Carbon nanofiber supported transition-metal carbide catalysts for the hydrodeoxygenation of guaiacol

Article abstract of DOI:10.1002/cctc.201300280

Hydrodeoxygenation (HDO) studies over carbon nanofiber-supported (CNF) W₂C and Mo₂C catalysts were performed on guaiacol, a prototypical substrate to evaluate the potential of a catalyst for valorization of depolymerized lignin streams. Typical reactions were executed at 55bar hydrogen pressure over a temperature range of 300-375°C for 4h in dodecane, using a batch autoclave system. Combined selectivities of up to 87 and 69% to phenol and methylated phenolics were obtained at 375°C for W₂C/CNF and Mo₂C/CNF at >99% conversion, respectively. The molybdenum carbide-based catalyst showed a higher activity than W₂C/CNF and yielded more completely deoxygenated aromatic products, such as benzene and toluene. Catalyst recycling experiments, performed with and without regeneration of the carbide phase, showed that the Mo₂C/CNF catalyst was stable during reusability experiments. The most promising results were obtained with the Mo₂C/CNF catalyst, as it showed a much higher activity and higher selectivity to phenolics compared to W₂C/CNF. Get off O′Me: Hydrodeoxygenation of guaiacol over W₂C and Mo₂C supported on carbon nanofibers was studied. These catalysts display high activities and selectivities towards phenol and cresols in comparison to previously reported CoMo/Al₂O₃ catalysts. Catalyst recycling experiments showed that the Mo₂C/CNF catalyst is stable under reaction conditions and can be reused without further treatment (CNF=carbon nanofiber).

Full text of DOI:10.1002/cctc.201300280

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300280
Carbon Nanofiber Supported Transition-Metal Carbide
Catalysts for the Hydrodeoxygenation of Guaiacol
Anna L. Jongerius, Robert W. Gosselink, Jelmer Dijkstra, Johannes H. Bitter,
Pieter C. A. Bruijnincx, and Bert M. Weckhuysen*[a]
Hydrodeoxygenation (HDO) studies over carbon nanofiber-sup-
ported (CNF) W2C and Mo2C catalysts were performed on
guaiacol, a prototypical substrate to evaluate the potential of
a catalyst for valorization of depolymerized lignin streams. Typ-
ical reactions were executed at 55 bar hydrogen pressure over
a temperature range of 300–3758C for 4 h in dodecane, using
a batch autoclave system. Combined selectivities of up to 87
and 69% to phenol and methylated phenolics were obtained
at 3758C for W2C/CNF and Mo2C/CNF at >99% conversion, re-
spectively. The molybdenum carbide-based catalyst showed
a higher activity than W2C/CNF and yielded more completely
deoxygenated aromatic products, such as benzene and tolu-
ene. Catalyst recycling experiments, performed with and with-
out regeneration of the carbide phase, showed that the Mo2C/
CNF catalyst was stable during reusability experiments. The
most promising results were obtained with the Mo2C/CNF cata-
lyst, as it showed a much higher activity and higher selectivity
to phenolics compared to W2C/CNF.
Introduction
The limited availability of fossil fuels combined with the ad-
verse effects of their use necessitates the search for renewable
feedstocks such as lignocellulosic biomass for the sustainable
production of fuels and chemicals. An important component
of lignocellulosic biomass is lignin, a three-dimensional amor-
phous network of cross-linked phenylpropane units, optionally
substituted with methoxy and hydroxyl functional groups.[1]The
highly aromatic structure of lignin makes it an interesting re-
source for the production of aromatic chemicals. Bulk aromat-
ics production from lignin requires (catalytic) depolymerization,
for which various routes have been explored including catalyt-
ic hydrogenation or hydrogenolysis,[2–9] oxidation,[10,11] hydroly-
sis,[12–15] and reforming.[16,17] Alternatively, lignin can first be
converted in a thermochemical (catalytic) pyrolysis process to
give a bio-oil.[18–23] For both the more dedicated catalytic lignin
depolymerization or general bio-oil production processes, the
product streams will consist of a mixture of oxygen-rich aro-
matic monomers and oligomers. Further deoxygenation of
these crude feeds is often desired to yield higher-value chemi-
cals such as phenolics or BTX (BTX=benzenes, toluenes and
xylenes) for example, through subsequent hydrodeoxygena-
tion (HDO). This requires the removal of the hydroxyl and me-
thoxy oxygen-containing functional groups that are typically
present in depolymerized lignin streams. Guaiacol, which con-
tains both of these functional groups and is commonly report-
ed as one of the main products of lignin depolymerization pro-
cesses, is often used as the model compound of choice for
such HDO studies. Model compound studies with guaiacol can
be used to assess the selectivity and efficiency of a HDO cata-
lyst or process for the conversion of real depolymerized lignin
streams to either phenolics, full HDO products (BTX) or ring hy-
drogenation products such as cyclohexane.[1,24–26]
Sulfided CoMo and NiMo catalysts or supported noble metal
catalysts are most commonly employed in HDO reactions. Ini-
tially, most studies on the HDO of lignin or lignin model com-
pounds also typically focused on the use of sulfided CoMo and
NiMo on alumina catalysts.[1,27–29] We previously reported, for
example, a systematic comparison of the HDO selectivity and
activity of CoMo/Al2O3 in reactions with guaiacol and various
other lignin-derived molecules, reactions which generally re-
sulted in the formation of (alkyl)phenols as the main HDO
product.[24] However, deactivation of the catalyst if no sulfur is
added to the feed and contamination of the products by
leaching of sulfur are general problems associated with this
class of catalysts. In addition, selective HDO over these cata-
lysts is hampered by the acidity of the support, as isomeriza-
tion reactions on Lewis acid sites lead to a broad product dis-
tribution even from a single model compound and coke for-
mation is accelerated during the reaction.[24] Furthermore, at
the typical elevated reaction temperatures alumina is known
to be unstable in the presence of water, which can be expect-
ed to be present in biomass-derived feedstocks.[30]
[a] A. L. Jongerius,+ R. W. Gosselink,+ J. Dijkstra, Dr. J. H. Bitter,
Dr. P. C. A. Bruijnincx, Prof. B. M. Weckhuysen
Utrecht University, Inorganic Chemistry and Catalysis
Debye Institute for Nanomaterials Science
Universiteitsweg 99, 3584 CG Utrecht (The Netherlands)
Fax: (+31)30-251-1027
Noble metal catalysts are commonly presented as a more
stable alternative for the traditional sulfided CoMo and NiMo
catalysts. However, oxygen removal over these catalysts is
often accompanied by complete hydrogenation of the aromat-
E-mail: B.M.Weckhuysen@uu.nl
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300280.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2964

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
ic ring, which is clearly undesirable if phenolics or BTX are tar-
geted as end products. Indeed, cyclohexanol and cyclohexane
were shown to be major products of the HDO of guaiacol over
supported Pt, Rh, Pd, and Ru catalysts.[28,29,31–33] The formation
of aromatic HDO products was observed for the conversion of
vaporized guaiacol and anisole over Pt/Al2O3. This process,
however, resulted in the formation of a large amount of differ-
ent products including oxygen-containing and oxygen-free hy-
drogenation products.[34–36]
carbon supports for the carbides is that catalyst preparation
does not require carburization gases. Furthermore, carbon
nanofibers (CNF) are ideally suited as support for molybdenum
or tungsten carbides, because of their high surface area and
mesoporosity whereas the microporosity of activated carbons
can cause diffusion limitations. Also, the CNF support is ex-
pected to be stable and inert under reaction conditions, even
in the presence of water, whereas activated carbons can con-
tain impurities originating from the original carbon source
used for their preparation.[50,51]
Alternative catalyst materials for the HDO of lignin-derived
feeds should preferably combine the relatively strong HDO
and weak hydrogenation properties of the traditional HDS cat-
alyst with the stability of noble metal hydrogenation catalysts.
Transition metal phosphides, nitrides, and particularly carbides
can potentially provide such an alternative. It has been demon-
strated that not only molybdenum sulphides, but also carbides
and nitrides of molybdenum and other transition metals show
good activity in various hydrotreating reactions.[37,38] In the
field of catalytic biomass conversion, bulk tungsten carbides
have already been used for lignin depolymerization[39] and for
cellulose degradation when supported on activated carbon.[40]
Recently, it was shown by some of us that CNF-supported
tungsten carbides combine high activity in the HDO of fatty
acids with low hydrogenation activity, resulting in the produc-
tion of high amounts of linear unsaturated hydrocarbons.[41]
Bulk molybdenum carbide was shown to be active in the HDO
of C2 and C3 oxygenates[42] and supported molybdenum car-
bide on carbon nanotubes and activated carbon were used by
Han et al. for the HDO of vegetable oils to diesel fuels.[43,44]
In addition, several bulk and supported molybdenum nitride
catalysts have been reported for the HDO of model com-
pounds that mimic lignin-derived products. Indeed, already in
1997, supported molybdenum nitrides and carbides were
shown to be active in the HDO of benzofuran to ethylben-
zene.[45] The HDO of guaiacol over molybdenum nitrides has
been reported more recently and high selectivities towards
phenol were obtained.[46–49] For molybdenum nitrides, Mo2N
was reported to be the most active phase whereas catalyst
prepared with other Mo/N ratios were less active.[46] In contrast
to the CoMo/Al2O3-catalyzed reactions that are reported to
proceed by consecutive demethylation/dehydroxylation
through catechol to phenol, the HDO of guaiacol over Mo2N is
thought to take place by a direct demethoxylation pathway
without a catechol intermediate. Demethylation of guaiacol to
catechol was only observed with Al2O3-supported molybde-
num nitrides, as a result of a support-catalyzed conversion.[48]
Importantly, the Mo2N catalyst generally showed a better con-
version of guaiacol than commonly used CoMo/Al2O3 catalysts
and yielded less ring-hydrogenation products, such as cyclo-
hexene and cyclohexane.[1]
Here we report our studies on the applicability of group 6
metal carbides for the selective HDO of guaiacol, which serves
as a model for the upgrading of depolymerized lignin feeds,
with the aim of obtaining phenolics. The conversion of guaia-
col over CNF-supported tungsten and molybdenum carbides
at different temperatures shows that high selectivities towards
the mono-oxygenated products phenol and cresol can be ob-
tained. In addition, spent catalysts are analyzed and recycled
(with and without additional heat treatment) to demonstrate
that in the case of molybdenum no substantial deactivation of
the catalyst takes place under the applied reaction conditions.
Results and Discussion
Catalyst characterization
The X-ray diffraction (XRD) patterns of the fresh 15 wt%
W2C/CNF and 7.5 wt% Mo2C/CNF catalysts are shown in
Figure 1, as is the pattern for bare CNF for comparison. The
graphite-like reflections of the CNF support are clearly ob-
served at 31, 51 and 648 2q, corresponding to (002), (101), and
(004) reflections.[52]
Both the W2C and Mo2C phases show main reflections at
similar angles, that is, 40, 44 and 468 2q corresponding to the
(100), (002), and (101) lattice planes, due to their structural re-
semblance. In both cases an oxide phase (WOx or MoOx, with
2<x<3) is also present as determined from the shoulder peak
at 418 2q.[52] This can be explained by partial oxidation of the
metal carbide upon exposure to air. From XRD analysis it is
shown that the observed oxide phase does not increase after
prolonged exposure to air. The observed line broadening of
the carbide reflections is similar for both metal carbide cata-
lysts, resulting in comparable average particle sizes of 4 nm, as
calculated using the Scherrer equation (see also Table 1).[53]
This is in accordance with statistical calculations on transmis-
sion electron microscopy (TEM) images (Table 1).
Table 1. Physicochemical characteristics of CNF-supported metal carbide
catalysts under study.
Notably, no studies have yet been reported that use either
bulk or supported molybdenum or tungsten carbides for the
HDO of guaiacol.[28,29] The results obtained with the fatty acids,
however, clearly show that the CNF-supported transition metal
carbide catalysts hold some promise also for selective hydro-
deoxygenation of guaiacol, as they combine high HDO with
low hydrogenation activity. A particular advantage of using
Particle size
N2-physisorption
Pore volume
[cm3 gÀ1
TEM
[nm]
XRD
[nm]
BET surface area
[m2 gÀ1
]
]
CNF
Mo2C/CNF
W2C/CNF
–
5
5
–
4
4
190
120
111
0.40
0.34
0.30
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2965

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 1. a) Full X-ray diffraction patterns and b) zoom-in X-ray diffraction patterns of bare CNF, W2C/CNF, and Mo2C/CNF. Reflections are given for graphitic
carbon (*), XOx (&) and X2C (~), with X=W or Mo.[52]
To study the influence of the carbothermal reduction pro-
Table 2. Conversion, selectivity to phenol and cresols, and mass balance
for the HDO of guaiacol without catalyst, over heat-treated CNF and over
cesses on the CNF support, N2-physisorption was performed.
The BET surface area and pore volume results are depicted in
Table 1 for bare CNF, Mo2C/CNF, and W2C/CNF (BET=
Brunauer–Emmett–Teller). A significant decrease in the BET sur-
face area and total pore volume is observed after heat treat-
ment, which indicates a structural change in the CNF structure.
No evidence for any structural change can be seen in TEM
images or XRD diffraction patterns, however; previous studies
with W2C/CNF also showed a decrease of the BET surface area
upon an increase of the temperature of the carbothermal pro-
cess, which could be attributed to, e.g., partial collapse or
plugging of the pores.[41]
Mo2C/CNF and W2C/CNF at 3508C and approximately 55 bar H2 pressure
after 4 h.
Conversion
[%]
Product class
selectivity [%]
Mass balance
[%]
a)
b)
c)
d)
e)
No catalyst
CNF
W2C/CNF
Mo2C/CNF
40
40
66
7
8
11
12
13
<1
<1
<1
2
2
3
4
9
4
4
2
2
68
70
76
74
7
46
45
>99
a) phenol; b) o-cresol, p-cresol, and dimethylphenol isomers; c) benzene
and toluene; d) anisole, methylated anisole, and dimethoxybenzene;
e) cyclohexane, cyclohexene, and cyclohexanone.
Catalytic Activity
Reactions performed in the absence of a catalyst resulted in
a 40% conversion of guaiacol. A small amount was converted
to phenol and cresols with a total selectivity of 14%, but the
majority of products could not be detected by GC analysis
(mass balance 68%, Table 2). A blank reaction with only bare
CNF showed the same guaiacol conversion of 40%, and selec-
tivities for phenol (7%) and cresol (11%). Low concentrations
of methoxylated and hydrogenated products were detected
with a total selectivity of 6%. Under identical conditions the re-
actions catalyzed with both W2C/CNF and Mo2C/CNF resulted
in significantly higher conversions, selectivity towards phenol-
ics and mass balances (Table 2). Combination of the product
yields obtained by GC analysis resulted in mass balances of
76% for W2C/CNF and 74% for Mo2C/CNF-catalyzed reactions
at 3508C. An increase in reaction temperature to 3758C result-
ed in higher mass balances for both catalysts (96% (W2C) and
91% (Mo2C)); the reactions at temperatures other than 3508C
are discussed below. The blank reactions also give low mass
balances, which can only be caused, in this case, by uncata-
lyzed thermal reactions. The results for the catalyzed reaction
point in the same direction, as recovery of the catalyst did not
show any increase in weight of the catalyst and no solids were
A
typical HDO reaction of guaiacol over W2C/CNF and
Mo2C/CNF was performed in dodecane at 3508C with 55 bar
hydrogen pressure for 4 h. The main products observed in all
runs were phenol, o-cresol, and p-cresol; other products were
obtained in smaller amounts with a total selectivity of less
than 10%. All products can be grouped into classes according
to extent and type of HDO: 1) phenol; 2) the cresols o- and p-
cresol and dimethylphenol isomers; 3) full HDO products ben-
zene and toluene; 4) products still containing methoxy func-
tionalities (anisole, methylated anisole and dimethoxyben-
zene); 5) ring-hydrogenation products (cyclohexane, cyclohex-
ene and cyclohexanone). In the Mo2C/CNF-catalyzed reactions,
larger amounts of group 3–5 products were generally formed,
with a relative increase in the fractions of full HDO and ring-hy-
drogenation products compared to the W2C/CNF-catalyzed re-
actions. The exact composition of all product mixtures can be
found in Tables 2 and S1 of the Supporting Information. It
should be noted that the conversion levels over W2C/CNF vary
between 66–82% for a typical 4 h reaction at 3508C; the ratio
between different products was always found to be constant,
however.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2966

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
formed on the reactor wall, indi-
cating that the 10–30% weight
loss that occurred during the re-
action cannot be attributed to
the formation of large amounts
of coke. Indeed,
a common
problem of the use of guaiacol
as substrate is that guaiacol con-
densation reactions result in
soluble higher molecular weight
products that cannot be detect-
ed by our GC methods, thus
lowering the mass balance.
Guaiacol is, for instance, known
Scheme 1. HDO of guaiacol gives phenol and cresols as the main products.
to undergo a radical-induced re-
arrangement leading to the for-
mation of oligomers and ultimately coke. This rearrangement
is initiated by OMe bond homolysis, but takes place only at
temperatures above 4008C.[54] Condensation products can
nonetheless also be formed at lower temperatures, as a result
of direct ring coupling as was shown by Bui et al. in the HDO
mildly acidic guaiacol and phenols. Similar to what was report-
ed before for the HDO of guaiacol over CoMo/Al2O3, phenol is
much more stable under the applied conditions than guaiacol,
indicating that the removal of the second oxygen to form ben-
zene is more difficult[24] and is not favorable when guaiacol is
still present. A reaction with phenol as the substrate over the
W2C/CNF catalyst at 3508C indeed showed a much lower con-
version of 28%, with a mass balance of 82% and benzene as
the major product with 30% selectivity.
1
of guaiacol over CoMo/Al2O3 and MoS2.[55] In our case, H NMR
analysis of a reaction mixture with a 85% mass balance also
shows that at least 5% more aromatic products are present in
the product mixture than can be identified by GC analysis.
Analysis of the reaction mixture obtained after a run without
catalyst (mass balance 68%, Table 1) shows a new singlet at
2.2 ppm, indicating the presence of a large amount of uniden-
tified aromatic ring-methylated products. Similar products
were again reported by Bui et al. after guaiacol HDO over
CoMo/Al2O3 and MoS2.[55] We therefore attribute the observed
loss in mass balance to the formation of condensed, higher
molecular weight products. Although the formation of such
higher molecular weight products cannot be easily prevented,
their contribution can be considerably reduced by speeding
up guaiacol conversion, as the phenol and cresol products are
much more stable against self-condensation.
To optimize the reaction conditions, reactions were per-
formed at different reaction temperatures (300–3758C) for 4 h
at approximately 55 bar hydrogen pressure. Results of the tem-
perature optimization are shown in Figure 2. For the W2C/CNF-
catalyzed reactions, conversion and selectivity towards phenol-
ics increased with reaction temperature. Full conversion was
obtained at 3758C after 4 h with a selectivity of 66% to phenol
and 21% to cresols. The Mo2C/CNF catalyst showed a much
higher activity with almost full conversion (95%) obtained
after 4 h already at 3258C. The total selectivity for phenolics
(53+17% for phenol and cresols, respectively) is lower with
Mo2C/CNF than with W2C/CNF, and a higher total selectivity
(17%) to other products, particularly benzene (8%) and anisole
(6%) is observed for the former at 3758C after 4 h. Ring hydro-
genation activity of the Mo2C/CNF catalyst also increased at
higher temperatures. As mentioned previously, higher temper-
atures result in better mass balances for both catalysts with
a maximum of 96% and 91% at 3758C for W2C/CNF and
Mo2C/CNF, respectively. Notably, the mass balances of >90%
obtained for the reactions at 3758C are among the highest re-
ported for guaiacol HDO reactions. The selectivities towards
phenol, or more generally, phenolics are much higher than
typically obtained with traditional HDS catalysts, as illustrated
by our previously reported CoMo/Al2O3-catalyzed HDO of
guaiacol at comparable substrate concentration and hydrogen
pressure. For this system, the highest phenolics yields were ob-
tained at 3008C after 4 h, amounting to a selectivity to phenol
of only 34% and 11% to cresols at 84% conversion. A broad
range of methylated and demethylated products were also
formed, including catechol (11%) and methylated guaiacol and
catechol (8%).[24] The use of noble metal catalysts also general-
ly results in lower phenolics yields, as they show much higher
Notably, no catechol was observed in any of the reaction
mixtures, indicating that the conversion of guaiacol to phenol
does not involve sequential demethylation (DME) and HDO
steps (as is commonly observed for sulfided CoMo/Al2O3), but
rather a direct demethoxylation pathway (DMO; Scheme 1).[24]
This is in accordance with results reported for bulk and
carbon-supported Mo2N-catalyzed HDO of guaiacol, although
small amounts of catechol were still found with the nitride cat-
alysts. Most likely the high temperatures that were used
during the carburization step resulted in removal of all mildly
Brønsted acidic sites on the CNF. Although Lewis acid sites on
alumina-supported catalysts are responsible for DME activity,
the CNFs used here can be considered inert under our reaction
conditions, as the applied heat treatment removes the relevant
functional groups from the support.[56] Indeed, blank reactions
performed in the presence of heat-treated CNF give almost
identical results as blank reactions run without any catalyst or
support present (Table 2). The small amounts of cresols and
other methylated products that are formed during the reaction
probably arise from a transalkylation reaction catalyzed by the
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2967

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 3. Conversion (^), selectivity to phenol (&), cresols (~), and mass
balance (*) for the HDO of guaiacol over a) W2C/CNF and b) Mo2C/CNF at
3508C and approximately 55 bar H2 pressure, followed over time.
Figure 2. Conversion, selectivity to phenol and cresols, and mass balance for
the HDO of guaiacol over a) W2C/CNF and b) Mo2C/CNF at approximately
55 bar H2 pressure for 4 h at different temperatures (& 3008C, & 3258C,
& 3508C, & 3758C.).
The results show that Mo2C/CNF is more active than
W2C/CNF, but also shows a higher selectivity to the full HDO
products benzene and toluene and the methoxylated product
anisole. The selectivity to full HDO products increases with re-
action time, with a maximum selectivity of 4.8% observed
after 6 h reaction. The higher activity of Mo2C/CNF also result-
ed in a ring hydrogenation selectivity (7%) after 6 h that was
higher than observed for W2C/CNF (2%). This is in line with
a previous observation that that tungsten carbides are less
active olefin hydrogenation catalysts than molybdenum car-
bides in the HDO of unsaturated fatty acids.[57]
ring hydrogenation than HDO activity and produce mainly cy-
clohexanol and eventually cyclohexane.[31,32]
A Mo2N catalyst supported on activated carbon showed
a similar phenol selectivity at 3008C in guaiacol HDO as the
Mo2C/CNF catalyst presented here at 3008C.[49] However, signif-
icantly lower conversion levels were obtained whilst all reac-
tion parameters were comparable.[46] The total selectivity for
mono-oxygenated aromatics was also lower with the activated
carbon-supported Mo2N catalyst as no cresols or anisole were
obtained. Demethylation activity, illustrated by the small quan-
tities of catechol that were detected, was furthermore reported
for these supported nitride catalysts.
Catalyst recycling
The conversion of guaiacol over time at 3508C and selectivi-
ty towards different products are shown for both catalysts in
Figure 3. The difference in activity between the two catalysts is
evident, as initial reaction rates of 0.16 and 0.64 molgmetalÀ1 hÀ1
were obtained for W2C/CNF and Mo2C/CNF, respectively. Full
conversion of guaiacol over Mo2C/CNF was reached after 2 h
while guaiacol conversion over W2C/CNF is still increasing after
6 h. Phenol selectivity is found to increase initially, but leveled
off when a conversion of around 60% was reached for both
catalysts. In the Mo2C/CNF-catalyzed reaction the phenol selec-
tivity decreased after 6 h in favor of full HDO and hydrogena-
tion products (see also Table S1 in the Supporting Information).
In both reactions the mass balance decreased sharply to
around 80% after the first hour, after which it levelled off.
W2C/CNF and Mo2C/CNF catalyst stability was tested in two
ways. A standard reaction was first performed on a freshly pre-
pared catalyst (run 1). After reaction, the spent catalyst was re-
trieved, rinsed with diethyl ether and reused directly without
any reactivation step (run 2). Hereafter, the spent catalyst of
run 2 was recarburized and again tested in a catalytic reaction
(run 3). All runs were performed at 3508C for 4 h and the cata-
lysts were analyzed by XRD after every run (Figures 4 and 5).
For the Mo2C/CNF catalyst, an additional recycling experiment
has been performed with reactions of 70 min in order to
obtain a conversion level comparable to W2C/CNF after 4 h.
The recovery of each catalyst after every run was close to
100% and reaction volumes and guaiacol concentrations were
in every run scaled to the amount of catalyst recovered to
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2968

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 4. a) Full X-ray diffraction patterns and b) zoomed in X-ray diffraction patterns of fresh W2C/CNF (fresh), spent W2C/CNF (run 1), reused W2C/CNF
(run 2), spent W2C/CNF after reactivation at 10008C (regenerated), and spent reactivated W2C/CNF (run 3). Reflections are given for graphitic carbon (*),
XO2 (&), and X2C (~), with X=W or Mo.[52]
Figure 5. a) Full X-ray diffraction patterns and b) zoomed in X-ray diffraction patterns of fresh Mo2C/CNF (fresh), spent Mo2C/CNF (run 1), reused
Mo2C/CNF (run 2), and reactivated Mo2C/CNF after reactivation at 9008C (regenerated) and reactivated Mo2C/CNF after reaction (run 3). Reflections are given
for graphitic carbon (*), XO2 (&), and X2C (~), with X=W or Mo.[52]
allow for a proper comparison of the runs. Catalytic data for
both catalysts is presented in Table 3.
ic test with the spent W2C/CNF catalyst (without any further
treatment, run 2) showed an increase in guaiacol conversion
from 66 to 80% and increased selectivities towards all prod-
ucts. Reused Mo2C/CNF showed similar conversions and selec-
tivities in run 2 for both 4 h and 70 min reactions, as could be
expected from the unchanged XRD pattern and high activity
of this catalyst.
Remarkably, the 70 min reaction with the molybdenum cata-
lyst resulted in a much higher phenol selectivity, 62% com-
pared to 46% after 4 h, and higher mass balances (92% com-
pared to 76%. Differences are seen in the XRD pattern of the
W2C/CNF catalyst after run 1 (Figure 4). A clear increase in the
diffraction peak attributed to WOx is observed for the spent
W2C/CNF catalyst, whereas for Mo2C/CNF, both after the
70 min and 4 h reactions, it does not change significantly
(Figure 5). The tungsten-based catalyst therefore seems more
prone to oxidation than the molybdenum carbide. The catalyt-
It was already noted above that the conversion levels of typ-
ical guaiacol HDO experiments catalyzed by freshly prepared
batches of W2C/CNF show some variation, that is, ranges be-
tween 66 and 82% in multiple experiments; conversion levels
for Mo2C/CNF, on the other hand, showed little variation. The
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2969

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
er, the XRD data does not show
such an increase for our cata-
lysts. Furthermore, our results
are based on 4–9 nm particles,
which are significantly smaller
than the bulk particles tested in
literature. Remarkably, although
no changes in the Mo2C/CNF
XRD pattern could be observed
after the recarburization treat-
ment, guaiacol conversion drop-
ped from 68 to 51% for the
70 min reaction with a concomi-
tant drop in phenol selectivity.
The characterization data none-
theless shows the carbide phase
of Mo2C/CNF to remain rather
stable during the reactions and
the carburization treatment.
Table 3. Conversion, selectivity to various products, and mass balance for the HDO of guaiacol over W2C/CNF
and Mo2C/CNF (run 1), the reused (run 2) and reactivated catalysts (run 3), at 3508C and approximately 55 bar
H2 pressure for 4 h or 70 min.
Conversion [%]
Product class selectivity [%]
Mass balance [%]
a)
b)
c)
d)
e)
W2C/CNF
run 1
run 2
4 h
66
81
48
46
51
36
12
15
12
1
1
1
4
7
2
2
2
3
76
80
79
run 3
Mo2C/CNF
run 1
run 2
4 h
>99
>99
>99
46
45
45
13
16
15
2
4
5
9
10
8
2
7
7
74
81
79
run 3
Mo2C/CNF
run 1
run 2
70 min
67
68
51
62
56
50
13
12
12
1
1
1
6
10
8
2
2
2
92
87
86
run 3
a) phenol; b) o-cresol, p-cresol, and dimethylphenol isomers; c) benzene and toluene; d) anisole, methylated
anisole, and dimethoxybenzene; e) cyclohexane, cyclohexene, and cyclohexanone.
increased conversion observed in the second run for the tung-
sten catalyzed reaction could therefore simply be the result of
this observed fluctuation in conversion level. A second set of
W2C/CNF recycling experiments, however, again showed
a higher activity in run 2 with the W2C/CNF catalyst. This indi-
cates that this increased activity of the spent catalyst could
also arise from a structural change that takes place under reac-
tion conditions. It is very likely that the formation of an oxide
phase plays an important role in these reactions, however,
more research is needed before the exact role and nature can
be revealed. These experiments are beyond the scope of this
paper.
Both catalysts show activity loss after the recarburization
step, which for the W2C/CNF can be attributed to an increase
in particle size. The decrease in activity of the Mo2C/CNF cata-
lyst after recarburization indicates, however, that other
changes also must take place in the material during the heat
treatment. For example, encapsulation of the carbide particles
in the carbon support or irreversible coke formation could be
responsible for the observed catalyst deactivation. Notably, no
loss in catalytic activity and selectivity was observed for the
molybdenum catalyst after washing and reusing the catalyst
without further heat treatment, though (Table 3, run 2, 70 min).
This shows that recarburization is not required for the Mo2C/
CNF catalyst, but is essential for the W2C/CNF catalyst to recov-
er the carbide phase. Indeed, the Mo2C/CNF catalyst proved to
be rather stable under reaction conditions and can be reused
for several runs without difficulty.
XRD analysis of the catalyst after run 2, but before recarburi-
zation showed again an increase in the WOx phase for the
tungsten catalyst, whereas no change in the reflections for the
molybdenum catalyst was observed. This unexpected differ-
ence between the stability towards oxidation of both catalysts
cannot be readily explained. Although the Gibbs free energy
for oxidation of a (bulk) W2C phase is more negative than for
Mo2C (as calculated by HSC chemistry for Windows), the differ-
ence is not significant enough to explain the observed dissimi-
larity in catalyst stability. A possible explanation would be that
it is caused by a kinetic factor, such as the diffusion of oxygen
through the particle.
Conclusions
High conversions and high selectivities towards phenolics are
obtained in the hydrodeoxygenation of guaiacol over W2C/CNF
and Mo2C/CNF at 300–3758C and approximately 55 bar hydro-
gen pressure. At higher temperatures, higher conversions, and
selectivities to the desired products are obtained. Mo2C/CNF is
more active than W2C/CNF, but also leads to the formation of
more full HDO and methoxylated products. Longer reaction
times and higher temperatures result in more of the full HDO
products benzene and toluene. Reactions with low catalytic ac-
tivity lead to the formation of higher molecular weight methy-
lated aromatics, thus lowering the mass balance. W2C/CNF is
sensitive to oxygen and the small amount of WOx phase de-
tected after reaction had a significant influence on catalytic ac-
tivity. After recarburization the W2C phase could be restored
completely, however, sintering of the particles occurred, result-
ing in loss of catalytic activity. In contrast, Mo2C/CNF showed
excellent stability under reaction conditions and could be
reused without further treatment and without loss of catalytic
activity. Compared to previously reported results with
Subsequently, both catalysts were subjected to a recarburiza-
tion treatment at 10008C and 9008C for W2C/CNF and
Mo2C/CNF, respectively. The X-ray diffractograms show that the
carbide phase was restored for the tungsten-based catalyst;
the thermal treatment did, however, lead to a clear increase in
particle size to 9 nm that corresponds to the size of the
formed WOx crystallites, which also were 9 nm. This increase in
particle size has a clear effect on the catalytic activity, with
a 20% decrease in conversion in run 3 with respect to run 1,
a 10% drop in selectivity for phenol and a sudden increase in
hydrogenation selectivity to 3.4%. Interestingly, larger carbide
particle sizes have been reported to show higher activity in hy-
drogenation and HDS reactions, which is attributed to a relative
increase in the ratio of the {111}/{200} crystal planes.[27] Howev-
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2970

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Sigma) as internal standard and 32 g of the solvent dodecane
(99%, Acros). A fresh batch of catalyst was used for every run. The
reactor was purged for 5 min with nitrogen (5.0) at a flow of
CoMo/Al2O3, the tungsten and molybdenum carbides show
much higher selectivities towards phenolics and lower (de)me-
thylation activity. Their low ring-hydrogenation activity makes
them excellent catalyst for the production of phenolics from
depolymerized lignin streams.
100 mLminÀ1 and 5 min with hydrogen (5.0) at
a flow of
100 mLminÀ1 while stirring at 600 rpm. The reactor was then filled
with hydrogen with a flow of 100 mLminÀ1 while heating to the
desired temperature (usually 3508C) with a heating ramp of
58CminÀ1, typically leading to a hydrogen pressure between 50
and 60 bar during the reaction. When reaction temperature was
reached, the stirring speed was increased to the maximum stirring
speed of 1000 rpm to minimize any gas liquid mass transfer limita-
tions. The catalytic reaction was carried out for 4 h. The reaction
was stopped by cooling and release of pressure, after which the re-
action mixture was diluted with an equal volume of diethyl ether
in order to dissolve all products. The solid catalyst was removed by
filtration and washed with diethyl ether.
Experimental Section
Catalyst synthesis
A 5 wt% Ni/SiO2 growth catalyst was prepared by homogeneous
deposition precipitation using 30 g silica (Aerosol 300, Degussa),
7.9 g Ni(NO3)2 (Acros Organics), and 4.85 g of urea (Acros Organics)
at 908C for 18 h in 1.5 L demineralized water, as reported earlier.[58]
5 g Ni/SiO2 was then reduced at 7008C (ramp 58CminÀ1 in N2 flow)
in a H2/N2 flow for 2 h. Fishbone-type CNFs were obtained by flow-
ing a CO/H2/N2 mixture (266:102:450 mLminÀ1) over this catalyst at
5508C (ramp 58CminÀ1 in N2/H2 flow) for 24 h. The crude product,
approximately 35 g, was collected and refluxed three times for 1 h,
with intermediate washing with deionized water, in 400 mL boiling
1.0m aqueous KOH (Acros) to remove SiO2. Subsequently, the
fibers were treated with 400 mL boiling, concentrated HNO3 (65%,
Merck) for 1.5 h in order to remove exposed nickel and introduce
oxygen-containing functional groups. The acid-treated fibers are fi-
nally washed with demineralized water until the pH of the filtrate
was neutral.
Product analysis
The reaction mixture was analyzed using a Varian 430-GC gas chro-
matograph equipped with a VF-5 ms capillary column and a FID
detector. Compounds were identified with gas chromatography
coupled with a mass spectrometer using a Shimadzu QP2010
GCMS instrument equipped with a VF-5 ms capillary column and
compared with retention times of pure standards. Response factors
of guaiacol and phenol relative to hexadecane were determined
experimentally. 1H NMR measurements were performed on the
pure reaction mixture with a few drops of deuterated chloroform
on a Varian 400 MHz NMR spectrometer.
The CNF were dried under vacuum at 808C for 4 h prior to impreg-
nation. For the tungsten-based catalysts, the support was impreg-
nated using an ammonium metatungstate solution (66.5 wt% W,
Aldrich) by pore volume (0.7 mLgÀ1 determined by water uptake)
to arrive at a 15 wt% loading. After impregnation the sample was
kept under vacuum at 808C for 24 h. Subsequently a heat treat-
ment was applied for 3 h at 10008C under N2 atmosphere to yield
W2C/CNF. Mo2C/CNF was also prepared by pore volume impregna-
tion, using an ammonium molybdate (>99%, Acros) solution to
arrive at a 7.5 wt% loading, followed by a heat treatment at 9008C
under N2 atmosphere.[41] Different weight loadings were used to
arrive at similar molar loadings of the carbide catalysts.
Acknowledgements
This research has been performed within the framework of the
CatchBio program. The authors gratefully acknowledge the sup-
port of the Smart Mix Program of the Netherlands Ministry of
Economic Affairs and the Netherlands Ministry of Education, Cul-
ture and Science.
Keywords: carbides · guaiacol · hydrodeoxygenation · lignin ·
phenols
Catalyst characterization
N2-physisorption isotherms were recorded with a Micromeritics
Tristar 3000 at 77 K. The samples were dried prior to performing
the measurement for at least 16 h at 473 K in a N2 flow. The sur-
face area was determined using the Brunauer–Emmett–Teller (BET)
theory.[59] The total pore volume was defined as the single-point
pore volume at p/p0 =0.95. X-ray powder diffraction (XRD) patterns
were obtained on a Bruker-AXS D2 Phaser powder X-ray diffrac-
tometer using Co Ka1,2 with l=1.79026 ꢁ. Measurements were
carried out between 10 and 1008 2q using a step size of 0.098 2q
and a scan speed of 1 s. Bright-field TEM images were obtained on
an FEI Tecnai 12, operated at 120 keV.
[1] J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, B. M. Weckhuysen, Chem.
Rev. 2010, 110, 3552–3599.
[2] J. S. Shabtai, W. W. Zmierczak, E. Chornet, US patent 6172272 B1, 2001.
[3] W. L. Schinski, A. E. Kuperman, J. Han, D. G. Naae, US patent 2009/
0218061A1, 2009.
[4] R. W. Thring, J. Breau, Fuel 1996, 75, 795–800.
[5] R. J. A. Gosselink, W. Teunissen, J. E. G. van Dam, E. de Jong, G. Geller-
stedt, E. L. Schott, J. P. M. Sanders, Bioresour. Technol. 2012, 106, 173–
177.
[6] J. Y. He, C. Zhao, J. A. Lercher, J. Am. Chem. Soc. 2012, 134, 20768–
20775.
[7] W. Xu, S. J. Miller, P. K. Agrawal, C. W. Jones, ChemSusChem 2012, 5,
667–675.
[8] C. Li, M. Zheng, A. Wang, T. Zhang, Energy Environ. Sci. 2012, 5, 6383–
6390.
[9] Q. Song, F. Wang, J. Cai, Y. Wang, J. Zhang, W. Yu, J. Xu, Energy Environ.
Sci. 2013, DOI: 10.1039/c2ee23741e.
[10] H. Werhan, J. M. Mir, T. Voitl, P. P. van Rohr, Holzforschung 2011, 65, 703–
709.
Catalytic testing
The catalytic reactions were performed in a 100 mL stainless steel
high pressure Autoclave Engineers batch autoclave reactor. The
temperature was monitored using a thermocouple and stirring was
performed using a magnetic driver equipped with an impellor. In
a typical reaction, the autoclave was loaded with 200 mg of the
catalyst material, 1.7 g guaiacol (Sigma), 0.6 g hexadecane (>99%,
[11] K. Stꢂrk, N. Taccardi, A. Bosmann, P. Wasserscheid, ChemSusChem 2010,
3, 719–723.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2971

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[12] M. Nagy, K. David, G. J. P. Britovsek, A. J. Ragauskas, Holzforschung 2009,
63, 513–520.
[37] S. T. Oyama, Handbook of Heterogeneous Catalysis, 1st ed., Wiley, 1996,
Chapter 2.3.14, pp. 342–356.
[13] V. M. Roberts, V. Stein, T. Reiner, A. Lemonidou, X. Li, J. A. Lercher, Chem.
Eur. J. 2011, 17, 5939–5948.
[14] J. M. Lavoie, W. Barꢃ, M. Bilodeau, Bioresour. Technol. 2011, 102, 4917–
4920.
[15] A. Toledano, L. Serrano, J. Labidi, J. Chem. Technol. Biotechnol. 2012, 87,
1593–1599.
[16] J. J. Zakzeski, A. L. Jongerius, P. C. A. Bruijnincx, B. M. Weckhuysen,
ChemSusChem 2012, 5, 1602–1609.
[38] E. Furimsky, Appl. Catal. A 2003, 240, 1–28.
[39] S. Kotrel, M. Emmeluth, A. Bençhr, WO patent WO/2009/037281, 2009.
[40] N. Ji, T. Zhang, M. Zheng, A. Wang, H. Wang, X. Wang, J. G. Chen,
Angew. Chem. 2008, 120, 8638–8641; Angew. Chem. Int. Ed. 2008, 47,
8510–8513.
[41] R. W. Gosselink, D. R. Stellwagen, J. H. Bitter, Angew. Chem. 2013, 125,
5193–5196; Angew. Chem. Int. Ed. 2013, 52, 5089–5092.
[42] H. Ren, W. Yu, M. Salciccioli, Y. Chen, Y. Huang, K. Xiong, D. G. Vlachos,
J. G. Chen, ChemSusChem 2013, 6, 798–801.
[17] J. J. Zakzeski, B. M. Weckhuysen, ChemSusChem 2011, 4, 369–378.
[18] P. F. Britt, A. C. Buchanan, M. J. Cooney, D. R. Martineau, J. Org. Chem.
2000, 65, 1376–1389.
[43] J. Han, J. Duan, P. Chen, H. Lou, X. Zheng, H. Hong, ChemSusChem
2012, 5, 727–733.
[19] M. Misson, R. Haron, M. F. A. Kamaroddin, N. A. S. Amin, Bioresour. Tech-
nol. 2009, 100, 2867–2873.
[44] J. Han, J. Duan, P. Chen, H. Lou, X. Zheng, H. Hong, Green Chem. 2011,
13, 2561–2568.
[20] P. de Wild, R. V. der Laan, A. Kloekhorst, E. Heeres, Environ. Prog. Sustain-
able Energy 2009, 28, 461–469.
[45] G. M. Dolce, P. E. Savage, L. T. Thompson, Energy Fuels 1997, 11, 668–
675.
[21] P. R. Patwardhan, R. C. Brown, B. H. Shanks, ChemSusChem 2011, 4,
1629–1636.
[22] E. Dorrestijn, L. J. J. Laarhoven, I. W. C. E. Arends, P. Mulder, J. Anal. Appl.
Pyrolysis 2000, 54, 153–192.
[23] Z. Ma, E. Troussard, J. A. van Bokhoven, Appl. Catal. A 2012, 423–424,
130.
[24] A. L. Jongerius, R. Jastrzebski, P. C. A. Bruijnincx, B. M. Weckhuysen, J.
Catal. 2012, 285, 315–323.
[46] C. Sepffllveda, K. Leiva, R. carcia, L. R. Radovic, I. T. Ghampson, W. J. De-
Sisto, J. L. Garcia Fierro, N. Escalona, Catal. Today 2011, 172, 232–239.
[47] I. T. Ghampson, C. Sepulveda, R. Garcia, B. G. Frederick, M. C. Wheeler, N.
Escalona, W. J. DeSito, Appl. Catal. A 2012, 413–414, 78–84.
[48] I. Tyrone Ghampson, C. Sepulveda, R. Garcia, J. L. Carcia Fierro, N. Esca-
lona, W. J. DeSisto, Appl. Catal. A 2012, 435–436, 51–60.
[49] I. T. Ghampson, C. Sepulveda, R. Garcia, L. R. Radovic, J. L. Carcia Fierro,
W. J. DeSisto, N. Escalona, Appl. Catal. A 2012, 439–440, 111–124.
[50] V. Jimꢃnez, P. Sµnchez, P. Panagiotopoulou, J. L. Valverde, A. Romero,
Appl. Catal. A 2010, 390, 35–44.
[25] Y.-C. Lin, C. L. Li, H. P. Wan, H. T. Lee, C. F. Liu, Energy Fuels 2011, 25,
890–896.
[26] F. de Miguel Mercader, M. J. Groeneveld, S. R. A. Kersten, C. Geantet, G.
Goussaint, N. W. J. Way, C. J. Schaverien, K. J. A. Hogendoorn, Energy En-
viron. Sci. 2011, 4, 985–997.
[51] J. H. Bitter, J. Mater. Chem. 2010, 20, 7312–7321.
[52] ICDD PDF-2 database package http://www.icdd.com/products/
pdf2.htm.
[27] E. Furimsky, Appl. Catal. A 2000, 199, 147–190.
[28] H. Wang, J. Male, Y. Want, ACS Catal. 2013, 3, 1047–1070.
[29] I. GraÅa, J. M. Lopes, H. S. Cerquiera, M. F. Ribeiro, Ind. Eng. Chem. Res.
2013, 52, 275–287.
[30] R. M. Ravenelle, J. R. Copeland, W. G. Kim, J. C. Crittenden, C. Sievers,
ACS Catal. 2011, 1, 552–561.
[31] C. R. Lee, J. S. Yoon, Y. W. Suh, J.-W. Choi, J.-M. Ha, D. J. Suh, Y.-K. Park,
Catal. Commun. 2012, 17, 54–58.
[32] A. Gutierrez, R. K. Kaila, M. L. Honkela, R. Slioor, A. O. I. Krause, Catal.
Today 2009, 147, 239–246.
[53] A. L. Patterson, Phys. Rev. 1939, 56, 978–982.
[54] M. Asmadi, H. Kawamoto, S. Saka, J. Anal. Appl. Pyrolysis 2011, 92, 88–
98.
[55] N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Appl. Catal. B 2011, 101,
239–245.
[56] R. W. Gosselink, R. van den Berg, W. Xia, M. Muhler, K. P. de Jong, J. H.
Bitter, Carbon 2012, 50, 4424–4431.
[57] S. A. W. Hollak, R. W. Gosselink, J. H. Bitter, D. S. van Es, unpublished re-
sults.
[58] M. L. Toebes, M. K. Van der Lee, L. M. Tang, M. H. Huis in t Veld, J. H.
Bitter, J. Van Dillen, K. P. De Jong, J. Phys. Chem. B 2004, 108, 11611–
11619.
[33] C. Zhao, J. A. Lercher, ChemCatChem 2012, 4, 64.
[34] T. Nimmanwudipong, R. C. Runnebaum, D. E. Block, B. C. Gates, Catal.
Lett. 2011, 141, 779–783.
[59] S. Brunauer, P. H. Emmett, E. Teller, J. Am. Chem. Soc. 1938, 60, 309–319.
[35] R. C. Runnebaum, T. Nimmanwudipong, D. E. Block, B. C. Gates, Catal.
Sci. Technol. 2012, 2, 113–118.
Received: April 14, 2013
Published online on July 16, 2013
[36] T. Nimmanwudipong, R. C. Runnebaum, D. E. Block, B. C. Gates, Energy
Fuels 2011, 25, 3417–3427.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2964 – 2972 2972