Hydrodeoxygenation of Guaiacol over Ceria-Zirconia Catalysts

Article abstract of DOI:10.1002/cssc.201500317

The hydrodeoxygenation of guaiacol is investigated over bulk ceria and ceria-zirconia catalysts with different elemental compositions. The reactions are performed in a flow reactor at 1 atm and 275-400 °C. The primary products are phenol and catechol, whereas cresol and benzene are formed as secondary products. No products with hydrogenated rings are formed. The highest conversion of guaiacol is achieved over a catalyst containing 60 mol % CeO₂ and 40 mol % ZrO₂. Pseudo-first-order activation energies of 97-114 kJ mol^-1 are observed over the mixed metal oxide catalysts. None of the catalysts show significant deactivation during 72 h on stream. The important physicochemical properties of the catalysts are characterized by X-ray diffraction (XRD), temperature-programmed reduction, titration of oxygen vacancies, and temperature-programmed desorption of ammonia. On the basis of these experimental results, the reasons for the observed reactivity trends are identified.

Full text of DOI:10.1002/cssc.201500317

DOI: 10.1002/cssc.201500317
Full Papers
Hydrodeoxygenation of Guaiacol over Ceria–Zirconia
Catalysts
Sarah M. Schimming,[a, b] Onaje D. LaMont,[a, c] Michael Kçnig,[a, d] Allyson K. Rogers,[a, e]
Andrew D. D’Amico,[c] Matthew M. Yung,[e] and Carsten Sievers*[a, b]
The hydrodeoxygenation of guaiacol is investigated over bulk
ceria and ceria–zirconia catalysts with different elemental com-
positions. The reactions are performed in a flow reactor at
1 atm and 275–4008C. The primary products are phenol and
catechol, whereas cresol and benzene are formed as secondary
products. No products with hydrogenated rings are formed.
The highest conversion of guaiacol is achieved over a catalyst
containing 60 mol% CeO2 and 40 mol% ZrO2. Pseudo-first-
order activation energies of 97–114 kJmolÀ1 are observed over
the mixed metal oxide catalysts. None of the catalysts show
significant deactivation during 72 h on stream. The important
physicochemical properties of the catalysts are characterized
by X-ray diffraction (XRD), temperature-programmed reduction,
titration of oxygen vacancies, and temperature-programmed
desorption of ammonia. On the basis of these experimental re-
sults, the reasons for the observed reactivity trends are identi-
fied.
Introduction
Our society relies heavily on energy and chemicals for our
daily lives. In most countries, transportation fuels are produced
primarily from nonrenewable petroleum.[1–3] As oil reserves are
finite, it is important to find alternative feedstocks for the pro-
duction of transportation fuels and chemicals.[3] Fuels pro-
duced from alternative feedstocks should be compatible with
existing infrastructure.[4] Consequently, the fuels will need to
be in liquid form and mimic the properties of petroleum-based
fuels. The feedstocks for fuels and chemicals must be sustaina-
ble, allow for conversion with limited CO2 emissions, and be as
environmentally benign as possible. Moreover, the use of a spe-
cific feedstock should not create a competition between the
production of food and fuels or chemicals.[5,6] Many types of
biomass can meet these requirements.[3,7]
Biomass can be processed in many ways, including gasifica-
tion, pyrolysis, and hydrolysis.[8] Among these options, pyrolysis
has the advantage that it can convert most carbonaceous ma-
terials including wood, grasses, agricultural waste, sewage
sludge, and chicken litter to bio-oils that are similar to certain
petroleum-based process streams.[4,9] Specifically, pyrolysis pro-
duces bio-oils through the heating of biomass in the absence
of oxygen to temperatures of 375–5258C under pressures of
1–5 atm. Flash pyrolysis, the most economically feasible pyroly-
sis at an industrial level, occurs under the above conditions
with a residence time of only a few seconds.[10,11] The initial
products are vapors, which then condense to a liquid product.
These bio-oils can be obtained in high yields of up to 80 wt%
based on dry feed and typically contain 10–40 wt% oxygen
and 10–30 wt% water depending on the feed.[6,8,9]
The properties of bio-oils from flash pyrolysis differ from
those of petroleum-based oils in terms of their low heating
values, high polarities, high acidities, and high viscosities.[8,10–13]
Most of these unfavorable characteristics are associated with
oxygen-containing compounds in pyrolysis oils. Owing to the
presence of these oxygenates, bio-oils are generally much
more acidic (pH 2–4) than petroleum-based oil, and this results
in challenges for storage, transport, and processing. Specifical-
ly, traditional carbon steel and aluminum containers cannot be
used because of the high corrosiveness of bio-oils. Moreover,
many oxygenates can polymerize in air, which leads to increas-
es in the average molecular weight, viscosity, and water con-
tent as well as a decrease of the oil yield.[4,14] However, if the
oxygen content is reduced, it is easier to separate the organic
product from the water, the heating value increases, and the
oil becomes more stable.
[a] S. M. Schimming, O. D. LaMont, M. Kçnig, A. K. Rogers, Prof. C. Sievers
School of Chemical & Biomolecular Engineering
Georgia Institute of Technology
311 Ferst Dr., NW, Atlanta, GA, 30332 (USA)
E-mail: carsten.sievers@chbe.gatech.edu
[b] S. M. Schimming, Prof. C. Sievers
Renewable Bioproducts Institute
Georgia Institute of Technology
Atlanta, GA, 30332 (USA)
[c] O. D. LaMont, A. D. D’Amico
Micromeritics Instrument Corporation
Norcross, GA, 30093 (USA)
[d] M. Kçnig
Technische Universität München
Catalysis Research Center
85748 Garching (Germany)
[e] A. K. Rogers, M. M. Yung
National Renewable Energy Laboratory
Golden, CO 80401 (USA)
An evolving technology for oxygen removal from bio-oils is
hydrodeoxygenation (HDO).[14] In principle, HDO can be per-
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201500317.
ChemSusChem 2015, 8, 2073 – 2083
2073
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
formed for a wide variety of feedstocks, and its flexibility ena-
bles the processing of pyrolysis oils with different composi-
tions. HDO requires a hydrogen source and results in the pro-
duction of water as a byproduct. Some compounds have a low
HDO reactivity and require a higher hydrogen pressure and
temperature for complete oxygen removal.
highly efficient HDO catalysts must be selective towards de-
oxygenated products and minimize hydrogen consumption
and coking.[35]
In this study, the HDO activity of ceria–zirconia catalysts is
demonstrated and correlated with the physicochemical proper-
ties of different ceria and ceria–zirconia samples. The effects of
the temperature and the reactant flow rates on the conversion
and product distribution are elucidated. The influence of the
composition of the mixed oxide catalysts on the density of
oxygen vacancies is discussed. The ceria–zirconia catalysts are
selective towards HDO products at atmospheric pressure, mini-
mize hydrogen consumption, and maximize carbon retention
in the product.
In many cases, supported noble and transition metal cata-
lysts that were developed for hydrodesulfurization (HDS) are
used for HDO.[14] Although supported noble metal catalysts
show considerable activity, their high cost presents a significant
challenge for economic viability. Other commercial HDS cata-
lysts are comprised of supported MoS2 with promoters such as
Co2+ or Ni2+ cations. The Co2+ or Ni2+ cations donate electrons
to the Mo atoms. Thus, the bonds between the Mo atom and
adjacent S atoms in the catalyst become weaker, and a sulfur
vacancy can be formed. If a sulfur atom in a feed molecule in-
teracts with a vacancy, the CÀS bond can be broken, and the
sulfur content of the product decreases.[15–17] Hydrogen is nec-
essary to remove the sulfur from the surface of the catalyst as
H2S and regenerate the vacancy.[17–20] The sulfur vacancies in
CoMoS2 and NiMoS2 can also bind oxygen and nitrogen and
remove these heteroatoms from
Results
Characterization
Composition and Morphology
The characterization results for all of the catalysts are summar-
ized in Table 1. The labels for the different catalysts were
organic reactants.[14,21,22] Unfortu-
Table 1. Physicochemical properties of ceria and ceria–zirconia catalysts.
nately, MoS2-based catalysts de-
activate over time during HDO
reactions because they are con-
verted to an inactive oxide
phase.[23–25] The active sulfide
phase can be retained if H2S is
co-fed with the hydrogen. Al-
though the co-feeding of H2S
can provide the desired activity,
it may result in the contamina-
tion of bio-oils from sources
with little or no sulfur content.[26]
Composition of Label BETsurface
Average pore Pore volume Crystallite size Acid site concentration
[c]
[e]
mixed oxide[a]
area [m2 gÀ1
]
size [nm][b]
[cm3 gÀ1
]
[nm][d]
[mmolgÀ1
]
CeO2
Ce100 63
9.5
5.0
2.7
2.6
0.144
0.110
0.059
0.043
9.3
4.3
3.2
5.3
16
24
25
32
Ce0.82Zr0.18O2
Ce0.60Zr0.40O2
Ce0.46Zr0.54O2
Ce82
Ce60
Ce46
85
93
79
[a] Composition of mixed oxide determined by PIXE analysis. [b] Calculated by BJH adsorption method. [c] BJH
cumulative pore volume. [d] Calculated from XRD. [e] Determined by TPD of ammonia.
Such contaminants can poison noble metal catalysts in down-
stream processes. Moreover, safety concerns associated with
H2S make its use undesirable.
chosen on the basis of their elemental compositions, which
were determined by proton-induced X-ray emission (PIXE)
analysis. In addition to cerium, zirconium, and oxygen, only
a small hafnium impurity (less than 0.3 mol%) was detected.[36]
All of the mixed oxides had surface areas that were higher
than that of pure ceria but lower than that of pure zirconia.[34]
Among the mixed oxide samples, the highest surface area of
93 m2 gÀ1 was observed for Ce60. The external surface areas of
all samples were within 2% of the BET surface areas; therefore,
the samples had negligible microporosity (Table S1). All of the
ceria–zirconia catalysts had a smaller average pore width and
a smaller pore volume than those of pure ceria. The X-ray dif-
fractograms of the mixed oxides were dominated by the same
four prominent peaks as those in the diffractogram of pure
ceria (Figure S1).[37–39] The crystallite sizes calculated by the
Scherrer equation were between 3.2 and 5.3 nm and reached
a minimum for Ce60 (Table 1). All of the ceria–zirconia samples
had smaller crystallite sizes than those of pure ceria or pure zir-
conia. A more detailed description of the physiochemical prop-
erties of the ceria–zirconia catalysts was reported elsewhere.[34]
The concentration of the Lewis acid sites increased with in-
creasing zirconia content of the catalyst (Table 1). None of the
This study aims to develop ceria–zirconia-based mixed metal
oxide catalysts for the HDO of pyrolysis oil with similar reactivi-
ties to those of sulfided NiMo and CoMo catalysts. An impor-
tant requirement for HDO activity over a solid oxide catalyst is
the availability of suitable sites for the activation of the CÀO
bonds in organic oxygenates. Much like that over sulfided
NiMo catalysts in HDS, the adsorption on the surface should
occur primarily at vacancies. One way to provide vacancies is
to use transition metal compounds with redox activity such as
ceria, titania, or vanadia.[27–29] Ceria transitions readily from the
3+ to the 4+ oxidation state and can alternate between these
states depending on the environmental conditions.[30,31]
Doping ceria with zirconia lowers the temperature of surface
and bulk reduction and also increases the number of oxygen
vacancies.[29,32] A second requirement for HDO catalysts is the
ability to dissociate hydrogen, which is needed to break the
CÀO bonds and to remove the oxygen from the catalyst sur-
face as water.[33] Ceria–zirconia dissociates hydrogen and cre-
ates oxygen vacancy sites under mild conditions.[34] Lastly,
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2074
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
samples contained any detectible Brønsted acid sites (Fig-
ure S2).[34]
Table 2. Concentration of oxygen vacancies on ceria and ceria–zirconia
at 200, 300, and 4508C.
Mixed
oxide
Concentration of oxygen vacancies [mmolgÀ1
]
Temperature-programmed reduction
2008C
3008C
4508C
Ce100
Ce82
Ce60
Ce46
0
0
0
0
80
146
144
92
276
596
780
656
The redox properties of the ceria-based catalysts were ana-
lyzed by temperature-programmed reduction (TPR) in
10 mol% hydrogen in argon (Figure 1). In particular, the forma-
drogen was consumed by the samples below 2008C. This ob-
servation illustrates that hydrogen was adsorbed on the sur-
face before the reduction of the surface began. It also indicates
that the adsorbed hydrogen did not contribute to the con-
sumption of O2 during the titration. At 3008C, a notable
number of oxygen vacancies were formed for ceria and all
ceria–zirconia samples. The concentration of oxygen vacancies
per mass of catalyst was the highest for Ce82 and Ce60. At
4508C, a significantly larger number of oxygen vacancies were
formed for all samples (Table 2). The concentrations of oxygen
vacancies in ceria–zirconia catalysts were at least two times
that of ceria. The highest concentration of vacancies was
found for Ce60.
Figure 1. TPR profiles of ceria and ceria–zirconia catalysts in 10 mol% hydro-
gen in argon.
Reactivity for the HDO of guaiacol
Stability of the catalysts
tion of oxygen vacancies by surface reduction under typical re-
action conditions for the conversion of hydrocarbon feedstocks
(i.e., below 4508C) is expected to have a strong influence on
the performance of these catalysts.[40–42] Pure ceria had two
main peaks from hydrogen consumption at 440 and 7258C
(Figure 1). The low-temperature peak at 4408C can be attribut-
ed to surface reduction, whereas the high-temperature peak is
assigned to bulk reduction.[41] The addition of zirconia caused
the high-temperature peak to shift to lower temperatures.
Consequently, bulk reduction became difficult to distinguish
from surface reduction in ceria–zirconia mixed oxides as the
zirconia content increased. The maximum of the low-tempera-
ture peak shifted to higher temperatures as the fraction of zir-
conia was increased owing to an increasing contribution from
bulk reduction. It is important to note that the shoulder on the
low-temperature side of the first peak likely contains a contri-
bution from the chemisorption of hydrogen on the surface
(vide infra).[34] Therefore, it was not possible to distinguish be-
tween the uptake owing to adsorption and the uptake owing
to reduction at approximately 2008C from the TPR measure-
ments alone.
The tests for stable catalytic performance started at the high-
est temperature of 3758C. The temperature was lowered by
258C every 10 h until a temperature of 2758C was reached. Fi-
nally, the temperature was increased back to 3758C, and the
reactivity was compared to that at the start of the run to
probe whether catalyst deactivation had occurred. Over a 72 h
reaction, the change in conversion at 3758C was less than 2%
for all catalysts tested. This indicated that none of the catalysts
were deactivated over the course of the reaction.
Product distribution over different catalysts
The conversion of guaiacol and the corresponding product dis-
tribution over different ceria and ceria–zirconia catalysts were
compared at 3758C and a W/F ratio (i.e., mass of catalysts di-
vided by mass flow rate) of 1.4 h (Table 3). Over ceria, the prod-
uct mixture contained 67 mol% phenol, 10 mol% catechol,
Table 3. Product distribution for the HDO of guaiacol.[a]
Mixed Conversion Product distribution [mol%] (C basis)
oxide
phenol benzene cresols catechol oligomers
Low-temperature oxygen storage capacity
Ce100 37
67
73
88
93
1
1
1
0
7
19
10
6
10
0
0
9
0
0
0
Ce82
Ce60
Ce46
59
67
46
The concentration of oxygen vacancies formed during reduc-
tion at 200, 300, or 4508C was quantified by reduction in 10%
H2 in Ar followed by titration with O2 (Table 2). At 2008C, no
oxygen consumption was measured for any of the samples.
However, from the TPR experiments, it seemed that some hy-
0
[a] Reaction conditions: 3758C, 1 bar H2, W/F=1.4 h, feed flow rate=
0.008 mLminÀ1 or 0.36 ghÀ1, W=0.5 g catalyst.
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2075
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
7 mol% cresols, and 1 mol% benzene. In addition, three differ-
ent oligomeric products accounted for 9% of the carbon fed,
assuming the same response factor as that determined by the
GC calibration for phenol. Unfortunately, these compounds
could not be identified by GC–MS with any degree of certainty.
Methanol and methane were not detected as products for
Ce100. The lack of methanol in the product stream could be
explained by the production of cresols as well as the decom-
position of methanol to CO and hydrogen.[43] The selectivity to
phenol increased with increasing zirconia content of the cata-
lyst. Cresols were the other major products produced over the
mixed metal oxide catalysts. In addition, limited amounts of
anisole, catechol, and methanol were formed under certain
conditions (vide infra). Oligomeric products were not formed
in detectable quantities over ceria–zirconia catalysts at 3758C
and a W/F of 1.4 h. Only a small amount of benzene was found
in the product stream from reactions over pure ceria and
ceria–zirconia.
Influence of the weight-to-feed ratio
To elucidate the reaction path of guaiacol, the product distri-
bution over Ce82 at 4008C and 1 bar was studied at W/F=
0.1–11.2 h (i.e., feed flow rates of 0.001–0.08 mLminÀ1 or 0.04–
3.56 ghÀ1; Figure 2). As the W/F ratio increased, the conversion
of guaiacol to products increased. Phenol was the major prod-
uct under all reaction conditions tested. The yield of catechol
reached a maximum at W/F=1.4 h, which indicates that cate-
chol is a primary product of guaiacol that can be converted
further in subsequent reactions (Figure 2). The yield of cresols
increased with increasing W/F ratio. Benzene remained a minor
product over the range of W/F ratios tested, although the ben-
zene yield increased steadily as the W/F ratio increased.
More information about the reaction network can be ob-
tained from the selectivity to different products (Figure 2b).
The selectivity to catechol was 28% at the lowest W/F ratio
and decreased as the W/F ratio increased. The selectivity to-
wards phenol increased initially with increasing W/F ratio and
then began to decrease as the W/F ratio increased further. As
the phenol selectivity decreased at high W/F ratio, the selectiv-
ity towards cresols increased notably. This observation implies
that cresols are produced from phenol. The selectivity towards
benzene increased steadily with increasing W/F ratio but re-
mained below 4% even at the highest W/F ratio tested.
An additional experiment was conducted, in which metha-
nol (4.4 gLÀ1) and phenol (12.8 gLÀ1) were co-fed over Ce82 at
4008C and 1 bar with a W/F ratio of 1.4 h. The product mixture
consisted of 90 mol% cresols and 10 mol% benzene. No guaia-
col was detected. Methanol was not observed in the product
stream even though the consumption of methanol for the pro-
duction of cresols did not account for the entire amount of
methanol fed.
Figure 2. (a) Product yield and (b) selectivity during hydrodeoxygenation of
guaiacol over 0.5 g of Ce82 with varying W/F at T=4008C and P=1 bar. The
feed flow rates were 0.001–0.08 mLminÀ1 or 0.04–3.56 greactant hÀ1
.
Table 4. Conversion of guaiacol over ceria–zirconia mixed oxides.[a]
Mixed
oxide
Conversion [%]
Ea
2758C
3008C
3258C
3508C
3758C
[kJmolÀ1
]
Ce100
Ce82
Ce60
Ce46
2
3
5
2
10
8
5
22
24
16
15
39
40
27
37
59
67
46
131
97
111
114
<1
5
[a] Reaction conditions: 1 bar H2, a W/F ratio of 1.4 h, and a feed flow rate
of 0.008 mLminÀ1 or 0.36 greactant hÀ1
.
from 2% to 37% as the temperature increased from 275 to
3758C (Table 4). At 3008C and above, all mixed oxides ach-
ieved higher conversions than ceria at the same temperature.
The highest conversions were observed over Ce60 and Ce82.
At 3758C, 67% and 59% of the guaiacol were converted over
Ce60 and Ce82, respectively.
The yield and selectivity of different products changed with
temperature for all ceria–zirconia catalysts. The product distri-
bution for Ce82 is shown in Figure 3. The most-abundant prod-
uct at all temperatures was phenol. As the temperature in-
creased, the phenol yield increased (Figure 3a). The selectivity
to phenol decreased with increasing temperature, whereas the
selectivity towards cresol increased. The yield of cresol was
below 5.6% up to 3508C (Figure 3b). Benzene and anisole
Influence of temperature
All catalysts showed activity over the entire temperature range
studied (i.e., 275–3758C). For Ce100, the conversion increased
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2076
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the oxygen-containing functional groups of the reactant coor-
dinate to vacancies of the sulfide particles.[19,26,47] Unfortunate-
ly, sulfide catalysts suffer from deactivation when sulfur atoms
are replaced by oxygen during HDO.[4,48] Therefore, oxides with
comparable sites have great potential as catalysts for HDO.[47,49]
We demonstrated recently that hydrogen is activated readily
over ceria and ceria–zirconia catalysts.[34] Specifically, isotopic
scrambling between H2 and D2 occurs over these catalysts at
considerable rates, and the activation energy is ca. 23 kJmolÀ1
if oxygen vacancies are involved. Up to 10.5 hydrogen atoms
per nm2 participate in isotopic scrambling at 3008C. Based on
these results, it is safe to assume that HDO reactions over
these catalysts are not limited by the amount of reactive hy-
drogen on the surface, if the reaction is performed in a hydro-
gen-rich atmosphere.
In addition, ceria-based catalysts have a substantial oxygen
storage capacity in the temperature range of interest for HDO
(Table 2). Under reducing conditions, the first oxygen vacancies
are generated on the surface between 200 and 3008C through
the condensation of singly bound hydroxyl groups.[50] In cer-
tain ceria–zirconia samples, bulk oxygen atoms become mobile
and migrate to the surface once the temperature reaches 330–
4708C.[29,41] In this temperature range, bulk and surface reduc-
tion become difficult to distinguish. The incorporation of zirco-
nia increases the oxygen storage capacity notably. Zirconium
has a slightly smaller ionic radius than cerium, and this causes
a slight strain in the ceria–zirconia crystal lattice and facilitates
structural relaxation after the formation of an oxygen vacan-
cy.[29,32,51] Consequently, ceria–zirconia catalysts fulfill all re-
quirements for an HDO catalyst.
Figure 3. (a) Major product yields, (b) minor product yields, and (c) selectivi-
ties over Ce82 at 1 bar, a W/F ratio of 1.4 h, and a feed flow rate of
0.008 mLminÀ1 or 0.36 greactant hÀ1
.
were not detected below 3258C, and catechol was not detect-
ed below 4008C.
The reactivity studies show that these characteristics trans-
late into catalytic activity for the HDO of guaiacol. The catalytic
activity became notable at approximately the same tempera-
ture (i.e., ꢀ3008C) as the formation of oxygen vacancies.
Guaiacol conversion over ceria–zirconia catalysts was higher
than that over pure ceria over the entire temperature range
tested and increased with increasing temperature as expected
(Table 4). Although differences in surface area might contribute
to some of the differences in activity, they are insufficient to
account for these differences by themselves. A linear correla-
tion was observed between the concentration of oxygen va-
cancies at 3008C and the guaiacol conversion at the same tem-
perature (Figure 4). At 3008C the conversion of guaiacol was
still low enough to provide differential conditions. The correla-
The apparent activation energy for guaiacol conversion over
all catalysts was estimated based on the reactivity data be-
tween 300 and 3758C (Table 4). The conversion at 2758C was
neglected, as the values were too close to the experimental
error of the gas chromatographic analysis. The simplifying as-
sumption was made that guaiacol is converted in a single, irre-
versible, first-order reaction (Table 4). An apparent activation
energy of 131 kJmolÀ1 was found for Ce100. The apparent acti-
vation energies for the ceria–zirconia catalysts were between
97 and 114 kJmolÀ1 and followed the order Ce82<Ce60<
Ce46.
Discussion
Structure–property relationships for HDO over ceria–zirco-
nia catalysts
Hydrodeoxygenation catalysts must be able to cleave CÀO
bonds and provide dissociatively adsorbed hydrogen to re-
place oxygen atoms in organic molecules. The most commonly
discussed mechanism for HDO over supported noble metal
catalysts involves oxygen removal by the acid-catalyzed dehy-
dration and metal-catalyzed hydrogenation of C=O and C=C
bonds.[14,44–46] For sulfided NiMo and CoMo catalysts, an inverse
Mars–van Krevelen mechanism has been suggested, in which
Figure 4. Correlation of the concentration of oxygen vacancy sites (Table 2)
with guaiacol conversion over ceria and ceria–zirconia at 3008C (Table 4).
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2077
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tion between the concentration of oxygen vacancies and the
activity of the catalyst is a clear indication that oxygen vacan-
cies play a critical role in the conversion of guaiacol. However,
the trend line in Figure 4 crosses the horizontal axis at a vacan-
cy concentration of 49.2 mmol of O vacancies per g, which indi-
cates that a minimum density of vacancies is required for cata-
lytic activity. We speculate that two oxygen vacancy sites are
necessary to deoxygenate guaiacol. To remove an oxygen
atom from guaiacol over ceria-based catalysts, it seems to be
necessary for another functional group to coordinate to anoth-
er vacancy. Consequently, two vacancy sites must be close
enough to each other to activate efficiently a CÀO bond in
guaiacol.
of the reaction temperature. Specifically, the generation of va-
cancies in sulfide and oxide catalysts can be facilitated at ele-
vated temperatures, and it has been proposed that these are
the active sites for the inverse Mars–van Krevelen mechanism
over such catalysts.[47]
It is recognized that the issues outlined above will likely
affect the apparent activation energies for the HDO of guaiacol
over the ceria-based oxide catalysts used in this study. Never-
theless, the apparent activation energies were determined
based on the assumption of a single, first-order reaction to
provide a semi-quantitative comparison with other types of
catalysts. The ceria–zirconia mixed oxides had activation ener-
gies between 97–114 kJmolÀ1, which is within the range re-
ported for sulfided CoMo and NiMo catalysts.[14] A comparison
with the apparent activation energy for pure ceria
(131 kJmolÀ1) shows that the incorporation of zirconia lowers
the apparent activation energy.
Differences in the concentration of surface oxygen vacancies
could also explain the differences in activity over different cata-
lysts at higher temperatures (Table 3). Specifically, the same
conversion was observed over Ce60 and Ce82 up to 3508C,
whereas the conversion was significantly higher over Ce60 at
3758C. A similar trend was observed for the oxygen storage ca-
pacity over these catalysts at a given temperature (Table 2).
However, it is not possible to quantify directly the number of
oxygen vacancy sites on the surface through oxygen titrations
above approximately 3308C owing to the increasing mobility
of bulk oxygen atoms.[29,41] Consequently, titration with O2 at
4508C will likely also saturate subsurface and bulk vacancies.
Thus, the oxygen titration experiment at 4508C determines an
upper limit for the number of oxygen vacancies that are avail-
able as active sites at the surface. Subsurface vacancies could
possibly have an influence on the catalytic performance in
HDO reactions, but this phenomenon is beyond the scope of
this work.
In other studies, it has been speculated that Lewis[4] and
Brønsted acidity[14,44–46] are important for HDO activity. The
ceria-based catalysts used in the present study only contain
Lewis acid sites (Table 1).[34] However, there was no clear quan-
titative correlation between the concentration of Lewis acid
sites and the conversion of guaiacol over these catalysts. The
lack of such a correlation does not prove that Lewis acid sites
are not involved in the reaction. Specifically, it should be men-
tioned that oxygen vacancies can act as Lewis acids. However,
Lewis acid sites can also be found in other environments.
Based on the present data, a fraction of the Lewis acid sites is
apparently not involved directly in the rate-determining step
of the HDO reaction.
The range of activation energies for HDO over ceria–zirconia
was comparable to the oxygen vacancy formation energies for
these materials, as determined by DFT+U calculations (DFT
with the inclusion of on-site Coulomb correction).[51] This indi-
cates that the regeneration of the oxygen vacancy at the end
of the catalytic cycle is likely the rate-limiting step for HDO. Zir-
conium cations lower the required energy for oxygen vacancy
formation through electronic effects and by facilitating struc-
tural relaxation after the removal of the oxygen atom.[51] In ad-
dition, zirconium distorts the CaF2 lattice of pure ceria. This ap-
pears to result in less-efficient crystallization during the calcina-
tion of the mixed metal oxides, as indicated by their smaller
crystallite sizes (Table 1). Consequently, edge and kink sites are
more abundant in the mixed oxides. The reduction (i.e.,
oxygen vacancy formation) occurs more readily at such sites.[53]
Moreover, it is suggested that smaller crystallites allow for the
formation of oxygen vacancies in closer proximity, so that the
surface interactions of both functional groups in guaiacol can
occur more readily.
Reaction network for HDO over ceria–zirconia catalysts
The hydrodeoxygenation of guaiacol has been studied over
a variety of different catalysts, including sulfided NiMo and
CoMo,[16,54] noble metals,[44,55] nitrides,[56] and phosphides.[52]
However, to our knowledge, there have been no reports of the
HDO of guaiacol over reducible metal oxides. Guaiacol is an
ideal model compound for assessing the performance of differ-
ent catalysts because it is a typical constituent of pyrolysis oils
and one of the most difficult methoxyphenols to deoxygen-
ate.[14]
Apparent activation energy for the HDO of guaiacol
Reports on the apparent activation energies for HDO vary over
a considerable range even if catalysts with similar composi-
tions are compared.[14,52] For example, for the HDO of guaiacol
over sulfided CoMo and NiMo supported on Al2O3, values of
50–112 kJmolÀ1 have been reported.[14] In most studies, the ap-
parent activation energies are calculated based on simplifying
assumptions. In particular, the conversion of a reactant is treat-
ed as a single, first-order reaction, whereas parallel reactions of
different orders may be involved. In addition, it has to be con-
sidered that the concentration of active sites may be a function
Several reaction networks have been suggested for the HDO
of guaiacol.[4,44,55] The bond energies of the CÀO bonds in
guaiacol increase in the order MeÀOAr<MeOÀAr<HOÀAr.[6]
Thus, it is not surprising that the hydrogenolysis of the H3CÀ
OAr bond (i.e., demethylation) was proposed as the first step
in many reaction paths. Catechol and methane are formed in
this reaction. Methane was observed over various catalysts in-
[6,16]
cluding Pt–Sn monoliths,[44] sulfided CoMo/Al2O3,
Pt/MgO,[57]
and Fe/SiO2.[58] Alternatively, a methyl group can remain on the
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2078
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
surface. In many cases, demethylation activity is attributed to
the metal sites on these catalysts.[44,55] An alternative pathway
is the demethoxylation (i.e., the cleavage of the MeOÀAr
bond) to produce phenol and methanol. This pathway was ob-
served over CoMo sulfides supported on alumina and zirco-
nia.[7,47,54,59] Bui et al. reported that this reaction step is favored
over catalysts with amphoteric supports, such as zirconia.[6]
The primary reaction products can be dehydroxylated to yield
completely deoxygenated products. However, over many cata-
tion to catechol. This observation is in line with a previous
report, in which the amphoteric character of zirconia was pro-
posed to favor the demethoxylation of guaiacol.[7] Moreover,
the lack of detectable quantities of methane in the product
stream is in agreement with the limited relevance of the deme-
thylation path. Methanol, which is formed during the conver-
sion of guaiacol to phenol, is converted in other reactions
(vide infra). Not surprisingly, the low yield to anisole showed
that the cleavage of the strong HOÀAr bond in guaiacol was
lysts, the aromatic ring must be hydrogenated before the last not kinetically relevant.
ÀOH group can be removed in a dehydration step.[60] In addi-
tion to deoxygenation, transalkylation reactions are often ob-
served during the HDO of guaiacol, in particular, over Lewis
acidic catalysts.[7] Consequently, the major reaction products in
different HDO studies included phenol, catechol, cresol, ben-
zene, cyclohexane, and cyclohexene.[61]
The maximum of the catechol yield at intermediate W/F
ratios showed that this compound is further deoxygenated to
form phenol. Consequently, only a limited amount of catechol
was found as the W/F ratio increased. In contrast to the high
reactivity of catechol, the deoxygenation reaction of phenol to
benzene was slow, and the benzene yield never exceeded 4%
even at the longest W/F ratio of 11.2 h. Similar observations
were reported for the deoxygenation of guaiacol and the inter-
mediate cresol over a vanadium oxide on alumina catalyst.[63] It
was proposed that the oxygen atoms in the 1- and 2-positions
of the aromatic ring form a chelated complex with a partially
reduced vanadium atom. As the redox activities of vanadia-
and ceria-based oxides are comparable, we suggest that the
same mechanism may be relevant for the catalysts in the pres-
ent study. It is interesting to note that guaiacol was less reac-
tive than anisole over Pd/C in the presence of H3PO4, which
was explained with the stabilization of the carbocation transi-
tion state by the adjacent hydroxyl group.[46]
Although most studies on HDO were conducted over sup-
ported metals on metal oxides or sulfided catalysts, there are
a limited number of reports on oxide catalysts. The hydrodeox-
ygenation of acrolein, acetone, hexanone, cyclohexane, and
[47,49]
anisole was performed over a metal-free MoO3.
For anisole
as a reactant, deoxygenated aromatic products were formed in
significant yields, and no products with saturated rings were
detected.[47] MoO3 accumulated a certain amount of coke but
was regenerated readily by calcination. The concentration of
oxygen vacancies plays a critical role in the HDO activity of
MoO3.[49] Ceria–zirconia is a well-known redox catalyst that
forms vacancy sites readily, and its reducibility can be tuned by
adjusting the Ce/Zr ratio and by controlling the morphology of
the sample.[62] Thus, it provides a unique opportunity for tailor-
ing its performance in HDO reactions.
Owing to the absence of the activating chelate interaction,
the cleavage of the last ArÀOH bond is slow, as indicated by
the high yields to phenol and limited benzene yields. The ArÀ
OH bond is the strongest CÀO bond in guaiacol (and other
phenol derivatives), and several studies suggested that it
needs to be weakened by the hydrogenation of the aromatic
ring before it can be cleaved.[22,46,64] In the present study, the
product stream contained no partially or completely hydrogen-
ated ring structures, and benzene was formed in limited
amounts. This shows that ceria–zirconia catalysts have some
activity for the complete HDO of guaiacol without hydrogena-
tion of the aromatic ring. We speculate that this activity could
be enhanced by the presence of surface sites that interact
with the aromatic ring as the phenol group binds to an
oxygen vacancy.
The reaction pathway of guaiacol over the present ceria–zir-
conia catalysts can be derived from the changes of the yields
to different products as a function of the W/F ratio (Figure 2).
The results presented in Figure 2 are representative for all
ceria–zirconia catalysts tested in this study. If the curves are ex-
trapolated to the origin, positive slopes are obtained for
phenol and catechol; therefore, these compounds are the pri-
mary products of the conversion of guaiacol (Figure 5). At the
lowest W/F ratio of 0.14 h, the yield of phenol was 2.3 times
higher than the yield of catechol, which shows that the deme-
thoxylation of guaiacol to phenol is faster than the demethyla-
In addition to HDO, phenol can undergo a transalkylation to
different cresol isomers (Figure 5).[46,64] Several studies reported
that this reaction requires Lewis acidic sites that facilitate the
formation of a methyl cation as the transition state of the alky-
lation reaction.[6,7,65] On the other hand, it was proposed that
amphoteric metal oxides, such as ceria, are good catalysts for
the ortho-methylation of phenol with methanol.[66] We specu-
late that the surface methoxy species resulting from the con-
version of guaiacol to phenol can react with oxygen vacancies
to be reduced to surface-bound methyl groups, which partici-
pate in the transalkylation reaction. In agreement with this hy-
pothesis, the initial slope for the yield of cresol with respect to
the W/F ratio was 0, which implies that cresol is a secondary
product.
Figure 5. Reaction pathways for the HDO of guaiacol and transalkylation of
products.
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2079
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
The reversibility of the deoxygenation of guaiacol was
probed by cofeeding methanol and phenol. The absence of
detectable amounts of guaiacol indicates that the demethoxy-
lation reaction is irreversible. Although methanol was found in
the product mixtures for the HDO of guaiacol under certain
conditions, its concentration was never significant. As men-
tioned above, surface methoxy species are consumed in the al-
kylation of phenol to produce cresol. However, the production
of cresol does not account for all of the methoxy species or
methanol produced by the methoxylation of guaiacol. The
likely explanation for this discrepancy is that methanol can be
decomposed over ceria-based catalysts to produce CO and
with hydrogenated rings are formed, and the catalysts show
no signs of deactivation after 72 h on stream.
The reaction is suggested to occur by an inverse Mars–van
Krevelen mechanism, in which oxygen vacancies act as active
sites. The correlation between the concentration of oxygen va-
cancies and the conversion of guaiacol under differential con-
ditions indicates that the HDO of guaiacol appears to occur ef-
ficiently on paired oxygen vacancies. The pseudo-first-order ac-
tivation energy for the HDO of guaiacol over ceria–zirconia cat-
alysts is comparable to the energy required for the formation
of an oxygen vacancy; this indicates that the latter is the rate-
limiting step in the HDO. The catalysts have available oxygen
vacancies and also dissociate hydrogen on the surface, another
prerequisite for HDO activity. In contrast, the acidity of these
catalysts does not seem to affect their performance in HDO re-
actions. The incorporation of zirconium ions into the CaF2-type
lattice of ceria leads to a structural contraction, which facili-
tates the formation of oxygen vacancies. In addition to elec-
tronic effects, the presence of zirconium partially suppresses
crystallization during the synthesis of the catalysts and results
in smaller, more active crystallites. It is suggested that opti-
mized synthesis methods could increase the concentration of
oxygen vacancies under specific reaction conditions and
reduce the energy required for their formation.
[43]
H2. With the instruments available for the present study, it
was not possible to detect CO and H2.
In addition to HDO products, oligomeric products were
formed over pure ceria (Table 3). Ceria and ceria–zirconia cata-
lysts have been used in several studies on the dimerization of
aldehydes or carboxylic acids by aldol condensation[67–69] or ke-
tonization.[68–70] However, these pathways are unlikely to be rel-
evant in the present study owing to the absence of the rele-
vant functional groups. Instead, the oligomerization of guaia-
col or its products could occur via free radical species. Kaev
et al. reported that this pathway leads to the formation of
chromenones, xanthones, and fluorenones as the oligomeriza-
tion products from the wet oxidation of phenol over ceria-
based catalysts at 1608C.[71]
Experimental Section
Chemicals
Stability of ceria-based HDO catalysts
Cerium(III) nitrate hexahydrate (99% trace metals basis), zirconiu-
m(IV) oxynitrate hydrate (99% trace metals basis), ammonium hy-
droxide (A.C.S regent grade, NH3 content 28–30%), octane, decane,
guaiacol, methanol, and phenol were purchased from Sigma–Al-
drich. Gases (hydrogen, helium, 2000 ppm ammonia in helium,
argon, and oxygen) were purchased from Airgas with ultra-high
purity (UHP Grade 5). A Parker Balston Gas Generator 1000 was
used to generate air for the gas chromatograph and for calcination
of the catalysts. Deionized water was purified to 18.2 MWcmÀ1
with a Barnstead NANOpure ultrapure water system.
In many studies, traditional HDO catalysts have been deactivat-
ed by rapid coke formation.[14] The same problem was reported
for ceria catalysts in the wet oxidation of phenol at 120–
1608C.[71,72] In the present study, oligomeric products were
only detected over pure ceria, and no such species were de-
tected over the ceria–zirconia catalysts. As even the ceria cata-
lyst showed stable activity for 72 h, it is safe to suggest that
the oligomeric products do not cause deactivation under the
experimental conditions applied in this study.
In addition to coking, the loss of active sites has to be con-
sidered. Specifically, water is always formed as a byproduct in
HDO, and water dissociatively adsorbs on ceria–zirconia surfa-
ces and saturates the vacancies.[73] Therefore, it is imperative to
perform HDO reactions over ceria–zirconia catalysts under reac-
tion conditions that allow for surface reduction and, thus, the
regeneration of vacancies. The sustained activity observed over
all ceria-based catalysts in this study shows that this is achieved
above approximately 3008C under 1 bar of hydrogen (Table 4).
Catalyst preparation
All catalysts were prepared through the precipitation or coprecipi-
tation of cerium nitrate hexahydrate, zirconium oxynitrate hydrate,
or both. The precursors were dissolved in deionized water to form
a 0.1m solution. The synthesis method published by Rossignol
et al. was modified by adding the precursor solution dropwise to
an aqueous ammonium hydroxide solution (28–30%) with continu-
ous stirring.[74] Precipitation occurred as soon as each drop of pre-
cursor solution was added. The resulting precipitate was collected
by filtration, rinsed with deionized water, and subsequently dried
in an oven at 1008C overnight. The catalysts were calcined for 4 h
in 200 mLminÀ1 zero-grade air at 5008C with a ramp rate of
Conclusions
58CminÀ1
.
Ceria–zirconia catalysts without fully reduced metal particles
are active for the hydrodeoxygenation (HDO) of guaiacol.
Phenol and catechol are formed as the primary products, and
catechol is readily converted to phenol. Benzene is formed as
the completely deoxygenated product, but the final deoxyge-
nation step is much slower than the previous ones. In addition,
cresols are formed in transalkylation reactions. No products
Elemental analysis
The PIXE analysis was performed by Elemental Analysis Inc. to de-
termine the concentration of cerium and zirconium in each cata-
lyst.
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2080
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
oxygen in helium. Next, the sample was held at 200, 300, or 4508C
for 1 h and then reduced for 30 min at the same temperature in
a 10 mol% hydrogen in argon stream. Afterwards, the samples
were purged with He for 10 min to obtain a stable baseline at the
reduction temperature. Then, oxygen titration experiments were
performed by pulsing O2/He over the sample until no further
uptake was detected (three consecutive peaks of approximately
equal area). It was assumed that each O2 molecule saturated two
oxygen vacancies.
Nitrogen Physisorption
A Micromeritics ASAP 2020 physisorption analyzer was used to per-
form nitrogen physisorption at À1968C. Before analysis, each
sample (ca. 0.15 g) was degassed under vacuum at 2008C for 4 h.
The BET surface area was calculated from the nitrogen adsorption
isotherm between 0.05ꢁP/P0 ꢁ0.3.[75,76] The Barrett–Joyner–Halen-
da (BJH) method was used to calculate the pore volume.[77]
X-ray diffraction
A
Philips X’pert diffractometer equipped with an X’celerator
module and a CuKa radiation source (l=1.54 Š) was used to mea-
sure the powder XRD patterns. The diffractograms were measured
in the range of 2q=10–908 with 0.00835568 as the step size. The
Bragg equation was used to calculate the interplanar spacing. The
Scherrer equation was used to calculate the crystallite sizes based
on the diffractions for the (111), (200), (220), and (311) planes, and
the results were averaged to reduce the experimental error.[78–80]
Reactivity studies
The performance of the catalysts for the hydrodeoxygenation of
guaiacol was studied by using a vertical stainless-steel trickle-bed
reactor with an outer diameter of 0.25 inches. The catalyst was
used as a powder with a particle size in the range 75–90 mm. The
liquid feed was introduced into the reactor with a Teledyne ISCO
500D series syringe pump through a line preheated to 808C.
Hydrogen was supplied with a Brooks SLA 5850 mass-flow control-
ler. The temperature was controlled with a Eurotherm 2416 tem-
perature controller. The products were analyzed with an online
Aglient 7890A gas chromatograph with an HP-5 column and
a flame ionization detector (FID). If necessary, unknown products
were identified by the manual injection of known compounds into
the gas chromatograph or by GC–MS analysis with a Varian 450 GC
instrument attached to a Bruker 300 MS spectrometer. The GC
column was an Agilent VF-35ms.
Temperature-programmed desorption of ammonia
The temperature-programmed desorption (TPD) of NH3 was per-
formed with 50 mg of catalyst. The sample was placed in a quartz
U-tube and held in place by quartz wool. Gases were passed
through the tube with a flow rate of 50 cm3 minÀ1. The sample was
heated in 100% H2 to 4508C at 108CminÀ1 and held at this tem-
perature for 60 min. Then, it was cooled to 1208C in H2 flow and
purged with He for at least 15 min. NH3 adsorption was conducted
by flowing 10 mol% NH3/He for 30 min before the tube was
purged with He for at least 15 min. The TPD was performed by in-
creasing the temperature to 5508C at a rate of 208CminÀ1 in
a pure He flow. The sample was held at the final temperature for
15 min. The stainless-steel gas lines downstream of the catalyst
bed were heated to 1508C. Desorbing ammonia (m/z=17) was de-
tected with an MKS Instruments Cirrus mass spectrometer. The
contribution of water to m/z=17 was calculated from the intensity
of the signal at m/z=18 and subtracted. A single-point calibration
was performed with 10 mol% NH3 in helium before each experi-
ment.
The reactions were performed at various temperatures and flow
rates. Each set of conditions was held constant for 10 h for a total
time on stream of 70 h. In a typical experiment, the catalyst
(ꢀ 0.5 g) was placed in the reactor, and a reactant stream contain-
ing guaiacol (33.9 gLÀ1), decane (702 gLÀ1) as solvent, and octane
(5.9 gLÀ1) as an internal standard was fed to the reactor. A different
reactant stream comprising methanol (4.4 gLÀ1
) and phenol
(12.8 gLÀ1) was co-fed over Ce82 at 4008C and 1 bar H2 and
a weight-to-feed ratio (W/F) of 1.4 h in one experiment to under-
stand the reaction pathway to HDO products. The flow rate of hy-
drogen was kept constant at 40 mLminÀ1. The flow rate of the
liquid reactant mixture was varied between 0.001 and
0.08 mLminÀ1 to obtain W/F ratios of 11.2 and 0.1 h, respectively.
The temperature was varied from 275 to 4008C, and the pressure
in the reactor was held constant at 1 bar. The conversion of guaia-
col (X) was calculated as:
Temperature-programmed reduction (TPR)
A Micromeritics Autochem II 2920 instrument was used to perform
temperature-programmed reduction (TPR) of each sample
(ꢀ 0.2 g). The catalyst was heated at 108CminÀ1 in 25 mLminÀ1 of
10 mol% oxygen in helium to 4508C, held at 4508C for 1 h, and
then cooled to 08C at 158CminÀ1. Then, the catalyst was heated to
10008C with a ramp rate of 108CminÀ1 in 10 mol% hydrogen in
argon. The gas mixture exiting the reactor was monitored with
a thermal conductivity detector (TCD). A temperature-programmed
experiment was performed with Ce82 under a pure helium flow to
verify the baseline stability and ensure that no desorbing species
were contributing to the TCD signal over the entire temperature
range.
Fg,inÀFg,out
X ¼
ð1Þ
Fg,in
Fg,in is the molar flow rate of guaiacol into the reactor, and Fg,out is
the molar flow rate of guaiacol out of the reactor.
The yields of different products (Yp) are defined as:
Fp,out
Yp
¼
Fg,in
ð2Þ
Fp,out is the molar flow rate of a specific product leaving the reac-
tor.
Quantification of oxygen vacancies
The selectivity of a specific product is defined as:
Oxygen storage capacity measurements were performed to quanti-
fy the concentration of oxygen vacancies of each sample (ꢀ 0.2 g)
by using a Micromeritics Autochem II 2920 instrument. First, the
sample was heated to 4508C in a flow of 25 mLminÀ1 of 10 mol%
Yp
ð3Þ
Sp ¼
X
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2081
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[33] E. S. Putna, T. Bunluesin, X. L. Fan, R. J. Gorte, J. M. Vohs, R. E. Lakis, T.
Egami, Catal. Today 1999, 50, 343–352.
Acknowledgements
[34] S. M. Schimming, G. S. Foo, O. LaMont, A. K. Rogers, M. M. Yung, A. D.
D’Amico, C. Sievers, J. Catal. 2015, DOI: 10.1016/j.jcat.2015.05.027.
[35] D. E. Resasco, J. Phys. Chem. Lett. 2011, 2, 2294–2295.
[36] L. S. Birks, E. J. Brooks, Anal. Chem. 1950, 22, 1017–1020.
[37] S. Rossignol, F. Gerard, D. Duprez, J. Mater. Chem. 1999, 9, 1615–1620.
[38] P. Fornasiero, T. Montini, M. Graziani, J. Kaspar, A. B. Hungria, A. Marti-
nez-Arias, J. C. Conesa, Phys. Chem. Chem. Phys. 2002, 4, 149–159.
[39] V. Sµnchez Escribano, E. F. Lopez, M. Panizza, C. Resini, J. M. G. Amores,
G. Busca, Solid State Sci. 2003, 5, 1369–1376.
[40] S. Bernal, J. J. Calvino, G. A. Cifredo, J. M. Gatica, J. A. P. Omil, J. M. Pinta-
do, J. Chem. Soc. Faraday Trans. 1993, 89, 3499–3505.
[41] F. Giordano, A. Trovarelli, C. de Leitenburg, M. Giona, J. Catal. 2000, 193,
273–282.
Financial support by the Renewable Bioproducts Institute, New-
Page Corporation, International Paper, and the U.S. Department
of Energy’s Bioenergy Technologies Office (DOE-BETO) under con-
tract No. DE-AC36-08GO28308 with the National Renewable
Energy Laboratory is gratefully acknowledged. The authors would
like to thank Allison R. Rogers for the XRD measurements. M. K.
acknowledges financial support by the Friedrich-Ebert-Stiftung.
Keywords: arenes
· bio-oil · heterogeneous catalysis ·
hydrodeoxygenation · oxygen vacancies
[42] M. V. Ganduglia-Pirovano, A. Hofmann, J. Sauer, Surf. Sci. Rep. 2007, 62,
219–270.
[43] N. Yi, R. Si, H. Saltsburg, M. Flytzani-Stephanopoulos, Appl. Catal. B
2010, 95, 87–92.
[44] M. A. Gonzµlez-Borja, D. E. Resasco, Energy Fuels 2011, 25, 4155–4162.
[45] W. Zhang, J. Z. Chen, R. L. Liu, S. P. Wang, L. M. Chen, K. G. Li, ACS Sus-
tainable Chem. Eng. 2014, 2, 683–691.
[46] C. Zhao, J. He, A. A. Lemonidou, X. Li, J. A. Lercher, J. Catal. 2011, 280,
8–16.
[47] T. Prasomsri, T. Nimmanwudipong, Y. Roman-Leshkov, Energy Environ.
Sci. 2013, 6, 1732–1738.
[1] D. Mohan, C. U. Pittman, P. H. Steele, Energy Fuels 2006, 20, 848–889.
[2] B. van Ruijven, D. P. van Vuuren, Energy Policy 2009, 37, 4797–4808.
[3] A. Demirbas, Prog. Energy Combust. Sci. 2007, 33, 1–18.
[4] P. M. Mortensen, J. D. Grunwaldt, P. A. Jensen, K. G. Knudsen, A. D.
Jensen, Appl. Catal. A 2011, 407, 1–19.
[5] P. McKendry, Bioresour. Technol. 2002, 83, 37–46.
[6] V. N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Appl. Catal. B 2011, 101,
239–245.
[48] E. Furimsky, F. E. Massoth, Catal. Today 1999, 52, 381–495.
[49] D. R. Moberg, T. J. Thibodeau, F. G. Amar, B. G. Frederick, J. Phys. Chem.
C 2010, 114, 13782–13795.
[50] M. Daturi, C. Binet, J. C. Lavalley, A. Galtayries, R. Sporken, Phys. Chem.
Chem. Phys. 1999, 1, 5717–5724.
[7] V. N. Bui, D. Laurenti, P. Delichere, C. Geantet, Appl. Catal. B 2011, 101,
246–255.
[8] G. W. Huber, S. Iborra, A. Corma, Chem. Rev. 2006, 106, 4044–4098.
[9] L. Reijnders, M. A. J. Huijbergts, Biofuels for Road Transport: A Seed to
Wheel Perspective, Springer, London, 2009.
[51] H.-T. Chen, J.-G. Chang, J. Chem. Phys. 2010, 132, 214702.
[52] H. Y. Zhao, D. Li, P. Bui, S. T. Oyama, Appl. Catal. A 2011, 391, 305–310.
[53] U. M. Bhatta, D. Reid, T. Sakthivel, T. X. T. Sayle, D. Sayle, M. Molinari,
S. C. Parker, I. M. Ross, S. Seal, G. Mobus, J. Phys. Chem. C 2013, 117,
24561–24569.
[54] Y.-C. Lin, C.-L. Li, H.-P. Wan, H.-T. Lee, C.-F. Liu, Energy Fuels 2011, 25,
890–896.
[55] A. Gutierrez, R. K. Kaila, M. L. Honkela, R. Slioor, A. O. I. Krause, Catal.
Today 2009, 147, 239–246.
[56] I. T. Ghampson, C. Sepulveda, R. Garcia, B. G. Frederick, M. C. Wheeler, N.
Escalona, W. J. DeSisto, Appl. Catal. A 2012, 413, 78–84.
[57] T. Nimmanwudipong, C. Aydin, J. Lu, R. C. Runnebaum, K. C. Brodwater,
N. D. Browning, D. E. Block, B. C. Gates, Catal. Lett. 2012, 142, 1190–
1196.
[58] R. N. Olcese, M. Bettahar, D. Petitjean, B. Malaman, F. Giovanella, A.
Dufour, Appl. Catal. B 2012, 115, 63–73.
[59] G. de La Puente, A. Gil, J. J. Pis, P. Grange, Langmuir 1999, 15, 5800–
5806.
[10] T. R. Carlson, G. A. Tompsett, W. C. Conner, G. W. Huber, Top. Catal. 2009,
52, 241–252.
[11] A. V. Bridgwater, G. V. C. Peacocke, Renewable Sustainable Energy Rev.
2000, 4, 1–73.
[12] R. H. Venderbosch, W. Prins, Biofuels Bioprod. Biorefin. 2010, 4, 178–208.
[13] C. Briens, J. Piskorz, F. Berruti, Int. J. Chem. React. Eng. 2008, 6, R2.
[14] E. Furimsky, Appl. Catal. A 2000, 199, 147–190.
[15] P. R. Wentrcek, H. Wise, J. Catal. 1978, 51, 80–85.
[16] A. L. Jongerius, R. Jastrzebski, P. C. A. Bruijnincx, B. M. Weckhuysen, J.
Catal. 2012, 285, 315–323.
[17] E. Laurent, B. Delmon, J. Catal. 1994, 146, 281–291.
[18] A. Y. Bunch, X. Q. Wang, U. S. Ozkan, J. Mol. Catal. A 2007, 270, 264–
272.
[19] F. E. Massoth, P. Politzer, M. C. Concha, J. S. Murray, J. Jakowski, J.
Simons, J. Phys. Chem. B 2006, 110, 14283–14291.
[20] E. M. Ryymin, M. L. Honkela, T. R. Viljava, A. O. I. Krause, Appl. Catal. A
2010, 389, 114–121.
[21] D. C. Elliott, Energy Fuels 2007, 21, 1792–1815.
[22] P. M. Mortensen, J. D. Grunwaldt, P. A. Jensen, A. D. Jensen, ACS Catal.
2013, 3, 1774–1785.
[60] J. Horacek, G. St’avova, V. Kelbichova, D. Kubicka, Catal. Today 2013,
204, 38–45.
[61] M. Saidi, F. Samimi, D. Karimipourfard, T. Nimmanwudipong, B. C. Gates,
M. R. Rahimpour, Energy Environ. Sci. 2014, 7, 103–129.
[62] A. Trovarelli, Catal. Rev. Sci. Eng. 1996, 38, 439–520.
[63] J. Filley, C. Roth, J. Mol. Catal. A 1999, 139, 245–252.
[64] R. K. Sharma, N. N. Bakhshi, Can. J. Chem. Eng. 1991, 69, 1071–1081.
[65] R. Wandas, J. Surygala, E. Sliwka, Fuel 1996, 75, 687–694.
[66] S. Sato, K. Koizumi, F. Nozaki, Appl. Catal. A 1995, 133, L7–L10.
[67] F. C. Calaza, Y. Xu, D. R. Mullins, S. H. Overbury, J. Am. Chem. Soc. 2012,
134, 18034–18045.
[68] E. I. Gürbüz, E. L. Kunkes, J. A. Dumesic, Appl. Catal. B 2010, 94, 134–
141.
[69] A. Gangadharan, M. Shen, T. Sooknoi, D. E. Resasco, R. G. Mallinson,
Appl. Catal. A 2010, 385, 80–91.
[23] C. S. Song, Catal. Today 2003, 86, 211 –263.
[24] J. D. Holladay, J. Hu, D. L. King, Y. Wang, Catal. Today 2009, 139, 244–
260.
[25] X. L. Ma, K. Y. Sakanishi, I. Mochida, Ind. Eng. Chem. Res. 1994, 33, 218–
222.
[26] O. I. S¸enol, T. R. Viljava, A. O. I. Krause, Appl. Catal. A 2007, 326, 236–
244.
[27] L. Vivier, D. Duprez, ChemSusChem 2010, 3, 654–678.
[28] L. F. Liotta, A. Macaluso, A. Longo, G. Pantaleo, A. Martorana, G. Dega-
nello, Appl. Catal. A 2003, 240, 295–307.
[29] P. Fornasiero, G. Balducci, R. DiMonte, J. Kaspar, V. Sergo, G. Gubitosa, A.
Ferrero, M. Graziani, J. Catal. 1996, 164, 173–183.
[30] B. Bulfin, A. J. Lowe, K. A. Keogh, B. E. Murphy, O. Lübben, S. A. Krasni-
kov, I. V. Shvets, J. Phys. Chem. C 2013, 117, 24129–24137.
[31] S. Bernal, G. Blanco, J. M. Gatica, J. A. Perez-Omil, J. M. Pintado, H. Vidal
in Binary Rare Earth Oxides (Eds.: G. Adachi, N. Imanaka, Z. C. Kang),
Kluwer Academic Publishers, Dordrecht, 2004, pp. 9–55.
[32] F. Fally, V. Perrichon, H. Vidal, J. Kaspar, G. Blanco, J. M. Pintado, S.
Bernal, G. Colon, M. Daturi, J. C. Lavalley, Catal. Today 2000, 59, 373–
386.
[70] Y. A. Zaytseva, V. N. Panchenko, M. N. Simonov, A. A. Shutilov, G. A. Zen-
kovets, M. Renz, I. L. Simakova, V. N. Parmon, Top. Catal. 2013, 56, 846–
855.
[71] S. Keav, A. Martin, J. Barbier Jr., D. Duprez, C. R. Chim. 2010, 13, 508–
514.
[72] J. J. Delgado, X. Chen, J. A. PØrez-Omil, J. M. Rodríguez-Izquierdo, M. A.
Cauqui, Catal. Today 2012, 180, 25–33.
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2082
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[73] C. Liu, A. Karim, V. Lebarbier, D. Mei, Y. Wang, Top. Catal. 2013, 56,
[78] A. K. Cheetham, A. P. Wilkinson, Angew. Chem. Int. Ed. Engl. 1992, 31,
1557–1570; Angew. Chem. 1992, 104, 1594–1608.
1782–1789.
[79] B. D. Cullity, Elements of X-ray Diffraction, Addison-Wesley Pub. Co.,
Reading, 1956.
[80] A. E. Nelson, K. H. Schulz, Appl. Surf. Sci. 2003, 210, 206–221.
[74] S. Rossignol, Y. Madier, D. Duprez, Catal. Today 1999, 50, 261–270.
[75] J. Rouquerol, P. Llewellyn, F. Rouquerol, Stud. Surf. Sci. Catal. 2007, 160,
49–56.
[76] S. Brunauer, P. H. Emmett, E. Teller, J. Am. Chem. Soc. 1938, 60, 309–319.
[77] E. P. Barrett, L. G. Joyner, P. P. Halenda, J. Am. Chem. Soc. 1951, 73, 373–
380.
Received: March 2, 2015
Published online on June 3, 2015
ChemSusChem 2015, 8, 2073 – 2083
www.chemsuschem.org
2083
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim