Hydrodeoxygenation of guaiacol over carbon-supported metal catalysts

Article abstract of DOI:10.1002/cctc.201300096

Catalytic bio-oil upgrading to produce renewable fuels has attracted increasing attention in response to the decreasing oil reserves and the increased fuel demand worldwide. Herein, the catalytic hydrodeoxygenation (HDO) of guaiacol with carbon-supported non-sulfided metal catalysts was investigated. Catalytic tests were performed at 4.0MPa and temperatures ranging from 623 to 673K. Both Ru/C and Mo/C catalysts showed promising catalytic performance in HDO. The selectivity to benzene was 69.5 and 83.5% at 653K over Ru/C and 10Mo/C catalysts, respectively. Phenol, with a selectivity as high as 76.5%, was observed mainly on 1Mo/C. However, the reaction pathway over both catalysts is different. Over the Ru/C catalyst, the O-CH₃ bond was cleaved to form the primary intermediate catechol, whereas only traces of catechol were detected over Mo/C catalysts. In addition, two types of active sites were detected over Mo samples after reduction in H₂ at 973K. Catalytic studies showed that the demethoxylation of guaiacol is performed over residual MoO_x sites with high selectivity to phenol whereas the consecutive HDO of phenol is performed over molybdenum carbide species, which is widely available only on the 10Mo/C sample. Different deactivation patterns were also observed over Ru/C and Mo/C catalysts.

Full text of DOI:10.1002/cctc.201300096

CHEMCATCHEM
FULL PAPERS
FULL PAPERS
Convert hard, strike oil: The hydro-
deoxygenation of guaiacol, a model re-
action for bio-oil upgrading, is investi-
gated. Carbon-supported ruthenium
and molybdenum catalysts show prom-
ising activity and selectivity, although
the reaction occurs following different
reaction pathways.
J. Chang,* T. Danuthai, S. Dewiyanti,
C. Wang, A. Borgna*
&& – &&
Hydrodeoxygenation of Guaiacol over
Carbon-Supported Metal Catalysts
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
1
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
DOI: 10.1002/cctc.201300096
Hydrodeoxygenation of Guaiacol over Carbon-Supported
Metal Catalysts
Jie Chang,* Tanate Danuthai, Silvia Dewiyanti, Chuan Wang, and Armando Borgna*[a]
Catalytic bio-oil upgrading to produce renewable fuels has at-
tracted increasing attention in response to the decreasing oil
reserves and the increased fuel demand worldwide. Herein, the
catalytic hydrodeoxygenation (HDO) of guaiacol with carbon-
supported non-sulfided metal catalysts was investigated. Cata-
lytic tests were performed at 4.0 MPa and temperatures rang-
ing from 623 to 673 K. Both Ru/C and Mo/C catalysts showed
promising catalytic performance in HDO. The selectivity to
benzene was 69.5 and 83.5% at 653 K over Ru/C and 10Mo/C
catalysts, respectively. Phenol, with a selectivity as high as
76.5%, was observed mainly on 1Mo/C. However, the reaction
pathway over both catalysts is different. Over the Ru/C catalyst,
the OꢀCH3 bond was cleaved to form the primary intermediate
catechol, whereas only traces of catechol were detected over
Mo/C catalysts. In addition, two types of active sites were de-
tected over Mo samples after reduction in H2 at 973 K. Catalytic
studies showed that the demethoxylation of guaiacol is per-
formed over residual MoOx sites with high selectivity to phenol
whereas the consecutive HDO of phenol is performed over
molybdenum carbide species, which is widely available only
on the 10Mo/C sample. Different deactivation patterns were
also observed over Ru/C and Mo/C catalysts.
Introduction
Bio-oils produced from the fast pyrolysis of lignocellulosic bio-
mass contain complex mixtures of reactive oxygenate com-
pounds, such as carboxylic acids, aldehydes, ketones, furans,
sugars, carbohydrates and water.[1] The presence of these com-
pounds in bio-oils leads to the undesired fuel properties, such
as instability, high acidity, high viscosity and poor heating
value.[2] Therefore, the removal of oxygenate functional groups
is an important step in obtaining high-quality fuels from the
low-cost bio-oils.[3] Various processes such as cracking and hy-
drodeoxygenation (HDO) have been proposed to convert bio-
oils into alternative liquid fuels.[4]
er, their industrial application seems to be hindered by the
gradual sulfur removal during the long-term operation because
the sulfur content in bio-oil is quite low.[7] Co-feeding of H2S
can regenerate the sulfide sites and therefore stabilise the cat-
alyst. However, the resulting fuels are contaminated by sulfur,
which loses the advantage of sulfur-free fuels. Furthermore,
the presence of water and carboxylic acid in bio-oil can result
in the instability of alumina supports and serious coke forma-
tion.[8,9] Thus, significant efforts have been made to develop
carbon-supported
upgrading.[10]
non-sulfided
catalysts
for
bio-oil
Owing to the thermal instability of bio-oil, a two-stage up-
grading strategy was proposed. In the first stage the stabilisa-
tion of active organic compounds, such as ketones, aldehydes
and some acids, is performed under mild temperatures, where-
as in the second stage a deep HDO of stabilised bio-oil is per-
formed using higher temperatures. In this way, the yield loss
and coking can be inhibited effectively.
Transition metal-based catalysts, particularly noble metal cat-
alysts, can also be used in HDO reactions.[11] Heeres et al.[11a] re-
ported the valuable screening results for noble metal and sul-
fided Mo-based catalysts in the bio-oil HDO reaction. It was
found that the carbon-supported noble metal catalysts have
better performance than the latter. Particularly, the Ru/C cata-
lyst is the most promising candidate for bio-oil upgrading in
terms of oil yields, deoxygenation activity and hydrogen con-
sumption. Although the Pd/C catalyst provided higher oil
yields than the Ru/C catalyst, the higher hydrogen consump-
tion and higher oxygen content in the products are the major
drawbacks. However, char and heavily viscous products
formed in the batch reactor set-up from the raw bio-oil hin-
dered the quantitative analysis, and thus the understanding of
the reaction mechanisms.
The HDO of bio-oil is generally performed by using a hydro-
treating process under high pressure. Conventional hydrode-
sulfurization catalysts, particularly sulfided NiMo and CoMo cat-
alysts, have been used in this application.[5] These catalysts can
provide a good HDO activity and conversions up to 95%, to-
gether with a significant bio-oil yield at bench scale.[6] Howev-
Molybdenum carbides and oxycarbides were found to be
active in hydrotreatment and HDO reactions.[12] These materials
have electronic and catalytic properties similar to those of the
noble metals, but are more resistant to poisons such as sulfur
and to sintering; thus, they are potential substitutes for expen-
sive and scarce noble metals. However, there are few reports
[a] Dr. J. Chang, Dr. T. Danuthai, S. Dewiyanti, Dr. C. Wang, Dr. A. Borgna
Heterogeneous Catalysis Department
Institute of Chemical and Engineering Sciences
1, Pesek Road, Jurong Island (Singapore)
Fax: (+65)6316-6182
E-mail: chang_jie@ices.a-star.edu.sg
armando_borgna@ices.a-star.edu.sg
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
2
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
on the application of these
types of catalysts in bio-oil
upgrading.
Herein, we have systematically
screened different carbon-sup-
ported metal catalysts for HDO
under relatively moderate pres-
sure (4.0 MPa) in continuous
packed-bed reactor.[13] To gain
an insight into the mechanism
of catalytic HDO, guaiacol, a phe-
nolic compound with two types
of oxygen-containing functional
groups (phenolic and methoxyl
groups), was used as a model
compound for the second stage
of bio-oil upgrading to hydrocar-
bon fuels. Carbon-supported
metal catalysts, such as Pd/C,
Cu/C, Ru/C, W/C, Mo/C, Ir/C and
Ni/C, were screened in the HDO
of guaiacol. The catalyst screen-
ing indicated that Ru-based and
Scheme 1. Reaction network for guaiacol conversion.
Table 1. HDO of guaiacol over carbon-supported catalysts.[a]
non-sulfided Mo-based catalysts are the most promis-
ing ones. The reaction mechanisms over these two
types of catalysts are also discussed.
Parameter
Ir/C W/C Cu/C Pd/C Ni/C Ru/C 1Mo/C 10Mo/C
Metal loading [wt%]
Conversion of guaiacol [%]
Product distribution [wt%]
cyclohexane
benzene
toluene
1
1
5
1
5
1
1
10
3.2 3.5 3.6 15.5 30.7 34.2 74.1
88.9
Results and Discussion
–
–
–
–
–
–
–
–
–
–
0.7
2.2
3.4
0.3
1.5
–
5.4
35.4
–
44.7
3.3
2.3
2.5
3.4
–
15.3 11.7 34.2
Catalyst screening for the HDO of guaiacol
7.3
5.8
3.2
4.8
4.0
0.9
3.6
2.0
phenol
anisole
45.1 33.3
4.2 19.7 78.5
The possible reaction network for guaiacol conver-
sion, based on the product distribution observed
herein, is shown in Scheme 1. The initial HDO activi-
ties over different carbon-supported catalysts are
summarised in Table 1. Furthermore, the van Krevelen
plot (O/C versus H/C) for the liquid products obtained
over different catalysts is shown in Figure 1, which
provides a direct evidence of the HDO efficiency. No-
tably, Ir/C, W/C and Cu/C show negligible activity
compared to the thermal conversion of guaiacol
(2.4% conversion without catalyst under the same
testing conditions). 10Mo/C demonstrates the highest
11.2 10.0 3.7
40.4 34.0 22.2
48.5 1.9 33.1
–
–
–
–
–
–
2.6
1.5
1.8
0.3
4.6
1.2
1.3
0.4
6.2
3.6
2.1
5.0
–
methoxyl anisole
catechol
cyclohexanediol
cyclohexanone
cyclohexanol
methoxyl cyclohexane
methoxyl cyclohexanol
methoxyl cyclohexanone
8.6 4.0
–
–
–
–
–
–
2.6 14.4 21.7
1.5 16.1
4.3
26.7 23.5
1.5 4.2
11.2 25.5
4.9
6.7
–
–
–
–
–
–
0.1
1.9
–
0.8
1.0
–
–
[a] Reaction conditions: 623 K, 4 MPa, W/F=0.067 h, H2/feed=20.
guaiacol conversion and benzene selectivity at 623 K, followed
by 1Mo/C and Ru/C. Ru/C provides similar benzene selectivity
with 10Mo. However, the saturation of the aromatic ring is also
significantly observed and cyclohexanone is the main by-prod-
uct. This indicates that 10Mo/C has higher H2 efficiency during
the HDO of guaiacol, which is consistent with the observation
from the van Krevelen plot (Figure 1). In contrast to the former
two catalysts, even though a high guaiacol conversion is ob-
tained over 1Mo/C, phenol is the predominant product, with
a selectivity as high as 78.5 wt%. Although Ni/C and Pd/C sam-
ples demonstrate catalytic activity in the reaction, both cata-
lysts show high selectivity to products with a saturated C6 ring,
such as cyclohexanol, methoxyl cyclohexane, methoxyl cyclo-
hexanol and methoxyl cyclohexanone; this is in good agree-
Figure 1. van Krevelen plot for guaiacol conversion over different catalysts.
Reaction conditions: 623 K, 4 MPa, W/F=0.067 h, H2/feed=20.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
3
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
ment with a previous work,[11a] which reports that Pd-based
catalysts have a low HDO efficiency because they consume
more H2 for aromatic ring saturation rather than for oxygen
removal.
According to the screening results, Ru/C and Mo/C catalysts
are the most promising catalysts. Therefore, they were further
investigated to obtain optimised reaction parameters and
a better understanding of the reaction mechanism.
Effect of temperature on the HDO of guaiacol
The effect of the reaction temperature on the conversion of
guaiacol and product distribution was investigated over Ru/C,
1Mo/C and 10Mo/C catalysts (Table 2). The conversion of guaia-
Table 2. Conversion and selectivity over different catalysts.[a]
Parameter
Ru/C 1Mo/C 1Mo/C[b] 10Mo/C 10Mo/C[b]
Conversion [%]
Selectivity [wt%]
cyclohexane
benzene
toluene
100 98.4
78.7
100
99.4
7.2
–
–
1.4
–
4.1
83.5
–
–
1.6
–
69.5 2.4
–
–
phenol
anisole
methoxyl anisole
catechol
cyclohexanediol
cyclohexanone
cyclohexanol
methoxyl cyclohexane
methoxyl cyclohexanol
methoxyl cyclohexanone
5.1 76.5
1.8 3.5
0.3 0.8
0.2 0.1
0.8 14.6
65.5
7.2
12.6
0.9
9.4
0.5
0.1
–
7.7
0.4
–
–
2.2
–
0.5
0.1
1.0
–
71.2
7.6
7.0
0.6
9.3
–
0.1
–
–
Figure 2. Product distribution over a) 10Mo/C, b) 1Mo/C, and c) Ru/C cata-
lysts. Reaction conditions: 623 K, 4 MPa, W/F=0.067 h, H2/feed=20. Guaia-
~
&
*
col conversion ( ), benzene selectivity ( ), phenol selectivity ( ), catechol
&
selectivity ( ).
–
–
3.7 0.1
0.8 0.5
C sample. If the space–time is less than 0.0087 h, catechol is
the dominant product over the Ru/C catalyst. At higher con-
tact times, the selectivity to catechol decreases monotonically
whereas the selectivity to phenol reaches the maximum value
of 49 wt% at 0.017 h. If the space–time is more than 0.033 h,
the benzene yield increases significantly and becomes the
dominant product at 0.067 h. The evolution of product distri-
bution suggests that catechol is the primary intermediate,
which is formed via the demethylation of the methoxyl group
over Ru/C catalysts.
8.9
–
–
0.9
0.7 0.4
–
[a] Reaction conditions: 673 K, 4 MPa, W/F=0.067 h, H2/feed=20. [b] Mo
catalysts without pre-reduction.
col reaches 100% over the three catalysts if the reaction tem-
perature is increased from 623 to 673 K at a space–time of
0.067 h. Benzene selectivity over the Ru/C catalyst increases to
69.5 wt% with a dramatic decrease in selectivity to phenol and
cyclohexanone. The total selectivity toward aromatic ring satu-
ration decreases from 35.0 to 15.1 wt%. This is consistent with
the results reported by Gutierrez et al., which indicate that de-
oxygenation is favoured over hydrogenation at high tempera-
tures.[14] Over the 10Mo/C catalyst, benzene selectivity also in-
creases to 83.5 wt%. However, the selectivity to benzene over
the 1Mo/C sample remains unchanged to approximately
2.0 wt%. The selectivity to cyclohexanediol increases from 5.0
to 14.6%, which indicates the reverse trend of ring saturation
activity to Ru/C.
However, over both 10Mo/C and 1Mo/C catalysts, phenol is
detected as the dominant intermediate instead of catechol.
Only 6.8 wt% of catechol is detected at a space–time of
0.0087 h over the 10Mo/C catalyst. Catechol selectivity decreas-
es with increasing space–time. In contrast, phenol selectivity is
as high as 73.8 wt% at 0.0087 h and decreases slowly while
benzene production increases. The 1Mo/C catalyst shows an
even higher phenol yield. Approximately 80 wt% of phenol se-
lectivity is observed, which is constant in the entire range of
space–time, although the conversion level of guaiacol changes
from 12.2 to 98.5% with the change in space–time.
The van Krevelen plot shown in Figure 3 indicates that both
Ru/C and 10Mo/C catalysts can hydrodeoxygenate guaiacol to
fuel-like products with an O/C molar ratio lower than 0.05. In
addition, the 10Mo/C catalyst provides good H2 efficiency
whereas the Ru/C catalyst produces more saturated C6 ring if
guaiacol conversion approaches 100%. Over the 1Mo/C cata-
lyst, the O-containing groups of guaiacol cannot be removed
Effect of space–time on the HDO of guaiacol
The effect of space–time on product distribution over different
catalysts is shown in Figure 2. Because of the complexity of
product composition, only the selectivity to the main products
is shown. The selectivity changes with space–time over the Ru/
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
4
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
cies, which is represented by the area ratio of peak II/peak I, is
higher for the 1Mo/C catalyst than for the 10Mo/C catalyst (
35.6 vs. 4.5, respectively). Furthermore, the area ratio of peak
III/(peak I+peak II) for the 1Mo/C catalyst is much smaller than
that of the 10Mo/C catalyst (0.08 vs. 0.37, respectively). It indi-
cates that the dominant molybdenum oxide in the 1 wt% Mo
catalyst is Mo-T, which is more difficult to be reduced.
Because of the low loadings of Ru/C and 1Mo/C samples,
the active phases are well dispersed over the carbon surface
and crystallite sizes measured from CO chemisorption are 2.3
and 2.5 nm, respectively. Thus, only two broad and not well-
defined peaks, which can be assigned to activated carbon,
were observed for these samples. XRD was used only to ex-
plore the phase change during H2 reduction for the 10Mo/C
catalyst. The XRD patterns for the fresh and reduced 10Mo/C
catalysts are shown in Figure 5. The only Mo species observed
in the fresh 10Mo/C catalyst is MoO3, with diffraction peaks at
~
&
Figure 3. van Krevelen plot over catalysts 10Mo/C ( ), 1Mo/C ( ), and Ru/
*
C ( ). Reaction conditions: 623 K, 4 MPa, W/F=0.067 h, H2/feed=20. The
input of guaiacol feedstock is marked by I.
completely and phenol is the dominant product if guaiacol
conversion approaches 100% at a space–time of 0.068 h.
Catalyst characterisation
According to the catalytic tests over Ru- and Mo-based cata-
lysts, the catalytic behaviour is significantly different over these
samples. To identify the nature of active phases in Ru/C and
Mo/C catalysts, temperature programmed reduction (TPR), XRD
and XPS characterisations were performed.
The TPR profiles illustrated in Figure 4 show a sharp H2 con-
sumption peak and a weak broad peak for the Ru-based
sample at 394 and 495 K, respectively. This indicates that RuOx
is reduced to Ru0.[16] The TPR profile of carbon-supported Mo
Figure 5. XRD patterns of a) fresh Ru/C, b) fresh 1Mo/C, c) fresh 10Mo/C and
~
*
*
d) reduced 10Mo/C catalysts at 973 K. MoO3 ( ), MoO2 ( ), b-Mo2C ( ) and
0
&
metal Mo ( ).
23.3, 25.7, 27.3, 33.7 and 38.98. After reduction in flowing H2 at
973 K, three Mo phases can be identified on the basis of
JCPDS database: MoO2 (25.9, 36.9 and 53.58), beta-Mo2C (34.5,
38.0 and 39.68) and metallic Mo0 (40.4 and 58.78). The forma-
tion of metallic Mo0 during the reduction of Mo/C catalysts
was observed by Li.[17c] It was reported that metallic Mo0 can
be formed over Mo/C catalysts if the reduction temperature is
lower than 973 K. If the temperature is increased, Mo0 reacts
with the carbon support and forms beta-Mo2C. Furthermore,
Mo0 species can be carburised easily by carbon-containing
feedstock at approximately 510–570 K at the initial reaction
time.[17a,b,d]
Figure 4. Temperature programmed reduction profiles of a) Ru/C, b) 1Mo/C
and c) 10Mo/C catalysts.
The Mo3d X-ray photoelectron spectra of 1Mo/C and 10Mo/
C catalysts are shown in Figure 6. The deconvoluted XPS re-
sults, providing the distribution of Mo species, are summarized
in Table 3. Notably, the XPS spectra of the as-prepared 1Mo/C
and 10Mo/C catalysts are similar, which shows only one dou-
blet located at 232.4 eV and thus indicates the presence of
only Mo6+ species (MoO3); this is in good agreement with the
XRD results. The peak deconvolution of the reduced 1Mo/C
catalyst reveals two doublets. The new doublet with a Mo3d5/2
binding energy of 229.4 eV can be assigned to the formation
of Mo4+ during H2 reduction.[18] The atomic percentage of
MoO3 and MoO2 are 76.6 and 23.4%, respectively. Although
catalysts shows mainly three H2 consumption peaks appearing
at 680, 780 and approximately 900 K, marked by dotted lines.
It was suggested that the first two peaks, at temperatures
lower than 800 K, may be attributed to the reduction of Mo
species with octahedral coordination while the H2 consump-
tion peak at approximately 800 K may be attributed to the re-
duction of tetrahedrally coordinated Mo species (Mo-T).[15] Peak
III can be attributed to the carburisation and deep reduction[17]
of Mo species over carbon surfaces. The deconvolution of TPR
profiles shows that the content of more refractory Mo-T spe-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
5
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
similar to the one obtained on the reduced 1Mo/C catalyst.
The spent catalysts were analysed further by using XPS. Molyb-
denum carbides cannot be detected, whereas Mo6+ and Mo4+
are the dominant species over both catalysts (Table 3, col-
umns 3 and 4). These results suggest that MoOx is the active
phase for the conversion of guaiacol into phenol. Phenol could
be further converted into benzene on carbide sites.
Additional catalytic tests using intermediate reaction
products
Additional catalytic tests using intermediate reaction prod-
ucts—anisole and catechol—as feedstock were performed
over Ru/C and Mo/C catalysts. Owing to the high boiling point
of catechol, it was fed as a 20 wt% solution in guaiacol.
Anisole can be readily converted over the three catalysts
(Table 4). Ru/C and 10Mo/C catalysts demonstrate a product
distribution similar to that for the HDO of guaiacol (Table 2).
The cleavage of the anisole arylꢀOCH3 bond was studied by
Afifi et al.[20] The cleavage of the OꢀCH3 bond [Eq. (1)] is ener-
getically favoured in comparison to that of the arylꢀO bond
[Eq. (2)]. Over metal catalysts, a weak coordinative bond can
be formed between the methoxyl group and the active site
through the free electron pairs of the oxygen atom. Then, the
OꢀCH3 bond dissociates homolytically and forms phenoxide
and methyl radicals, which are hydrogenated to phenol and
methane..[21] Thus, catechol was observed as the primary inter-
Figure 6. Mo 3d XPS spectra of 1Mo/C and 10Mo/C catalysts: a1) fresh 1Mo/
C, a2) reduced 1Mo/C, a3) spent 1Mo/C, b1) fresh 10Mo/C, b2) reduced
10Mo/C and b3) spent 10Mo/C.
Table 3. Chemical states of reduced Mo-based catalysts derived from the
peak deconvolution of XPS curves.
Mo species XPS peaks
Relative percentage [%]
Mo3d3/2 Mo3d5/2 1Mo/C 10Mo/C 1Mo/C[a] 10Mo/C[a]
Mo6+
235.6
233.0
232.0
232.7
229.4
228.6
76.6
23.4
–
25.6
33.9
40.5
65.4
34.6
–
56.8
43.3
–
Mo4+
Carbide
[a] Spent Mo-based catalysts (without H2 reduction before reaction).
two Mo species are detected on the low Mo loading
catalyst, the XPS spectrum of the reduced 10Mo/C
catalyst reveals the existence of three doublets. The
Mo3d5/2 peak located at 229.4 eV is assigned to Mo4+
species of MoO2, whereas the doublet at 228.6 eV is
assigned to a molybdenum carbide phase, which is
in good agreement with the binding energy reported
in the literature on carbide species.[19] The corre-
sponding atomic percentage of MoO2 and Mo2C are
33.9 and 40.5%, respectively. The XPS surface analysis
reveals significant differences between both Mo cata-
lysts. Although oxide species are detected only in the
1Mo/C sample, molybdenum carbide is the dominant
species in the 10Mo/C catalyst.
Table 4. Conversion and selectivity over different catalysts in anisole and catechol
HDO reactions.[a]
Parameter
Anisole
Ru/C
20% catechol in guaiacol
1Mo/C 10Mo/C Ru/C
1Mo/C
10Mo/C
Conversion [%]
52.3[b] 19.2[b]
19.4[b]
85.5[c]
95.9[d]
27.3[c]
44.1[d]
47.5[c]
67.9[d]
Selectivity [wt%]
cyclohexane
benzene
toluene
phenol
1.1
77.5
–
3.4
–
–
74.9
1.4
20.6
–
–
–
1.2
0.9
1.0
–
–
84.4
5.7
8.1
–
–
–
–
–
8.7
42.4
1.0
31.6
3.4
0.1
1.8
3.6
6.2
0.8
–
0.1
1.0
1.1
76.5
3.0
0.8
11.1
0.2
5.7
0.2
–
–
29.0
5.9
52.8
1.6
1.0
8.2
0.5
–
anisole
methoxyl anisole
cyclohexanediol
cyclohexanone
cyclohexanol
methoxyl cyclohexane
methoxyl cyclohexanol
methoxyl cyclohexanone
–
5.1
12.9
–
–
–
To clarify the relationship between the chemical
states of Mo species and the catalytic performance,
catalytic tests using unreduced Mo samples were per-
formed (Table 2, columns 3 and 5). Phenol is the
main product over both catalysts. Therefore, the un-
reduced catalysts demonstrate a product distribution
1.9
–
–
–
–
–
–
0.2
0.1
[a] Reaction conditions: 673 K, 4 MPa, W/F=0.067 h and H2/feed=20; [b] Conversion
of anisole; [c] Conversion of guaiacol; [d] Conversion of catechol.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
6
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
mediate of the HDO of guaiacol over the Ru/C catalyst because
it is the preferred reaction pathway from an energetic point of
view. Therefore, the HDO of guaiacol over the Ru/C catalyst
should follow the reaction pathway suggested by Delmon and
Laurent (Scheme 2).[22]
Figure 7. FTIR spectra of guaiacol: a) guaiacol, b) guaiacol adsorbed on
MoO3 and c) guaiacol adsorbed on reduced MoO3.
Scheme 2. Mechanism of the HDO of guaiacol proposed by Delmon and
Laurent.[22]
1610 and 1512 cmꢀ1, respectively. In addition, significant con-
tribution from d(OH) vibrations, shifting from 1362 to
1370 cmꢀ1, is observed. It could be attributed to interactions
through hydrogen bonds. Over reduced MoO3, the main bands
for aromatic ring vibration at 1597 and 1501 cmꢀ1 shift to
lower wavenumbers, 1585 and 1494 cmꢀ1, respectively. This in-
dicates the presence of an electrodonation effect of the
oxygen atom on the aromatic ring over molybdenum oxide
with lower oxidation states. In addition, the d(OH) vibration at
1362 cmꢀ1 is no longer detected while the other bands are still
quite strong. This could indicate the removal of ꢀOH groups
and the formation of phenate species. The intensity of the
band at 1253 cmꢀ1, assigned to the asymmetric CꢀOꢀC bond,
decreases significantly, whereas the band at 1225 cmꢀ1, as-
signed to the CꢀOCH3 bond, disappears and a new band at
1276 cmꢀ1 is detected, which could be attributed to the inter-
action between methoxyl groups and MoOx. An IR spectro-
scopic study on the adsorption and activation of guaiacol over
oxides was reported by Popov et al.[25] The peaks at approxi-
mately 1330 and 1225 cmꢀ1 were assigned to arylꢀOH and
arylꢀOCH3 bonds, and both are eliminated at a high tempera-
ture (673 K). Therefore, the formation of a doubly-anchored
guaiacol surface species was suggested over oxides with Lewis
sites. This is consistent with our observations over MoOx,
which indicates that a doubly-anchored intermediate could be
formed.
However, this mechanism cannot explain the product distri-
bution obtained over the 1Mo/C catalyst during guaiacol ex-
periments (Table 2), in which phenol is the main product with
a selectivity of approximately 80%. Similarly, a high phenol se-
lectivity during the HDO of guaiacol over carbon-supported
CoMo sulfide catalysts was reported by Delmon et al.[23] This
was explained assuming that the direct hydrogenolysis of the
arylꢀOCH3 bond may occur on metal sulfides; however, no
clear evidence was provided. Notably, over the 1Mo/C catalyst,
in which MoOx is the dominant species, benzene is formed as
the main product during anisole experiments. This indicates
that demethoxylation rather than demethylation is the main
reaction pathway over MoOx. This finding is also supported by
our previous guaiacol experiments, which shows that a minor
benzene yield is obtained over Mo-based catalysts without car-
bide phases. No further HDO of phenol occurs over MoOx spe-
cies, as clearly indicated in Table 2.
Co-feeding catechol with guaiacol significantly inhibits
guaiacol conversion while a higher phenol/benzene ratio is ob-
served. This may be attributed to the competitive adsorption
between catechol and guaiacol, particularly on Mo-based cata-
lysts. It is known that Mo6+ ions can react with catechol to
form a stable complex.[24] These results also suggest a strong
interaction between catechol and other Mod+ ions, as indicat-
ed by the decrease in the reaction rate in catechol co-feeding
experiments. Furthermore, the increase in phenol selectivity on
the 10Mo/C catalyst, in which carbides are the dominant sur-
face species and have less effect on catechol competitive ad-
sorption, indicates that although catechol can also be convert-
ed, without the synergistic effects of MoOx sites its oxygen-re-
moval efficiency is significantly reduced. Thus, the direct cleav-
age of the arylꢀOCH3 bond resulting in phenol formation
seems to be the predominant pathway over MoOx sites to
form catechol as an intermediate in comparison to the deme-
thylation route.
Furthermore, a redox mechanism of guaiacol deoxygenation
via doubly-anchored guaiacol intermediate over Al2O3-support-
ed vanadium oxide catalysts was proposed by Filley and
Roth.[26] In this mechanism, methyl catechol is formed firstly
over the Lewis acid sites of the Al2O3 support. Then, oxygen
species available on the neighbour site of phenolic groups are
subtracted by V3+ to restore the V5+ sites while hydrogen is
transferred to an aryl carbon to break the arylꢀO bond, which
yields cresol.
Therefore, for Mo/C catalysts, an activation mechanism of
guaiacol via a doubly-anchored surface intermediate over re-
sidual MoO3 (Scheme 3) could be envisaged. In contrast to the
case of vanadium oxide, no significant production of alkylation
products, such as methyl anisole and cresol, was detected over
our Mo/C catalysts, owing probably to the lack of acidity of
the carbon surface. However, methanol is observed in the
FTIR study of adsorbed guaiacol
The typical FTIR spectrum of guaiacol between 1100 and
1700 cmꢀ1 as a reference is shown in Figure 7a. Over MoO3,
the aromatic C=C band positions shift to higher wavenumbers,
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
7
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Scheme 3. Doubly-anchored activation mechanism of the HDO of guaiacol
over MoOx.
liquid product obtained if Mo-based catalysts are used. This in-
dicates that the cleavage of the arylꢀOCH3 bond and consecu-
tive methylation before activation is not required, at least over
carbon-supported Mo catalysts. It could be speculated that
guaiacol adsorbs molecularly on Mo4+ sites to form a doubly-
chelated phenate intermediate directly instead of getting con-
verted into methyl catechol first. Thus, the oxygen atom is cap-
*
Figure 8. Catalyst deactivation during the HDO of guaiacol over 10Mo/C ( ),
~
&
1Mo/C ( ) and Ru/C ( ) catalysts. Reaction conditions: 673 K, 4.0 MPa, W/
F=0.033 h, H2/feed=20.
Conclusions
The hydrodeoxygenation of guaiacol, a model reaction for bio-
oil upgrading to produce renewable hydrocarbon-like fuels,
was investigated. The reaction occurs readily over carbon-sup-
ported Ru and Mo catalysts under moderate pressure (4.0 MPa)
and yields benzene and phenol as dominant products. The
product distribution indicates that the conversion of guaiacol
occurs through different reaction pathways over Mo/C and Ru/
C catalysts. Over MoOx sites, guaiacol could be converted di-
rectly into phenol via the formation of a doubly-anchored in-
termediate, followed by the direct removal of the methoxyl
group. In contrast, the formation of phenol over Ru/C catalysts
occurs preferentially via the cleavage of the CH3ꢀO bond of
the methoxyl group and catechol is formed as an intermediate
product. Metallic ruthenium and molybdenum carbides are the
active phases for deep deoxygenation to benzene, which re-
sults in complete oxygen removal. Significantly different deac-
tivation behaviour was observed over Mo-based and Ru-based
catalysts. While the Mo-based catalyst deactivates monotoni-
cally, the Ru-based catalyst shows a fast initial deactivation,
then reaches a pseudo-steady-state residual activity.
tured from guaiacol and Mo site is oxidised to Mo6+ Then, hy-
.
drogenation occurs, which leads to the formation of reaction
products and reduces Mo6+ species to Mo4+, and this com-
plete the catalytic cycle.
Therefore, the reaction mechanism of the HDO of guaiacol
over Mo/C catalysts could be envisaged. In this mechanism,
partially reduced MoOx species act as active sites for the direct
demethoxylation of guaiacol to form phenol. The resulting
phenol can be deoxygenated further on adjacent carbide sites
to produce benzene. However, because catechol adsorbs more
strongly on MoOx, the demethoxylation route is partially
blocked if catechol is co-fed. Thus, demethylation to catechol
followed by deoxygenation to phenol and then to benzene be-
comes the main pathway over carbide sites. In addition, owing
to the competitive catechol adsorption on carbide sites, the
phenol HDO reaction rate is affected, which results in lower
guaiacol conversion and higher phenol selectivity in co-feeding
experiments over the 10Mo/C catalyst.
Catalyst deactivation
Experimental Section
The deactivation behaviour was evaluated for Ru/C, 1Mo/C
and 10Mo/C catalysts during 6 h of time on stream (Figure 8).
Two types of deactivation patterns are observed. 1Mo/C and
10Mo/C catalysts deactivate monotonically during the entire
time on stream, whereas the deactivation of the Ru/C catalyst
shows a two-stage deactivation regime: 1) a sharp decrease in
activity is observed at the beginning of the reaction and 2) a
slow deactivation occurs, which reaches a pseudo-steady-state
conversion level. The most possible cause for deactivation is
coking, as reported by Delmon and Laurent.[22] Some tar-like
heavy products were also observed over the inert supporting
materials (e.g., quartz wool and glass bead) in the reactor.
However, there is not enough direct evidence to attribute the
deactivation exclusively to coke formation because a carbon-
based support was used in this work. Further investigations
are necessary to explore deactivation mechanisms over Ru/C
and Mo/C catalysts.
Catalyst preparation
Carbon-supported metal catalysts (Pd, Cu, Ru, Ir, Ni and Mo) were
prepared through incipient wetness impregnation. Before metal
impregnation, a commercial activated carbon used as a support
(NORIT R 3 EXTRA) was grinded and sieved into 20–40 mesh. Then,
the carbon support was treated with nitric acid by using a refluxing
method at 353 K overnight. Next, it was washed with deionised
water and dried at 383 K. Metal impregnation was performed by
using the aqueous solution of the active metal salts at appropriate
concentrations to yield the desired metal loadings. Palladium(II) ni-
trate, copper(II) nitrate trihydrate, ruthenium(III) acetylacetonate,
iridium(III) chloride hydrate, nickel(II) nitrate hexahydrate and am-
monium heptamolybdate tetrahydrate (supplied by Strem Chemi-
cals Inc.) were used as metal precursors for Pd, Cu, Ru, Ir, Ni and
Mo, respectively. After impregnation, the catalysts were kept at RT
for 4 h and then dried at 373 K overnight. Finally, the resulting cat-
alysts were calcined at 673 K in flowing N2 for 5 h.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
8
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
dwelling time. Energy correction was performed by using the C1s
peak of adventitious carbon at 284.5 eV as a reference. Reduced
samples were treated ex situ in flowing H2 at 973 K and then passi-
vated. With minimal exposure to air, samples were loaded in the
XPS reactor cell located inside the XPS pre-chamber and re-re-
duced in flowing H2 at 773 K. After re-reduction and evacuation,
samples were transferred to the main XPS chamber and XPS spec-
tra were recorded.
Catalytic activity and control reactions
The catalytic activity measurements were performed in a stainless
steel fixed-bed tubular reactor (outer diameter=0.5 in.). The reac-
tion was performed at temperatures ranging from 623 to 673 K
and 4 MPa. A H2/guaiacol molar ratio of 20 was used in all experi-
ments. The space–time, expressed as W/F, in which W is the mass
of the catalyst (g) and F is the feed flow rate (ghꢀ1), was varied sys-
tematically by changing either the catalyst loading or the feed
flow rate. Before the reaction, the catalyst was reduced in situ in
flowing H2 for 1 h at 673 K and 4.0 MPa, except for Mo/C catalysts,
which were reduced at 973 K. After reduction, the reactor was
cooled to the reaction temperature. Subsequently, guaiacol was
fed into the reactor. The reaction products were condensed in
a sample cylinder and collected within 1 h intervals. The collected
liquid samples were weighed and analysed with an Agilent 6890
gas chromatograph equipped with an HP-5 column. For all experi-
ments reported here, the average value of the mass balance, calcu-
lated as the ratio between the weight of the condensed liquid
product and the weight of the reactant fed into the reactor, was
85.6%. An Agilent 6890 gas chromatograph was used for product
identification. Product identification was verified by using stand-
ards. Moreover, an element-sensitive GC analysis of the liquid prod-
ucts was also performed with an Agilent 6890 gas chromatograph
equipped with a JAS atomic emission detector.
Powder XRD measurements were performed with a Bruker D8 Ad-
vance X-ray diffractometer equipped with a RINT2000 wide-angle
goniometer. Diffraction patterns were recorded within the range of
2q=5–708 (step size: 0.028) by using CuKa radiation and a power
of 40 kVꢁ40 mA. Reduced samples were treated ex situ in flowing
H2 at 973 K and then passivated before recording XRD patterns.
The FTIR spectra were recorded with a Bio-Rad Excalibur Series FTS
3000 system equipped with a mercury cadmium telluride detector
and a high-temperature DRIFT cell fitted with KBr windows. In
a typical experiment, the sample was treated initially at 673 K for
2 h in flowing He (UHP, 99.999%; flow rate: 20 mLminꢀ1). The spec-
tra recorded under He flow at ambient temperature were used as
reference. Next, guaiacol was introduced by dropping the liquid
(20 mL) on the powder sample. Then, the cell was evacuated at
30 mbar (3 kPa) to remove excess guaiacol until the spectra
became constant. The spectra were acquired with a resolution of
4 cmꢀ1. Typically, 256 scans were recorded, averaged and trans-
formed by using the Kubelka–Munk method, and the spectra re-
corded before introducing guaiacol was used as reference.
Conversion and product selectivity were calculated by using Equa-
tions (3a,b):
1ꢀCGWlp
ð3aÞ
ð3bÞ
Xð%Þ ¼
Sjð%Þ ¼
ꢁ 100
Wi
Acknowledgements
Cj
P
ꢁ 100
Cj
This work has been financially supported by the Thematic Strate-
gic Research Program (grant no. 0921390037) from the Science
and Engineering Research Council (SERC), Agency for Science,
Technology, and Research (A*STAR), Singapore.
in which Cj is the concentration of the reaction product j measured
by GC, CG is the guaiacol concentration in the liquid product, Wi is
the weight of the reactant fed during the sampling interval and
Wlp is the weight of the liquid product.
Keywords: bio-oil
·
guaiacol
·
hydrodeoxygenation
·
Control reactions for guaiacol and anisole were performed over ac-
tivated carbon under common conditions used in catalyst evalua-
tions (673 K, 4.0 MPa, W/F=0.067 h). The conversion values were
15.1 and 4.6% for guaiacol and anisole, respectively. Catechol
(33.6%) and phenol (15.5%) were observed as the dominant prod-
ucts in the guaiacol control experiment, whereas in the anisole
control reaction, phenol (54.7%) was the main product.
molybdenum · ruthenium
[1] a) C. Amen-Chen, H. Pakdel, C. Roy, Biomass Bioenergy 2001, 79, 277–
287; b) E. Berl, Sciences 1944, 99, 309–330; c)
D. Mohan, C. Pittman, Jr., P. Steele, Energy Fuels 2006, 20, 848–889.
[2] a) Y. Sheu, R. Anthony, E. Soltes, Fuel Process. Technol. 1988, 19, 31–50;
b) S. Czernik, A. Bridgwater, Energy Fuels 2004, 18, 590–598; c) D. Resas-
co, S. Crossely, AIChE J. 2009, 55, 1082–1089.
[3] a) M. Fatih Demirbas, Appl. Energy 2009, 86, S151–S161; b) L. Rollmann,
J. Catal. 1977, 46, 243–252.
Catalyst characterisation
[4] a) I. GraÅa, F. Ribeiro, H. Cerqueira, Y. Lam, M. Almeida, Appl. Catal. B
2009, 90, 556–563; b) E. Furimsky, Appl. Catal. A 2000, 199, 147–190;
c) D. Elliott, Energy Fuels 2007, 21, 1792–1815; d) A. Ausavasukhi, T.
Sooknoi, D. Resasco, J. Catal. 2009, 268, 68–78.
[5] a) E. Laurent, B. Delmon, Appl. Catal. A 1994, 109, 77–96; b) Y. Romero,
F. Richard, S. Brunet, Appl. Catal. B 2010, 98, 213–223; c) R. Nava, B. Pa-
welec, P. CastaÇo, M. ꢂlvarez-Galvꢃn, C. Loricera, J. Fierro, Appl. Catal. B
2009, 92, 154–167; d) W. Baldauf, U. Balfanz, M. Rupp, Biomass Bioener-
gy 1994, 7, 237–244.
Catalyst reducibility was determined by using the TPR technique.
TPR measurements were performed with a Thermo Scientific
TPROD 1100. The calcined catalyst (50 mg), thermally treated under
Ar atmosphere at 773 K to remove water and other contaminants,
was heated from 303 to 1123 K (heating rate: 10 Kminꢀ1) under
a 5% H2/Ar mixture (flow rate: 50 mLminꢀ1) and kept at 1123 K for
15 min while the hydrogen consumption was monitored
continuously.
[6] a) S. Zhang, Energy Sources 2003, 25, 57–65; b) M. Samolada, W. Bal-
dauf, I. Vasalos, Fuel 1998, 77, 1667–1675.
[7] A. Oasmaa, E. Leppꢄmmꢄki, P. Koponen, J. Levander, E. Tapola, “Applica-
tion of Standard Fuel Oil Analyses”, VTT Publication 306, 1997.
[8] C. Fisk, T. Morgan, Y. Ji, M. Crocker, C. Crofcheck, S. Lewis, Appl. Catal. A
2009, 358, 150–156.
XPS was used to investigate the surface composition of the cata-
lyst. The XPS spectra were recorded on a Thermo ESCALAB 250
spectrometer using AlKa radiation (hn=1486.6 eV). Measurements
were performed with 20 eV pass energy, 0.1 eV step size and 0.1 s
[9] D. Elliott, T. Hart, Energy Fuels 2009, 23, 631–637.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&
9
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[10] A. Centeno, A. David, C. Vanbellinghen, R. Maggi, B. Delmon, Develop-
ments in Thermochemical Biomass Conversion (Eds.: A. V. Bridegwater,
D. G. Boocock), Blackie Academic and Scientific, London, UK, 1997,
pp. 589–601.
[11] a) J. Wildschut, F. Mahfud, R. Venderbosch, H. Heeres, Ind. Eng. Chem.
Res. 2009, 48, 10324–10334; b) C. Zhao, Y. Kou, A. Lemonidou, X. Li, J.
Lercher, Angew. Chem. 2009, 121, 4047–4050; Angew. Chem. Int. Ed.
2009, 48, 3987–3990.
Z. Song, T. Cai, J. Hrbek, Top. Catal. 2006, 39, 257; c) X. Li, M. Ding, X.
Bao, Chin. J. Catal. 2008, 29, 884; d) J. Lee, M. Boudart, Catal. Lett. 1993,
20, 97.
[18] A. Yong, H. Borg, L. Ijzendoorn, W. Soundant, V. Beer, J. Veen, J. Nie-
mantsverdriet, J. Phys. Chem. 1993, 97, 6477–6483.
[19] a) R. Barthos, A. Szꢅchenyi, ꢂ. Koꢆs, F. Solymosi, Appl. Catal. A 2007,
327, 95–105; b) L. Leclercq, M. Provost, H. Pastor, J. Grimblot, A. Hardy,
L. Gengembre, G. Leclercq, J. Catal. 1989, 117, 371–383.
[12] a) L.Thompson, P. Savage, D. Briggs, Novel catalysts for upgrading coal-
derived liquids. Final technical progress report, DOE/PC/92537-T12,
1995; b) T. Xiao, A. York, K. Coleman, J. Claridge, J. Sloan, J. Charnock,
M. Green, J. Mater. Chem. 2001, 11, 3094–3098.
[13] Q. Bu, H. Lei, A. Zacher, L. Wang, S. Ren, J. Liang, Y. Wei, Y. Liu, J. Tang,
Q. Zhang, R. Ruan, Biores. Tech. 2012, 124, 470–477.
[20] A. Afifi, J. Hindermann, E. Chornet, R. Overend, Fuel 1989, 68, 498.
[21] J. Bredenberg, M. Huuska, J. Raty, M. Koiwio, J. Catal. 1982, 77, 242.
[22] E. Laurent, B. Delmon, Appl. Catal. 1994, 109, 97.
[23] A. Centeno, E. Laurent, B. Delmon, J. Catal. 1995, 154, 288.
[24] K. Kustin, S. Liu, J. Am. Chem. Soc. 1973, 95, 2487.
[25] A. Popov, E. Kondratieva, J. Goupil, L. Mariey, P. Bazin, J. Gilson, A. Tra-
vert, F. Mauge, J. Phys. Chem. C 2010, 114, 15661.
[26] C. R. Filley, C. Roth, J. Mol. Catal. A-Chem. 1999, 139, 245–252.
[14] A. Gutierrez, R. Kaila, M. Honkela, R. Slioor, A. Krause, Catal. Today 2009,
147, 239–246.
[15] a) L. Feng, X. Li, D. Dadyburjor, E. Kugler, J. Catal. 2000, 190, 1–3; b) T.
Bhaskar, K. Reddy, C. Kumar, M. Murthy, K. Chary, Appl. Catal. A 2001,
211, 189–201.
Received: February 4, 2013
Revised: March 22, 2013
[16] P. Koopman, A. Kieboom, H. Bekkum, J. Catal. 1981, 69, 172.
[17] a) D. Mordenti, D. Brodzki, G. Djega-Mariadassou, J. Solid State Chem.
1998, 141, 114; b) E. Kugler, C. Clark, J. Wright, D. Dadyburjor, J. Hanson,
Published online on && &&, 0000
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 10
&10
&
ÞÞ
These are not the final page numbers!