (1994) 759.
[49] J. Weitkamp, M. Hunger, U. Rymsa, Micropor. Mesopor. Mater. 48 (2001) 255.
[50] M.A. Natal-Santiago, J.M. Hill, J.A. Dumesic, J. Mol. Catal. A Chem. 140 (1999)
199.
[51] M.A. Natal-Santiago, M.A. Sanchez-Castillo, R.D. Cortright, J.A. Dumesic, J.
Catal. 193 (2000) 16.
[52] J. Lopez, J. Valente, J.-M.C. Sanchez, F. Figueras, J. Catal. 208 (2002) 30.
[53] J. McMurry, Organic Chemistry, fifth ed., Brooks/Cole Pub. Co., Pacific Grove,
2000.
Acknowledgments
[54] D. Gamarra, A. Martinez-Arias, J. Catal. 263 (2009) 189.
[55] J.I. Gutierrez-Ortiz, B. Rivas, R. Lopez-Fonseca, J.R. Gonzalez-Velasco, J. Therm.
Anal. Calorim. 80 (2005) 225.
This work was supported by the US Department of Energy Of-
fice of Basic Energy Sciences, and the Great Lakes Bioenergy Re-
search Centre (GLBRC) funded by DOE at the University of
[56] L. Cider, N.H. Schöön, Appl. Catal. 68 (1991) 191.

E.L. Kunkes et al. / Journal of Catalysis 266 (2009) 236–249
[60] C.D. Leitenburg, A. Trovarelli, J. Kaspar, J. Catal. 166 (1997) 98.
249
[57] M.S. Li, J.Y. Shen, Mater. Chem. Phys. 68 (2001) 204.
[58] Y. Kamimura, S. Sato, R. Takahashi, T. Sodesawa, T. Akashi, Appl. Catal. A 252
[61] J. Kondo, H. Abe, Y. Skata, K. Maruya, K. Domen, T. Onishi, J. Chem. Soc. Faraday
Trans. 1 (84) (1988) 511.
(2003) 399.
[59] P. Panagiotopoulou, D.I. Kondarides, X.E. Verykios, Appl. Catal. A 344 (2008)
45.