You are on page 1of 14

Catalysis

Science &
Technology
View Article Online
PAPER View Journal
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

Carbon nanofiber-supported ReOx catalysts for


the hydrodeoxygenation of lignin-derived
Cite this: DOI: 10.1039/c5cy01992c
compounds†
I. Tyrone Ghampson,*a Catherine Sepúlveda,b Rafael García,b José L. G. Fierroc and
Néstor Escalona*ad

The effect of ReOx loading (2–13 wt%) and H2 pressure (0–5 MPa) for the hydrodeoxygenation of phenol
has been studied for carbon nanofiber-supported ReOx catalysts in a batch reactor at 573 K. Characteriza-
tion of the supports and catalysts has been obtained from N2 physisorption, TPD, FTIR, XRD, potentiometric
titration, TPR and XPS measurements, which revealed the presence of a crystalline and surface ReOx phase
whose particle size and surface coverage increased with loading. The reactivity of the catalysts was linked
to the in situ partial reduction of ReO3 to form Re4+ and Re6–7+ sites, whose presence and relative amounts
were determined by post-reaction XPS analysis. The reaction rate increased with ReOx loading up to 10
wt%, attributed to the increase in Re surface coverage; a decrease in reaction rate at higher loading was as-
cribed to the formation of aggregates. The study revealed a strong affinity for direct cleavage of the C–O
Received 19th November 2015, bond to form benzene. The similar relative abundance of the Re species is responsible for the similar trend
Accepted 26th January 2016
in product distribution of the catalysts. The dependence of activity and product distribution with respect to
H2 pressure has been related to kinetics and thermodynamics. The reactivity of the best catalyst for the
DOI: 10.1039/c5cy01992c
HDO of guaiacol (2-methoxyphenol), anisole (methoxybenzene), phenol and o-cresol (2-methylphenol)
www.rsc.org/catalysis further demonstrated the catalyst's preference for C–O bond scission.

1. Introduction properties are associated with the high level of water and oxy-
genated compounds in the feed.1 Hence, it is important to re-
The need to reduce the overreliance on fossil fuels and reduce duce the high oxygen content.
greenhouse gas (GHG) emissions provides an impetus to the Much of the research aimed at addressing this issue has
development of renewable energy. The recent interest in lig- focused on catalytic hydrodeoxygenation (HDO). There has
nocellulosic biomass in the literature underscores how bio- been an upsurge in the research on catalyst development in
mass has emerged as a viable alternative to conventional HDO over the past decade using model compounds, which
crude oil for the production of hydrocarbon fuels and value- has led to several classes of materials being demonstrated to
added chemicals. Among the choices of conversion technolo- be capable of catalytically removing oxygen in the liquid, va-
gies, the fast pyrolysis of woody biomass to produce bio-oil pour and aqueous phases.2 Lignin-derived monomers have
has become an important strategy despite significant barriers received the most attention in these studies because they
that limit its wide implementation, such as high viscosity, low have been identified to be present in significant quantities,
stability, corrosivity, low energy content, etc. These deleterious and are among the most recalcitrant compounds, in bio-
oil.3–5 In particular, phenol and substituted phenols such as
a
Departamento de Ingeniería Química y Bioprocesos, Escuela de Ingeniería, guaiacol, anisole and cresol are commonly studied due to
Pontificia Universidad Católica de Chile, Avenida Vicuña Mackenna 4860, Macul, their intransigence to selective deoxygenation and hence are
Santiago, Chile. E-mail: isaactghampson@udec.cl; Fax: +56 41 2245374;
utilized to probe the reactivities of catalysts and their tenden-
Tel: +56 41 2203353
b
Universidad de Concepción, Facultad de Ciencias Químicas, Casilla 160c,
cies in steering the reaction towards either ring hydrogena-
Concepción, Chile tion or direct deoxygenation.6
c
Instituto de Catálisis y Petroleoquímica, CSIC, Cantoblanco, 28049 Madrid, Studies on conventional hydrotreating sulfide catalysts were
Spain the initial focus on HDO reactions of lignin-derived com-
d
Facultad de Ciencias Químicas, Pontificia Universidad Católica de Chile,
pounds and were instrumental in laying the groundwork for
Santiago, Chile. E-mail: neescalona@ing.puc.cl; Fax: +56 2 3547962;
Tel: +56 2 3547962
future studies.7–11 Since then, meaningful strides have been
† Electronic supplementary information (ESI) available. See DOI: 10.1039/ made in several studies detailing the reactivity,5,12 mecha-
c5cy01992c nism13 and kinetics.14 A common theme in these studies is

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

the good selectivity of sulfide catalysts to aromatic hydrocar- These intriguing findings provide an incentive and a template
bons—including benzene, toluene, xylene, ethylbenzene— to expand this investigation into evaluating a possible struc-
which is at the heart of HDO reactions because of the ability ture–function relationship which will add insight into the de-
to cleave the carbon–oxygen bond and the prevention of com- sign of effective catalysts. In this study, a series of carbon nano-
plete saturation of CC bonds. However, the use of sulfide fiber (CNF)-supported ReOx catalysts with different loadings was
catalysts is complicated by the addition of a sulfiding agent to synthesized, characterized by N2 physisorption, TPD, FTIR,
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

replenish stripped sulfur active sites, leading to the risk of XRD, potentiometric titration, TPR and XPS, and tested with
contamination of naturally sulfur-free liquid. phenol HDO in a batch reactor. CNF was utilized as a support
There has been significant interest in exploring non- because it boasts properties not available to conventional
sulfide catalysts due to the deficiencies of sulfide catalysts. supports (Al2O3, SiO2, etc.), including minimizing the effect of
Noble metals, often used in conjunction with solid acid cata- the support on the active phase which will provide an opportu-
lysts (such as HZSM-5, HBEA, SO42−/ZrO2, and Nafion etc.), nity to observe the structural effects on catalytic HDO behaviour.
have proven extremely effective in HDO reactions of In addition, using a technique such as XPS to assess the struc-
phenolics.15–23 These studies reported complete conversion tural changes of metal oxide catalysts occurring during reactions
and high deoxygenation of lignin monomers, whereby the no- is significantly easier when supported on CNF than when oxidic
ble metals act as hydrogenation catalysts while the acid cata- supports are used.43 Carbon nanofiber has received scant atten-
lysts are responsible for dehydration. However, although no- tion in the literature for HDO reaction of phenolics: recent stud-
ble metal catalysts boast high activity, several challenges ies by Jongerius et al.5 and Santillan-Jimenez et al.44 on the
(their price, rarity, etc.) prevent their widespread use. HDO of guaiacol over CNF-supported metal carbides (Mo2C and
Transition metal carbide,24,25 nitride26 and phosphide cat- W2C) have demonstrated their viability.
alysts27,28 hold promise as a cheaper alternative to noble The HDO studies of the best ReOx/CNF catalyst was ex-
metal catalysts. Recent studies have demonstrated their capa- tended by testing at different H2 pressures for the conversion
bilities for catalysing HDO of phenolics to aromatic hydrocar- of phenol, as well as on a library of model compounds
bons.29 Many challenges, however, stand in the way of this (guaiacol, anisole, o-cresol and phenol) to further illustrate
less-expensive alternative, the most significant of which are the effectiveness of this catalyst for HDO of lignin-derived
the severe preparation conditions and their poor stability compounds.
during regeneration.
Nickel-based catalysts have recently made inroads into HDO 2. Experimental
reactions driven by their high hydrogenation activity and low 2.1. Sample preparation
cost.30–37 Studies on guaiacol,32,34 phenol,30 anisole,35,36 etc.
have found these catalysts to be highly active and selective to hy- A pyrolytically-stripped carbon nanofiber (CNF, PR-24-XT-PS,
drogenated products, including cyclohexane, methylcyclohexane, SBET = 43 m2 g−1, >98% carbon basis) was purchased from
etc. Recently, similar to bifunctional noble metal-based catalysts, Sigma-Aldrich and treated with HNO3 (65%, Merck) at 363 K
Ni catalysts have often been paired with acid catalysts such as for 16 h under vigorous stirring (10 g of CNF to 500 mL of
HZSM-5 to augment its activity towards saturated hydrocar- acid). Then, the mixture was cooled down to room tempera-
bons.38,39 Nevertheless, the drawbacks of using Ni-based cata- ture, filtered and extensively rinsed with distilled water. The
lysts are their hydrogen inefficiency as a result of their excessive sample was dried in air at 383 K for 24 h. The oxidized sup-
saturation ability and their questionable stability due to port is denoted CNFox.
sintering under HDO conditions. The supported rhenium oxide catalysts were prepared by
The results of these studies exemplify the need to improve wetness impregnation of an aqueous solution with requisite
the capabilities and performance of these catalysts still further amounts of ammonium perrhenate (≥99%, NH4ReO4, Sigma-
or explore other classes of catalysts. Reducible transition metal Aldrich) to obtain a range of ReOx loading (2–13 wt%). After
oxides represent an attractive alternative to the other catalytic impregnation, the samples were kept at room temperature
systems. Their reactivity is essentially linked to the partial reduc- for 24 h, dried at 383 K for 15 h, and then calcined in static
tion of surface metal oxide species under reaction conditions. air at 573 K (ramp rate of 2 K min−1) for 0.5 h. The samples
Recent studies have demonstrated their adeptness at directly are referred to as yReOx/CNFox, where y indicates the ReOx
cleaving C–O bonds in lignin derivatives, an important capabil- loading in wt%.
ity due to its potential impact on hydrogen efficiency.40–42 In ad-
dition, metal oxide catalysts can be regenerated by calcination,40 2.2. Sample characterization
and their emergence renders often costly pre-activation proce- Temperature-programmed decomposition (TPD) analyses of
dures such as sulfidation, reduction, carburization and the as-received and HNO3-treated CNF were carried out in a
phosphidation unnecessary. U-shaped quartz tube micro-reactor, placed inside a program-
In a recent study, ReOx/SiO2 catalyst has displayed high ac- mable electrical furnace. TPD profiles of CO, CO2 and H2O
tivity in the conversion of guaiacol and a high selectivity into ar- were obtained from room temperature to 1313 K, at 10 K
omatic hydrocarbons (benzene, toluene, and xylene, BTX), a min−1 and under a helium flow of 50 mL min−1 (99.995%,
valuable feedstock for downstream petrochemical processing.41 Linde). Evolution of desorbed gases was monitored by using

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016
View Article Online

Catalysis Science & Technology Paper

a thermal conductivity detector (TCD) coupled to a non- to a sum of Gaussian and Lorentzian lines (90G-10L) using a
dispersive infrared device. least squares minimization procedure for χ 2 with the help of
Electrophoresis measurement (EM) of the support was car- the XPS peak program. Relative surface atomic ratios were de-
ried out using a Zeta-Meter 3.0+ apparatus using 20 mg of termined from the corresponding peak areas, corrected with
sample suspended in 200 mL of a 1 × 10−3 mol L−1 KCl solu- tabulated sensitivity factors,46 with a precision of 7%.
tion. The pH was adjusted with either 0.1 mol L−1 HCl or
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

KOH solution. The IEP (isoelectric point) was obtained from 2.3. Catalyst testing
the plot of pH vs. zeta potential. The yReOx/CNFox catalysts were evaluated for phenol hydro-
FT-IR analyses of the supports were performed using a Ni- deoxygenation in a 300 mL stirred-batch autoclave set-up
colet Nexus FTIR in the wavenumber range (4000–400 cm−1) (Parr Model 4841) at 573 K and under a hydrogen pressure of
and with a scan of 64. The samples were prepared using a 1 : 3 MPa for 4 h. The liquid reactant consisted of 1.8 g of phe-
100 mg of CNF and KBr support. nol (99%, Merck) and 700 μL of hexadecane (as an internal
Acid site concentration and acid strength measurements standard, 99%, Sigma-Aldrich) dissolved in 80 mL of
of the support were determined using a potentiometric dodecane (as solvent, 99%, Merck). The catalysts were dried
method,45 whereby a suspension in acetonitrile (99.9%, in an oven at 383 K to remove physisorbed water prior to cat-
Merck) was titrated with n-butylamine (99%, Merck). The vari- alyst testing. This mixture and 200 mg of catalyst were loaded
ation in electric potential was registered using a Denver In- into the reactor, purged with a 100 mL min−1 flow of nitrogen
strument UltraBasic pH/mV meter. (99.995%, Linde) for 20 min to evacuate air from the system,
The BET specific surface area (SBET) and total pore volume heated to 573 K while stirring at 645 rpm, and charged with
(Vp) of the samples were obtained from nitrogen sorption iso- hydrogen (99.995%, Linde) to 3 MPa. The start of the reaction
therms at 77 K using a Micromeritics TriStar II 3020 instru- was assumed to be the time when the reaction temperature
ment. Prior to the measurements, the samples were degassed was reached and H2 was introduced into the reactor. The re-
under inert atmosphere at 473 K for 2 h. SBET was calculated action was punctuated by the addition of hydrogen to main-
from the adsorption branch in the range 0.05 ≤ P/P 0 ≤ 0.25 tain the pressure for the entire duration of the experiment.
and Vp was recorded at P/P 0 = 0.97. Liquid samples were periodically withdrawn during the
X-ray diffraction (XRD) patterns of the samples were course of the reaction and identified by using a Perkin Elmer
obtained using a Rigaku diffractometer equipped with Ni- GCMS-SQ8T (Clarus 680) gas chromatograph equipped with
filtered CuKα1 radiation (λ =1.5418 Å). The standard scan pa- an Elite-1 column (Perkin Elmer, 30 m × 0.32 mm, film thick-
rameters were 10–90° 2θ with a step size of 0.02° and a ness of 0.25 μm). The liquid samples were also quantified by
counting time of 0.4 s per step. Identification of the phases using a Perkin Elmer (Clarus 400) gas chromatograph
was achieved by reference to the ICDD files using EVA analy- equipped with a CP-Sil 5 column (Agilent, 30 m × 0.53 mm,
sis software. film thickness of 1.0 μm) and a FID. The products were also
H2-temperature programmed reduction (TPR) of the cal- identified by their column retention time in comparison with
cined catalysts was performed in a U-shaped quartz tube available standards.
micro-reactor, placed inside a programmable electrical fur- Phenol conversion as a function of H2 pressure of the best
nace and connected to a TCD detector. The effluent gas was yReOx/CNFox catalyst was also carried out at 573 K and at a
passed through a cold trap filled with a mixture of iso- pressure ranging from 0 to 5 MPa following the aforemen-
propanol and liquid nitrogen to remove the steam produced tioned procedure. The reaction at 0 MPa was conducted in
during the reduction before entering the TCD detector. Be- the absence of hydrogen and under 0.5 MPa of nitrogen.
fore the measurements, the sample (ca. 25 mg) was dried Finally, the HDO studies of the best yReOx/CNFox catalyst
overnight to remove weakly-adsorbed water molecules. Then, were extended by testing on lignin model compounds
the sample was heated in flowing 5 vol% H2/Ar (Linde) from (guaiacol, anisole, phenol and o-cresol) at 573 K and 5 MPa
ambient temperature to 1273 K at a heating rate of 10 K H2 pressure for 4 h. The substrates were guaiacol (98%,
min−1. Merck), anisole (99%, Merck), and o-cresol (99%, Merck).
X-ray photoelectron spectra of calcined and spent catalysts About 200 mg of the catalyst was added to the reactor
were recorded on a VG Escalab 200R electron spectrometer charged with 0.232 mol L−1 of the substrate dissolved in 80
using a Mg Kα (1253.6 eV) photon source. The spent catalysts mL of dodecane and 700 μL of hexadecane.
were removed from the reactor, stored in flasks containing The conversion of the substrate (XSUB) was calculated from
dodecane and then transferred to the pre-treatment chamber (eqn (1)):
of the spectrometer. The binding energies (BE) were
referenced to the C 1s level of the carbon support at 284.8
eV. An estimated error of ± 0.1 eV can be assumed for all (1)
measurements. The intensities of the peaks were calculated
from the respective peak areas after background subtraction where n0substrate is the initial moles of the substrate (typically
and spectrum fitting by the standard computer based statisti- phenol) in the solution (mol) and nisubstrate is the moles of the
cal analysis which included fitting the experimental spectra substrate at time i (mol). The initial reaction rate rs (mol g−1

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

s−1) was calculated from the initial slope of the conversion vs.
time plot b (s−1) according to (eqn (2)):

(2)

where m (g) is the mass of the catalyst. The intrinsic rate


Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

(molecPHEconverted at Re−1 s−1) was calculated using (eqn (3)):


Fig. 1 (a) TPD profiles and (b) FTIR of as-received CNF and nitric acid-
treated CNF.
(3)

where MWC is the molar mass of carbon (12 g mol−1); XReO3 Decomposition of these groups releases CO2 and is consis-
is the ReO3 loading (wt%) and (Re/C)XPS is the XPS surface tent with the presence of strong acid sites.51
atomic Re/C ratio (at./at.). The yield of product (Y) was calcu- The presence of these groups on CNF was supported by
lated from (eqn (4)): FT-IR analysis of the samples (Fig. 1b). Both the pristine and
the oxidized CNF displayed bands in the 1600–1500 cm−1 re-
(4) gion which correspond to a graphite structure,54 in addition
to bands in the 1500–1000 cm−1 region which can be attrib-
uted to the weak oxidation of CNF as a result of exposure to
where niproduct j is the moles of product j formed at time i ambient conditions.55 The oxidized CNF support also
(mol). The product selectivity was calculated using eqn (5): displayed bands at 1754 and 1042 cm−1, attributed to the
CO and C–O stretching vibration modes, respectively, in
(5) the carboxylic groups. These additional bands confirm that
oxidation introduced oxygen surface groups on the support,
where nproduct j is the moles of product j formed (mol) and in agreement with deductions from TPD results.
nsubstrate converted is the moles of substrate converted (mol). A Isoelectric point (IEP) results obtained from electrophore-
number of repeated runs under the same conditions were sis measurements showed an IEP at pH = 4.2 and 2.5 for the
performed to ensure satisfactory reproducibility of the data. as-received CNF and the oxidized CNF, respectively, indicative
of surface modification of CNF after oxidative treatment in
3. Results and discussion accordance with FTIR and TPD results. The lower IEP value
of CNFox can be attributed to an increase in surface acidity
3.1. Support characterization due to the oxidation of the functional groups on the
The carbon nanofiber support was oxidized with concen- support.56
trated HNO3 to decorate the inner and outer areas of the sup- The BET surface area and pore volume (Table 1) remained
port with oxygen-containing functional groups.47 In a recent essentially unchanged after oxidation, implying that textural
study, Ni catalysts supported on an as-received carbon nano- properties were not altered after oxidative treatment and that
tube (CNT) and an oxidized CNT (CNTox) have been com- the oxygen-containing surface groups were mostly introduced
pared, and it was found that the presence of surface oxygen on the surface and not on the pore walls and entrances.
groups enhanced the hydrogenolysis capacities of the Ni/ The surface acidity of the supports was estimated from po-
CNTox catalyst.48 Since CNT and CNF possess similar physi- tentiometric titration curves with n-butylamine as the probe
cal and chemical properties,49 we decided to pre-oxidize the molecule. The results include the maximum acid strength of
CNF support prior to ReOx impregnation. The methodology the surface sites (derived from the initial electrode potential,
used for the oxidation of CNF has already been reported to E0) and the acid site density (total number of acid sites nor-
be effective by Tessonnier et al.47 malized by the surface area). The pristine CNF and oxidized
The chemical nature of the oxygen functional groups of CNF presented weak acid sites (−82 mV) and very strong acid
the support was determined by means of TPD profiles shown
in Fig. 1. The untreated CNF displayed a broad peak and a
shoulder at high temperatures (1080–1180 K), while the oxi- Table 1 Textural and structural properties of materials under study

dized CNF displayed broad (1000–1170 K) shoulders. These Sample ReOx loading (wt%) BET (m2 g−1) Vp (cm3 g−1)
peaks typically arise from the decomposition of carbonyl and
CNF — 42 0.06
quinone groups, which leads to the desorption of CO, indi- CNFox — 45 0.06
cating the presence of weakly acidic, neutral and basic 2ReOx/CNFox 2 42 0.06
groups whose carbon atom is bonded to one oxygen 4ReOx/CNFox 4 39 0.06
6ReOx/CNFox 6 39 0.06
atom.50–52 In addition to these peaks, the oxidized CNF sup-
8ReOx/CNFox 8 39 0.06
port also exhibited a broad 545–670 K peak, which is indica- 10ReOx/CNFox 10 39 0.06
tive of the presence of lactonic and carboxylic groups.52,53 13ReOx/CNFox 13 39 0.06

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016
View Article Online

Catalysis Science & Technology Paper

sites (147 mV), respectively, according to the criterion pro- Table 2 XPS data of the calcined catalysts
posed by Cid and Pecchi.57 Additionally, the oxidized CNF Binding energy (eV) Surface atomic
had a higher amount of surface acid (37%). This acidity dif- (distribution/%) ratio (at%)
ference is due to the introduction of surface oxygen groups, Sample C1s Re 4f5/2 O1s O/C Re/C
in line with the findings of Tessonnier et al.47
CNF 284.8 (76)
XRD patterns of the pristine CNF (not shown) and the oxi- 286.3 (18) — 531.8 (37) 0.032 —
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

dized CNF (Fig. 2) show no observable differences, indicating 288.1 (6) 533.2 (63)
that the crystalline structure of the carbon material remained CNFox 284.8 (65)
286.3 (23) — 531.6 (41) 0.161 —
unchanged after oxidative treatment. This is not surprising
288.3 (12) 533.1 (59)
considering the similarities of the textural parameters of the 2ReOx/CNFox 284.8 (66)
two samples. The supports showed peaks indexed as 286.3 (25) 45.5 531.6 (38) 0.145 0.0022
graphite-like carbon (ICDD file no. 01-075-0444) and carbon 288.4 (8) 533.1 (62)
4ReOx/CNFox 284.8 (67) 531.1 (26)
(ICDD file no. 01-080-0017). There were some unidentified
286.3 (23) 45.7 532.2 (32) 0.144 0.0042
peaks which are possibly associated with the metal catalyst 288.4 (10) 533.3 (42)
used for the growth of the fibres. 6ReOx/CNFox 284.8 (61) 531.1 (38)
XPS analysis was used to further assess the effect of the 286.3 (26) 45.6 532.3 (23) 0.159 0.0051
288.4 (13) 533.4 (39)
oxidative treatment on the surface of the CNF support. It can 8ReOx/CNFox 284.8 (62) 531.1 (36)
be seen in Table 2 that the C 1s binding energies (BE) of the 286.3 (29) 45.7 532.3 (22) 0.151 0.0061
parent CNF and oxidized CNF consisted of three peaks with 288.5 (9) 533.3 (42)
BEs at 284.8 eV, 286.3 eV and 288.2 eV ± 0.1 eV, and the O 1s 10ReOx/CNFox 284.8 (56) 531.2 (39)
286.3 (29) 45.7 532.4 (24) 0.166 0.0077
peak consisted of two peaks at 531.7 eV ± 0.1 eV and 533.2 eV 288.2 (15) 533.5 (37)
which are all close to the binding energies previously 13ReOx/CNFox 284.8 (57) 531.1 (40)
reported for this type of material:58 the C 1s BE at 284.8 eV 286.3 (30) 45.7 532.3 (24) 0.159 0.0098
represents graphitic carbon, consistent with the XRD data; 288.3 (13) 533.5 (36)
the C 1s peak at 286.3 eV and the O 1s signal at 531.7 eV cor-
respond to the C–O bonds in the phenolic or ether groups;
the C 1s BE at 288.2 eV and the O 1s BE at 533.2 eV are con- IEP and FTIR results that the surface of CNF become
sistent with carboxyl and/or ester groups. The relative peak O-enriched after nitric acid oxidation.
intensities (shown in parentheses) indicate the predomi-
nance of graphitic carbon on the surface of all the samples.
Furthermore, the relative concentration of the C 1s BE corre- 3.2. Characterization of the metal oxide
sponding to carboxyl and/or ester groups doubled after nitric Textural characterization results of the catalysts are summa-
acid oxidation which is consistent with the characterization rized in Table 1. The BET specific surface areas of the cata-
data discussed earlier. The clearest evidence, however, of the lysts ranged from 39 to 42 m2 g−1. The decrease in BET sur-
effect of the oxidation treatment can be found in the XPS O/C face area after impregnation and subsequent calcination was
atomic ratio in Table 2: the oxidized CNF support showed a minimal, indicating that pore blockage was not significant
marked 5-fold increase in O/C atomic ratio compared with and that most of the ReOx nanoparticles are located on the
the pristine CNF, supporting the interpretation from TPD, outer surface of the support.
The XRD patterns of the support and calcined catalysts
are shown in Fig. 2. The patterns of all the catalysts showed
peaks indexed as graphite-like carbon (ICDD file no. 01-075-
0444) and carbon (ICDD file no. 01-080-0017), indicating that
the structural integrity of the CNFox support remained intact
after impregnation of NH4ReO4 and calcination. All the cata-
lysts except the lowest loaded 2ReOx/CNFox catalyst exhibited
diffraction lines corresponding to hexagonal ReO3 crystals
(ICDD card no. 00-040-1155). The absence of ReO3 diffraction
peaks in the 2ReOx/CNFox catalyst suggests that it contains
small crystallites of Re oxide below the detection limit or that
the reflections of CNF may have masked the low-intensity Re
oxide peaks. The catalysts with low Re loading—4ReOx/CNFox
and 6eOx/CNFox—displayed a barely discernible characteristic
peak for ReO3 at 2θ = 16.5°, 30.1° and 34.7°. These peaks be-
came more well defined with increasing loading due to the
formation of an increased amount of ReO3. At high loadings,
Fig. 2 X-ray diffraction patterns of calcined ReOx/CNFox catalysts. new reflections of ReO3 emerged with the most prominent at

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

2θ = 25.3°, 34.7°, 40.0°, 50.8° and 52.1°; the appearance of interacting with the CNF support surface. This suggests the
the reflection at 2θ = 25.3° confirms that the support peak at presence of two types of Re species on the catalysts. A frac-
2θ = 26.1° masked some of the ReO3 peak at lower loadings. tion of the higher temperature peak could be due to pyrolysis
The average ReOx crystallite size (nm) of the three high- or gasification of the carbon support to produce CO.59 The
loaded catalysts was calculated from the Scherrer equation TPR profile of the bare CNFox support show no H2 consump-
using the ReOx reflection at 2θ = 16.5°; the crystallite size for tion peak in the temperature range of 550–590 K, suggesting
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

the catalysts with lower loadings was not calculated because that peak I observed for the catalysts is associated with the
the peak could not be satisfactorily resolved from the back- reduction of ReOx species. However, the CNFox TPR profile
ground. Instrumental contribution to line broadening was did reveal a broad peak centred at 1060 K attributed to gasifi-
accounted for by measuring a silicon standard (NIST SRM cation of the CNFox support. There are some other notable
640e) using an identical optical configuration. The particle observations in the TPR profiles. The relative intensity of
size increased in the order 8ReOx/CNFox (20.5 nm) < 10ReOx/ peak I to peak II changed with Re loadings. In fact, the
CNFox (21.7 nm) < 13ReOx/CNFox (24.1 nm), in line with the 2ReOx/CNFox catalyst only presented the higher temperature
well-documented increase in particle size with loading. peak which implies that the majority of the ReOx species
The reduction behaviour as a function of ReOx content were in close contact with the CNFox support, consistent with
was examined by TPR as shown in Fig. 3. The TPR profiles of the XRD data in Fig. 2 which indicated that ReOx species in
CNF-supported ReOx catalysts show mainly two H2 consump- that catalyst were well dispersed on the support. At higher
tion peaks: a peak (I) observed in the 550–590 K temperature ReOx content (≥4 wt%), peak I developed with increasing Re
range for all catalysts, except 2ReOx/CNFox, may be assigned loading and the reduction maximum slightly shifted to a
to the reduction of an isolated ReOx crystallite; and a broad higher temperature, indicative of a decrease in reducibility of
peak (II) observed over the temperature range of 660–830 K rhenium oxide species with increasing Re surface coverage.
may be attributed to the deep reduction of Re species This trend is in agreement with BET surface area results
which suggested that the Re species were not confined in the
pore of the support but rather deposited on the surface.
Thus, as the ReOx loading increased, the amount of surface
Re oxide species increased and formed several surface layers
of ReOx clusters,60 which may have limited H2 diffusion to
the ReOx species in the innermost layer and decreased the
reducibility.
The catalysts were analysed by XPS to characterize the sur-
face by monitoring the Re 4f, C 1s and O 1s regions before
and after (phenol HDO at 573 K and 3 MPa H2) reaction. The
binding energies (BE) and atomic ratios for all the calcined
catalysts are listed in Table 2. Fig. 4a shows the Re 4f XPS
spectra of the selected calcined catalysts. The rest of the XPS
spectra can be found in the ESI (Fig. S1†). The C 1s binding

Fig. 4 XPS spectra of selected catalysts in the Re 4f spectral region of


Fig. 3 TPR profiles of the calcined ReOx/CNFox catalysts. the (a) calcined and (b) spent ReOx/CNFox catalysts.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016
View Article Online

Catalysis Science & Technology Paper

energies of the calcined catalysts were identical to those ob- other hand, the binding energies of the second doublet
served for the oxidized CNF support discussed earlier (the showed small negative shifts of −0.2 eV to −0.5 eV relative to
same holds true for the O 1s signals). The Re 4f XPS spectra the BE of Re6–7+ species in the calcined samples. This de-
of the calcined catalysts display the characteristic 4f doublet crease in positive Re6–7+ charge could be attributed to partial
whose spin–orbit splitting (Re 4f7/2 and 4f5/2) is 2.4 eV. Nota- replacement of oxygen with a less electronegative carbon
bly, the Re 4f7/2 component for the calcined catalysts with dif- from the support to form a Re oxycarbide species. It has been
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

ferent loadings are nearly identical, which could be attributed previously reported that defective carbon atoms from the
to the weak Re–carbon support interaction in comparison to CNFox support are capable of dissolving into metal and metal
other supports.41 The observed BE at 45.6 eV ± 0.1 eV corre- oxide lattices during reduction.66,67
sponds to Re6–7+ species because it is 0.7 eV higher than the The distribution of rhenium oxidation states derived from
BE normally observed for ReO3 (Re6+) but 1.1 eV less than the XPS spectra of fresh and spent catalysts is shown in pa-
that for Re2O7 (Re7+).61–63 This value was slightly less than rentheses in Tables 2 and 3 and illustrated in Fig. 5. The data
the Re 4f7/2 BE value observed for reference bulk ReOx (46.0 reveal similar relative abundance of Re species, suggesting
eV) which is assigned to ReO4− species.64 On the other hand, that the ReOx/CNFox catalytic sites were not significantly mod-
the same Re6–7+ species was present on the ReOx/SiO2 catalyst ified by an increase in Re loading. A corollary of this observa-
reported by Leiva et al.;41 however, unlike the ReOx/SiO2 cata- tion is that product distribution from phenol conversion is
lyst, only one Re oxide species was present on the surface of anticipated to be similar.
the ReOx/CNFox samples, suggesting that surface oxygen func- The coverage of the support by Re species in the catalyst
tional groups on the CNFox support may have stabilized the was estimated from the measurements of the Re/C intensity
ReOx structure during calcination. ratio in the XPS spectra. The results in Fig. 5 show that Re
The O1s BE of the bulk ReOx sample (not shown) surface coverage increases linearly with ReOx loading for the
appeared at 531.2 eV for lattice Re-O which is close to the BE calcined catalysts, indicating homogeneous distribution of
of 531.7 assigned to phenolic or ether groups of the CNFox rhenium oxide on the CNFox support within the range of
support. Thus, a part of the O 1s BE at 531.1 eV observed for loading used. Fig. 5 also shows that the Re/C atomic ratio of
the ReOx/CNFox catalysts is certainly related to the presence the spent catalyst decreased relative to the corresponding cal-
of Re6–7+ species. The absence of surface and crystalline cined catalyst and the difference became more pronounced
Re2O7 nanoparticles from XRD and XPS analyses is due to the at high loadings, leading to a slight deviation from linearity
volatility of dimeric and polymeric Re2O7 species65 of Re dispersion above 10 wt% ReOx attributed to the forma-
The XPS results of the spent catalysts are summarized in tion of ReOx agglomerates. Thus, in summary, the above re-
Table 3. Similarly, the data show that the observed Re 4f7/2 sults show that although the nature of the active sites on
and 4f5/2 signals are similar in all the catalysts. The most ReOx/CNFox is independent on loading, the amount increases
intense 4f7/2 component of the 4f doublet is located at 42.5 ± almost linearly with loading.
0.1 eV and 45.4 eV ± 0.3 eV: the Re 4f7/2 signal observed at
42.5 eV is assigned to ReIJIV) species of ReO2 (ref. 61 and 62),
indicating partial H2 reduction during the reaction. On the 3.3. Catalytic activity
The reactivity of the ReOx/CNFox catalysts is tested in the
hydrodeoxygenation of lignin-derived phenolic compounds.
Table 3 XPS data of the spent catalysts We firstly delved into the effect of rhenium oxide loading to
determine the possible structure–reactivity relationship for
Binding energy (eV) Surface atomic
(distribution/%) ratio (at%)
Sample C 1s Re 4f5/2 O1s O/C Re/C
2ReOx/CNFox 284.8 (63)
286.2 (28) 42.4 (37) 531.6 (38) 0.121 0.0017
288.5 (9) 45.1 (63) 533.1 (62)
4ReOx/CNFox 284.8 (61) 531.1 (26)
286.2 (26) 42.4 (25) 532.2 (32) 0.133 0.0029
288.6 (13) 45.2 (75) 533.3 (42)
6ReOx/CNFox 284.8 (55) 531.1 (38)
286.2 (33) 42.6 (27) 532.3 (23) 0.156 0.0038
288.5 (12) 45.2 (73) 533.4 (39)
8ReOx/CNFox 284.2 (63) 531.1 (36)
286.3 (26) 42.6 (26) 532.3 (22) 0.109 0.0054
288.5 (11) 45.2 (74) 533.3 (42)
10ReOx/CNFox 284.8 (63) 531.2 (39)
286.2 (26) 42.3 (27) 532.4 (24) 0.114 0.0063
288.4 (11) 45.5 (73) 533.5 (37)
13ReOx/CNFox 284.8 (65) 531.1 (40) Fig. 5 Distribution of rhenium oxidation states and Re/C surface
286.2 (27) 42.6 (22) 532.3 (24) 0.136 0.0073 atomic ratio of fresh and spent ReOx/CNFox catalysts as a function of
288.3 (8) 45.7 (78) 533.5 (36) ReOx loading.

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

the conversion of phenol, investigated the impact of hydro-


gen pressure, and compared the reactivity of other phenolic
model compounds.
3.3.1. Effect of ReOx loading. The conversion of phenol
and the evolution of reaction products are shown in Fig. 6.
The concentration of the reactants and the product yields
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

were determined relative to the hexadecane standard. The


ReOx/CNFox catalysts displayed similar concentration profiles:
benzene and cyclohexane were the major products, while
cyclohexanol and cyclohexene were the minor products.
Fig. 7 Reaction scheme for the conversion of phenol. The figure is
These compounds and the undetected cyclohexanone consti-
drawn on the basis of the information from ref. 14 and 31.
tute the product stream from phenol conversion over other
catalytic systems as well.30 The reaction pathway during the
conversion of phenol is depicted in Fig. 7,14,31 whereby phe-
nol proceeds through either direct deoxygenation (DDO) of For the ReOx/CNFox catalysts, XRD and XPS data showed the
the C–O σ bond to form benzene or hydrogenation (HYD) of presence of Re6–7+ species in the calcined catalysts. On the
the aromatic ring to form cyclohexanol. Benzene can then be other hand, XPS analysis of the spent catalyst showed the
saturated to form cyclohexane. On the other hand, cyclo- presence of incompletely reduced Re4+ and Re oxycarbide
hexanol can be dehydrated to form cyclohexene and then a fi- species. Taken together, it is plausible to deduce the presence
nal hydrogenation step to form cyclohexane. of oxygen vacancies and surface hydroxyls on these catalysts
The catalytic active sites on reducible metal oxides are during the reaction.
coordinatively unsaturated metal sites (i.e. oxygen vacancies) Analogous to sulfur vacancies on sulfide catalysts,14 the
and Brønsted acid sites of surface hydroxyls formed during in oxygen vacancy sites on the ReOx/CNFox catalyst serve as se-
situ partial reduction of the surface lattice oxygen of the lective sites for the adsorption of phenol. A possible sche-
metal oxide.42,68,69 The oxygen vacancies are created through matic representation adapted from the studies by Romero
the loss of water via dehydroxylation on terminal sites when et al.,13 Mortensen et al.,30 Moberg et al.68 and Prasomsri
H2 adsorbs on the surface oxygens forming hydroxyl groups. et al.40 is shown in Fig. 8. After the creation of an oxygen va-
In effect, the surface hydroxyls are instrumental in the forma- cancy and surface hydroxyls, selective deoxygenation of phe-
tion of both the oxygen vacancies and Brønsted acid sites.68 nol on reducible ReOx takes place in a stepwise fashion
according to this schematic: (1) the coordinatively unsatu-
rated Re4+ acts as a Lewis acid and interacts with the oxygen
lone pair on phenol; (2) then, the C-1 of phenol becomes

Fig. 6 Conversion of phenol and the yield of products with time over
(a) 2ReOx/CNFox, (b) 4ReOx/CNFox, (c) 6ReOx/CNFox, (d) 8ReOx/CNFox,
(e) 10ReOx/CNFox, and (f) 13ReOx/CNFox catalysts. Reaction conditions: Fig. 8 Possible mechanism of direct deoxygenation of phenol to
phenol (1.783 g), yReOx/CNFox (200 mg), dodecane (80 mL), 573 K, 3 benzene on the partially reduced ReOx phase. The figure is drawn on
MPa H2. the basis of the information from ref. 13, 30, 40, and 68.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016
View Article Online

Catalysis Science & Technology Paper

highly nucleophilic and accepts a proton (H+) from surface guaiacol to phenol. In addition, the creation of vacancies by
hydroxyl groups, leading to the formation of an adsorbed H2 as a reductant and the coordination of the OH group of
phenoxide ion;70 (3) C–O bond cleavage of the phenoxide and the substrate to the low-valent Re species mimic the mecha-
desorption; and (4) the recovery of the oxygen vacancy by the nism proposed by Ota et al.72 in the HDO of 1,4-
elimination of water. For the less preferred hydrogenation anhydroerythritol over the ReOx–Pd/CeO2 catalyst. The forma-
pathway, a flat adsorption through the aromatic π bonding tion of cyclohexanol and the significant amount of cyclohexane
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

involving at least two neighbouring vacancies as active sites demonstrate the bi-functionality of the reducible ReOx catalysts
is typically proposed.13,14 The product selectivity calculated at possessing both hydrogenolysis (metal oxide) and hydrogenating
10% conversion of phenol shown in Fig. 9a better illustrates (metal-like) sites. It is unclear which of the Re species deter-
the preferred route for the ReOx/CNFox catalyst. All the cata- mined from XPS measurements of post-reaction catalysts (ReO2
lysts demonstrated a strong preference for the DDO pathway or ReOxCy) is responsible for the hydrogenating capabilities of
as evidenced by benzene being the most dominant product these catalysts. Computational calculations and surface science
(51–59%). Cyclohexanol (3–8%) and cyclohexene (2–11%), experiments are warranted to provide insight into the respective
products of the HYD pathway, were produced in significantly reactivities of these Re species. Another noteworthy observation
less quantities. The products also included cyclohexane (26– is the similarity in the selectivity pattern of all the ReOx/CNFox
35%) which can be formed via either route according to the catalysts. This can be explained by the relative abundance of Re
reaction scheme shown in Fig. 7. These results indicate that active sites derived from XPS analysis (shown in Fig. 5). The ref-
despite the DDO pathway being preferred by a significant erence bulk ReOx catalyst produced a higher selectivity to ben-
margin, the HYD pathway cannot be ignored due to the ap- zene (73%) in comparison to the CNT-supported catalysts. The
preciable quantities of ring hydrogenated products. However, results further demonstrate the propensity of the ReOx phase
based on these results, the ReOx/CNFox catalysts likely for C–O bond scission.
followed the possible schematic representation shown in The influence of acid sites on the oxidized CNF support
Fig. 8. Our results agree well with the observation of Prasomsri on the selectivity cannot be completely discounted. In fact,
et al.40 who reported that MoO3 was selective for C–O bond the 10ReOx/CNFox catalyst displayed higher benzene selectiv-
cleavage of anisole. A recent kinetic study by Leiva et al.71 ity (59%) compared to a reference ReOx/SiO2 catalyst (45%)
has proposed the presence of only one active site on ReOx/ with similar Re surface density which can be attributed to en-
SiO2, likely coordinative unsaturated sites, which converts hanced hydrogenolysis capability of the former due to the
presence of acidic surface oxygen groups on the CNFox
support.
The catalytic activity of the ReOx/CNFox catalyst, expressed
as the initial reaction rate (μmolPHE converted gcat−1 s−1), is
given in Table 4 and illustrated in Fig. 9b. The unsupported
ReOx catalyst is included as reference. The activity increased
continuously with an increase in the ReOx loading, reaching
a maximum at 10 wt%, and then declined. A look at Fig. 5
shows that the trend of the initial reaction rates follows the
XPS Re/C atomic ratio of the spent catalyst and not the cal-
cined catalysts. It can thus be inferred that changes to the
catalyst surface induced by the reactive H2 at 573 K and 3 MPa
occurred within the first 40 min of the reaction and remained
stable afterwards. Thus, although the true nature of the active

Table 4 Initial reaction rates of ReOx/CNFox catalysts as a function of


ReOx loading and H2 pressure for the HDO of phenol at 573 K

Reaction rates
Catalyst (μmolPHE converted gcat−1 s−1)
2ReOx/CNFox 0.2
4ReOx/CNFox 1.2
6ReOx/CNFox 2.9
8ReOx/CNFox 4.7
10ReOx/CNFox 9.0
13ReOx/CNFox 4.8
Fig. 9 (a) Product selectivity calculated at 10% phenol conversion and Unsupported ReOx 5
(b) reaction and intrinsic rates as a function of ReOx loading. The 10ReOx/CNFox (0 MPa) 0
product selectivity did not include the 2ReOx/CNFox catalyst due to its 10ReOx/CNFox (2 MPa) 4.7
very low conversion value. Reaction conditions: phenol (1.783 g), 10ReOx/CNFox (3 MPa) 9.0
yReOx/CNFox (200 mg), dodecane (80 mL), 573 K, 3 MPa H2. 10ReOx/CNFox (5 MPa) 14.6

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

phase during the reaction is unknown, post-reaction surface Product selectivity was calculated at 20% phenol conversion
characterization of reducible metal oxide catalysts is more re- and shown in Fig. 10a, which reveals clear distinction of prod-
flective of the catalyst behaviour than the pre-reaction character- uct distribution with H2 pressure. An increase in hydrogen
ization. Hence, the activity of ReOx/CNFox catalyst is assigned to pressure decreased the yield of benzene and enhanced the se-
the presence of Re4+ and Re6–7+ species derived from XPS mea- lectivity of cyclohexanol and cyclohexane. This trend indicates
surements of the spent catalysts. This is in agreement with the enhanced hydrogenation activity at the expense of direct deox-
study by Ota et al.72 who attributed the high catalytic perfor-
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

ygenation activity, consistent with the recent findings by Jin


mance of ReOx-Pd/CeO2 catalyst in the HDO of 1,4- et al.35 for the HDO of anisole. The selectivity to cyclohexene
anhydroerythritol to rhenium species with the +4 valence state. was low and insensitive to hydrogen pressure, possibly because
In relation to the trend, the low activity at low loadings reflects the hydrogenation of cyclohexene is a significantly faster reac-
the paucity of Re active sites, while the increase in the number tion than the other steps.30 It thus appears that changes in re-
of catalytically active sites with Re loading enhanced the rate of action rates and product distribution as a function of H2 pres-
phenol conversion. However, a further increase in loading be- sure are kinetically controlled. However, we cannot discount
yond an optimum value (10 wt% ReOx loading) led to loss of ac- the effect of the increase in H2 pressure on the nature of the
tive sites from crystallite agglomeration and, as a consequence, catalytically active surface during the reaction. An increase in
a drop in activity. Fig. 9b also shows the intrinsic activity (reac- H2 pressure is expected to increase H2 dilution which might
tion rates normalized with exposed Re atoms derived from XPS induce significant changes to the catalyst's surface properties
Re/C surface atomic ratios). The trend obtained for the intrinsic and, thus, modify the mechanism of the reaction.
rate was similar to the reaction rates, indicating a clear depen- The dependence of phenol HDO on H2 pressure is even
dency of the activity on the dispersion of the ReOx sites. more evident at 4 h reaction time, as shown in Fig. 10b. Total
Based on the results discussed here, the 10ReOx/CNFox cat- conversion increased to 100 mol% at 3 MPa and above. Addi-
alyst was selected for further study. tionally, a more pronounced increase in cyclohexane selectiv-
3.3.2. Effect of hydrogen pressure. It is well known that ity, with a concomitant decrease in benzene production, with
low H2 pressure favours the direct production of aromatics, hydrogen pressure was observed. There were also loss of
while high H2 pressure tends to steer the reaction towards cyclohexanol and cyclohexene. This profound effect suggests
the HYD route.4 Thus, to elucidate whether the ReOx/CNFox possible thermodynamic effects. Furimsky4 showed, through
catalyst affinity for cleaving C–O bond was due to their inher- equilibrium calculations, that phenol conversion increases
ent properties or the favourable reaction conditions with H2 pressure due to enhanced oxygen elimination and in-
employed, the impact of hydrogen pressure was investigated creased aromatic ring hydrogenation. In a nutshell, although
at 2, 3 and 5 MPa H2 and at 300 °C. A reaction was also high H2 pressure hampered direct deoxygenation, benzene
conducted in the absence of hydrogen (under 0.5 MPa N2) and was still formed in a significant amount over the ReOx/CNFox
would be referred to here as 0 MPa H2. There was no detect- catalyst. The desirability of the DDO route has implications
able conversion after 4 h in the absence of H2 (at 0 MPa H2), on their potential as a hydrogen efficient HDO catalyst.
indicative of the indispensability of H2 in catalytic HDO. The 3.3.3. Conversion of lignin model compounds. To demon-
time course of the HDO of phenol at varying pressure is shown strate the efficacy of the reducible Re oxide catalysts for the
in the ESI (Fig. S2†). Fig. 10a shows the sensitivity of the HDO conversion of other lignin monomers, the 10ReOx/CNFox cata-
of phenol to hydrogen pressure. Similar hydrogen pressure ef- lyst was tested for the conversion of guaiacol, anisole,
fects on lignin-derived compounds have been reported for o-cresol and phenol under conditions comparable to the
other catalytic systems.8,35 The initial reaction rates increased studies reported by Jongerius et al.5 Fig. 11 illustrates the
almost linearly with H2 pressure, suggesting a pseudo first- evolution of the reaction products from the conversion of
order dependence on the rate of phenol conversion with hy- guaiacol (Fig. 11[a]), phenol (Fig. 11[b]), anisole (Fig. 11[c])
drogen pressure within the range studied. This observation and o-cresol (Fig. 11[d]). Guaiacol yielded mainly phenol,
mirrors the findings of Odebunmi and Ollis8 with sulfided benzene and cyclohexane, while toluene and anisole were
CoO–MoO3/Al2O3 catalysts for the conversion of cresols. detected in minor amounts. In addition, trace amounts of
cyclohexanol, xylene and cyclohexane were also detected. Phe-
nol and anisole were intermediate products judging by their
concentration profiles. In the conversion of phenol (Fig. 11[b]),
cyclohexane and benzene were the dominant products, while
cyclohexanol and cyclohexene were detected in significantly
less quantities. Fig. 11(c) shows that anisole was converted to
benzene and methoxycyclohexane and subsequently to cyclo-
hexane; a trace amount of methanol was also detected. The
conversion of o-cresol yielded toluene and methylcyclohexane
Fig. 10 Product selectivity from conversion of phenol as a function of
in almost equal quantities. Methylcyclohexanol was an inter-
H2 pressure: (a) at 20% conversion and (b) after 4 h reaction time. mediate product; in addition, benzene, cyclohexene and cyclo-
Phenol (1.783 g), 10ReOx/CNFox (200 mg), dodecane (80 mL), 573 K. hexane were detected in trace amounts.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016
View Article Online

Catalysis Science & Technology Paper


Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

Fig. 12 Reaction scheme of the conversion of guaiacol including the


conversions of anisole, phenol and cresol. The figure is drawn on the
basis of the information from ref. 74 and 75.
Fig. 11 Conversion of lignin-derived substrates and the yield of prod-
ucts as a function of time over the 10ReOx/CNFox catalyst. Substrate:
(a) guaiacol, (b) anisole, (c) phenol and (d) o-cresol. Reaction condi-
tions: substrate (0.232 mol L−1), 10ReOx/CNFox (200 mg), dodecane (80 guaiacol (Gua) underwent either demethoxylation (DMO) to
mL), 573 K, 5 MPa H2. form phenol (Phe) or dehydroxylation (DDO) to form anisole
(Ani). The non-detection of catechol cannot be unequivocally at-
tributed to the reaction bypassing the catechol intermediate; it
Table 5 shows that complete conversion of guaiacol and could be due to its rapid conversion to phenol since this reac-
phenol was achieved after 4 h of reaction, while conversions tion is facile.27 The DMO pathway was more prevalent than the
of 88% and 44% were observed for anisole and cresol, respec- DDO pathway (22.1 vs. 1.2 mol%) which is likely due to the
tively. The significantly lower reactivity of o-cresol could be lower bond dissociation energy (BDE) of Caromatic–OCH3 than
due to the catalyst's relative inertness towards C–C bond scis- Caromatic–OH (91.5 vs. 109.2 kcal mol−1);76 consequently, it is eas-
sion which would be discussed later on. However, steric and ier to remove methoxyl than hydroxyl from guaiacol. The second
electronic effects associated with the methyl group cannot be stage conversion such as the DDO of phenol to form benzene
discounted.73 (Ben) also occurred, which attests to the high reactivity of
The product yields calculated at 1 h and 4 h reaction times guaiacol with the ReOx/CNFox catalyst. On the other hand,
are also presented in Table 5. The data at 1 h depict the com- anisole conversion was initiated by demethoxylation to produce
mencement of the reaction after the initial induction period of benzene (8.6 mol%) or hydrogenation to form methoxy-
the catalyst and, hence, are used to elucidate the routes of pri- cyclohexane (0.9%), followed by ring hydrogenation to form
mary conversion of the substrate. A generalized reaction scheme cyclohexane (CyH, 5.7 mol%).
of the conversion of guaiacol—encompassing the conversions of The conversion of phenol proceeded via two routes: (i) direct
phenol, anisole and o-cresol—adapted from Bui et al.74 and deoxygenation to benzene (19.0%) or (ii) ring hydrogenation to
Nimmanwudipong et al.75 is shown in Fig. 12 to illustrate the form cyclohexanol (CyHnol, 6.8 mol%). The DDO and HYD
pathways taken by the catalyst. On the basis of the results, pathways have already been discussed in the earlier sections.

Table 5 HDO of lignin model compounds over the 10ReOx/CNFox catalyst in a batch reactor at 573 K and 5 MPa H2 for 1 and 4 h

Time Conv. HDO Selectivity (mol%)


(h) (mol%) % CyH Ben Phe CyHnol Ani Tol MetCyH MetCyHnol
Gua 1 27.7 10.4 1.3 1.3 22.1 1.3 1.2 0 0 0
4 100 88.6 66.1 19.0 7.9 1.2 1.8 3.3 0 0
Ani 1 14.8 98.5 5.7 8.6 0 0 — 0 0 0
4 88.1 99.5 52.7 30.2 0 0 — 0 0 0
Phe 1 50.6 86.6 24.1 19.0 — 6.8 0 0 0 0
4 100 100 82.4 17.6 0 0 0 0 0
Cre 1 7.2 77.9 0.1 0.1 0 0 0 3.2 2.2 1.6
4 43.9 96.6 1.1 0.5 0 0 0 19.3 21.4 1.5

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

A notable observation is the difference in benzene selectiv- support acidity is required for dealkylation. A Ga/HBEA cata-
ity from the conversion of phenol and anisole: the latter lyst was substantially more active than a Ga/SiO2 catalyst,
yielded more benzene than the former (53.0 vs. 37.5%). This and the authors suggested that the disparity associates with
is consistent with the higher bond strength of the Caromatic– the fact that SiO2 does not contain any significant acidity.
OH bond (106.1 kcal mol−1) compared with the Caromatic– However, the CNFox support contains significant acidity but
OCH3 bond (91.1 kcal mol−1), in a similar fashion as the de- was nonetheless unable to cleave the C–C bond on o-cresol.
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

methoxylation vs. dehydroxylation selectivity difference from Therefore, we can conclude that the very low degree of
guaiacol conversion.76 However, it should be noted that the dealkylation cannot be associated with the acidity of the
relative bond strength alone cannot explain the observed se- ReOx/CNFox catalyst but rather by the intrinsic preference of
lectivity differences. The ReOx/CNFox catalyst preferentially ReOx species for C–O bond scission.
cleaved the much stronger Caromatic–OCH3 bond over the O– Direct comparison of these results with the studies
CH3 bond (51.2 kcal mol−1) in guaiacol and anisole, in line reported by Jongerius et al.5 with the sulfided CoMo/Al2O3
with the findings of Prasomsri et al.76 over reducible MoO3 catalyst under comparable conditions underscores the poten-
catalysts. Furthermore, unlike phenol conversion which pro- tial of the ReOx/CNFox catalyst for the conversion of lignin-
duced cyclohexanol in appreciable amounts, the ring hydro- derived compounds. The authors reported guaiacol, phenol,
genated counterpart from anisole conversion (i.e. methoxy- anisole and o-cresol conversions of 50, 30, 40 and 27%, re-
cyclohexane) was only produced in trace amounts; this spectively, which lagged far behind the ReOx/CNFox catalyst
demonstrates the affinity of the ReOx/CNFox catalyst to DMO. which exhibited corresponding conversions of 100, 100, 88
Hence, it can be inferred that the –OCH3 on guaiacol and and 44%. Additionally, there was no evidence of demethyla-
anisole bonds with surface reducible metal oxide (ReO2 and tion and transalkylation with this catalyst, in contrast to the
ReOxCy) species, in agreement with the recent findings by sulfided catalyst. However, the ReOx/CNFox catalyst also
Prasomsri et al.40 who demonstrated from the DFT model exhibited an inherent hydrogenation character and, thus, was
that reduced MoO3 (Mo3O9) bonded with the methoxy group more prone to over-saturation of the aromatic ring. Addi-
on anisole. In addition, Chang et al.77 reported that partially tional studies are required to suppress its hydrogenation ac-
reduced MoOx (Mo4+) species were responsible for the direct tivity without sacrificing its high activity.
demethoxylation of guaiacol and anisole to form phenol and
benzene, respectively.
The o-cresol reaction proceeded solely via DDO to produce 4. Conclusions
toluene (Tol), part of which was subsequently hydrogenated
Phenol was used as a lignin model compound to determine
to form methylcyclohexane (MetCyH).
the reactivity of carbon nanofiber-supported rhenium oxide
The selectivity at 4 h shed insight into the products
catalysts as a function of ReOx loading and revealed a direct
formed at the tail end of the reaction. A very high to com-
correlation between the initial reaction rate and relative Re
plete degree of hydrodeoxygenation was obtained for all the
dispersion derived from XPS measurements of spent cata-
compounds (83.4–100%). The conversion of the lignin mono-
lysts. The reaction was catalysed in situ generated oxygen va-
mers over the ReOx/CNFox catalyst resulted in the predomi-
cancies and exhibited a preference for direct deoxygenation
nant formation of ring saturated products such as cyclohex-
of phenol to benzene. In addition, the product distribution
ane and methylcyclohexane: guaiacol, phenol and anisole
was largely congruent, indicative that the nature of the active
generated 59.5–82.4% cyclohexane, formed mainly from the
sites was not modified by an increase in loading. The activity
hydrogenation of benzene; o-cresol produced 48.7% of
and product distribution were also strong functions of H2
methylcyclohexane. However, a significant fraction of aroma-
pressure: a higher pressure enhanced the activity and the hy-
ticity was retained as evidenced by the 17.6–34.0% selectivity
drogenation capability but diminished the benzene selectiv-
towards benzene, toluene and xylene (BTX). The selectivity to-
ity. The best ReOx/CNFox catalyst was also applied to other
wards other products such as cyclohexanol and methyl-
lignin-derived substrates such as guaiacol, anisole and
cyclohexanol (MetCyHnol) did not exceed 3%. Some other
o-cresol and demonstrated an ability to selectively cleave the
noteworthy observations from Table 5 are as follows: (1) no
C–O bond without C–C bond cleavage, thereby retaining all
cresols (methylphenols) were detected from the conversion of
carbon atoms. This demonstrates that the ReOx phase is
guaiacol. Typically, methylation (ME) of phenol to produce
hydrogen-efficient and carbon-atom economical.
cresols could occur as a result of CH4 produced from demeth-
ylation (DME) of guaiacol to form catechol. However, there
was no evidence of DME which explains the absence of cre- Acknowledgements
sols in the product stream. (2) The catalyst's relative inert-
ness towards C–C bond cleavage is evident in Table 5, where The authors gratefully acknowledge the financial support
demethylated products account for a measly 3.9% of the from CONICYT-Chile for FONDECYT Project No. 3150033 and
product selectivity from the conversion of o-cresol. It was pre- 1140528 and Project Basal PFB-27. We thank Dr. Freddy Celis
viously noted by Ausavasukhi et al.78 in the study of the HDO and Dr. Marcelo Campos for helpful discussion on Raman
of m-cresol over Ga-modified beta zeolite catalysts that strong spectra measurements.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016
View Article Online

Catalysis Science & Technology Paper

Notes and references 27 H. Y. Zhao, D. Li, P. Bui and S. T. Oyama, Appl. Catal., A,
2011, 391, 305–310.
1 S. Czernik and A. V. Bridgwater, Energy Fuels, 2004, 18, 28 Y. Li, X. Yang, L. Zhu, H. Zhang and B. Chen, RSC Adv.,
590–598. 2015, 5, 80388–80396.
2 M. Saidi, F. Samimi, D. Karimipourfard, T. 29 S. Boullosa-Eiras, R. Lødeng, H. Bergem, M. Stöcker, L.
Nimmanwudipong, B. C. Gates and M. R. Rahimpour, Hannevold and E. A. Blekkan, Catal. Today, 2014, 223,
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

Energy Environ. Sci., 2014, 7, 103–129. 44–53.


3 H. Wang, J. Male and Y. Wang, ACS Catal., 2013, 3, 30 P. M. Mortensen, J.-D. Grunwaldt, P. A. Jensen and A. D.
1047–1070. Jensen, ACS Catal., 2013, 3, 1774–1785.
4 E. Furimsky, Appl. Catal., A, 2000, 199, 147–190. 31 S. Echeandia, P. L. Arias, V. L. Barrio, B. Pawelec and J. L. G.
5 A. L. Jongerius, R. Jastrzebski, P. C. A. Bruijnincx and B. M. Fierro, Appl. Catal., B, 2010, 101, 1–12.
Weckhuysen, J. Catal., 2012, 285, 315–323. 32 M. V. Bykova, D. Y. Ermakov, S. A. Khromova, A. A. Smirnov,
6 C. Newman, X. Zhou, B. Goundie, I. T. Ghampson, R. A. M. Y. Lebedev and V. A. Yakovlev, Catal. Today, 2014, 220–
Pollock, Z. Ross, M. C. Wheeler, R. W. Meulenberg, R. N. 222, 21–31.
Austin and B. G. Frederick, Appl. Catal., A, 2014, 477, 64–74. 33 M. V. Bykova, D. Y. Ermakov, V. V. Kaichev, O. A.
7 H. Weigold, Fuel, 1982, 61, 1021–1026. Bulavchenko, A. A. Saraev, M. Y. Lebedev and V. A. Yakovlev,
8 E. O. Odebunmi and D. F. Ollis, J. Catal., 1983, 80, 56–64. Appl. Catal., B, 2012, 113–114, 296–307.
9 E. Laurent and B. Delmon, Appl. Catal., A, 1994, 109, 77–96. 34 M. V. Bykova, O. A. Bulavchenko, D. Y. Ermakov, M. Y.
10 M. K. Huuska, Polyhedron, 1986, 5, 233–236. Lebedev, V. A. Yakovlev and V. N. Parmon, Catal. Ind.,
11 R. K. M. R. Kallury, W. M. Restivo, T. T. Tidwell, D. G. B. 2011, 3, 15–22.
Boocock, A. Crimi and J. Douglas, J. Catal., 1985, 96, 35 S. Jin, Z. Xiao, C. Li, X. Chen, L. Wang, J. Xing, W. Li and C.
535–543. Liang, Catal. Today, 2014, 234, 125–132.
12 C. V. Loricera, B. Pawelec, A. Infantes-Molina, M. C. Álvarez- 36 Y. Yang, C. Ochoa-Hernández, V. A. de la Peña O'Shea, P.
Galván, R. Huirache-Acuña, R. Nava and J. L. G. Fierro, Pizarro, J. M. Coronado and D. P. Serrano, Appl. Catal., B,
Catal. Today, 2011, 172, 103–110. 2014, 145, 91–100.
13 Y. Romero, F. Richard and S. Brunet, Appl. Catal., B, 37 W. Song, Y. Liu, E. Barath, C. Zhao and J. A. Lercher, Green
2010, 98, 213–223. Chem., 2015, 17, 1204–1218.
14 F. E. Massoth, P. Politzer, M. C. Concha, J. S. Murray, J. 38 W. Song, Y. Liu, E. Barath, C. Zhao and J. A. Lercher, Green
Jakowski and J. Simons, J. Phys. Chem. B, 2006, 110, Chem., 2015, 17, 1204–1218.
14283–14291. 39 C. Zhao, S. Kasakov, J. He and J. A. Lercher, J. Catal.,
15 C. Zhao, J. He, A. A. Lemonidou, X. Li and J. A. Lercher, 2012, 296, 12–23.
J. Catal., 2011, 280, 8–16. 40 T. Prasomsri, T. Nimmanwudipong and Y. Roman-Leshkov,
16 C. Zhao, Y. Kou, A. A. Lemonidou, X. Li and J. A. Lercher, Energy Environ. Sci., 2013, 6, 1732–1738.
Angew. Chem., Int. Ed., 2009, 48, 3987–3990. 41 K. Leiva, N. Martinez, C. Sepulveda, R. García, C. A. Jiménez,
17 C. Zhao, W. Song and J. A. Lercher, ACS Catal., 2012, 2, D. Laurenti, M. Vrinat, C. Geantet, J. L. G. Fierro, I. T.
2714–2723. Ghampson and N. Escalona, Appl. Catal., A, 2015, 490,
18 C. Zhao and J. A. Lercher, ChemCatChem, 2012, 4, 64–68. 71–79.
19 C. Zhao, D. M. Camaioni and J. A. Lercher, J. Catal., 42 V. M. L. Whiffen and K. J. Smith, Energy Fuels, 2010, 24,
2012, 288, 92–103. 4728–4737.
20 J. Y. He, C. Zhao and J. A. Lercher, J. Catal., 2014, 309, 43 T. Ressler, A. Walter, J. Scholz, J. P. Tessonnier and D. S. Su,
362–375. J. Catal., 2010, 271, 305–314.
21 D.-Y. Hong, S. J. Miller, P. K. Agrawal and C. W. Jones, 44 E. Santillan-Jimenez, M. Perdu, R. Pace, T. Morgan and M.
Chem. Commun., 2010, 46, 1038–1040. Crocker, Catalysts, 2015, 5, 424–441.
22 S. Echeandia, B. Pawelec, V. L. Barrio, P. L. Arias, J. F. 45 R. Cid and G. Pecchi, Appl. Catal., 1985, 14, 15–21.
Cambra, C. V. Loricera and J. L. G. Fierro, Fuel, 46 C. D. Wagner, L. E. Davis, M. V. Zeller, J. A. Taylor, R. H.
2014, 117(Part B), 1061–1073. Raymond and L. H. Gale, Surf. Interface Anal., 1981, 3,
23 W. Zhang, J. Chen, R. Liu, S. Wang, L. Chen and K. Li, ACS 211–225.
Sustainable Chem. Eng., 2014, 2, 683–691. 47 J.-P. Tessonnier, D. Rosenthal, T. W. Hansen, C. Hess, M. E.
24 A. L. Jongerius, R. W. Gosselink, J. Dijkstra, J. H. Bitter, Schuster, R. Blume, F. Girgsdies, N. Pfänder, O. Timpe, D. S.
P. C. A. Bruijnincx and B. M. Weckhuysen, ChemCatChem, Su and R. Schlögl, Carbon, 2009, 47, 1779–1798.
2013, 5, 2964–2972. 48 A. B. Dongil, I. T. Ghampson, R. Garcia, J. L. G. Fierro and
25 W.-S. Lee, Z. Wang, R. J. Wu and A. Bhan, J. Catal., N. Escalona, RSC Adv., 2016, 6, 2611–2623.
2014, 319, 44–53. 49 V. Castranova, P. A. Schulte and R. D. Zumwalde, Acc. Chem.
26 I. T. Ghampson, C. Sepulveda, R. Garcia, B. G. Frederick, Res., 2013, 46, 642–649.
M. C. Wheeler, N. Escalona and W. J. DeSisto, Appl. Catal., 50 J. L. Figueiredo, M. F. R. Pereira, M. M. A. Freitas and
A, 2012, 413–414, 78–84. J. J. M. Órfão, Carbon, 1999, 37, 1379–1389.

This journal is © The Royal Society of Chemistry 2016 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

51 L. R. Radovic and F. Rodriguez-Reinoso, Chemistry and 66 A. Rinaldi, J.-P. Tessonnier, M. E. Schuster, R. Blume, F.
Physics of Carbon, ed. P. A. Thrower, Marcel Dekker, New Girgsdies, Q. Zhang, T. Jacob, S. B. Abd Hamid, D. S. Su and
York, 1996, vol. 25, p. 243. R. Schlögl, Angew. Chem., Int. Ed., 2011, 50, 3313–3317.
52 U. Zielke, K. J. Hüttinger and W. P. Hoffman, Carbon, 67 B. Frank, K. Friedel, F. Girgsdies, X. Huang, R. Schlögl and
1996, 34, 983–998. A. Trunschke, ChemCatChem, 2013, 5, 2296–2305.
53 Y. Otake and R. G. Jenkins, Carbon, 1993, 31, 109–121. 68 D. R. Moberg, T. J. Thibodeau, F. o. G. Amar and B. G.
Published on 27 January 2016. Downloaded by Flinders University of South Australia on 09/02/2016 15:06:32.

54 R. J. Nemanich and S. A. Solin, Phys. Rev. B: Condens. Matter Frederick, J. Phys. Chem. C, 2010, 114, 13782–13795.
Mater. Phys., 1979, 20, 392–401. 69 H. H. Kung, Surface Coordination Unsaturation. Transition
55 Y. Nie and T. Hübert, Polym. Int., 2011, 60, 1574–1580. Metal Oxides: Surface Chemistry and Catalysis, Elsevier,
56 M. L. Toebes, J. M. P. van Heeswijk, J. H. Bitter, A. Jos van Amsterdam, 1989.
Dillen and K. P. de Jong, Carbon, 2004, 42, 307–315. 70 H. Z. Liu, T. Jiang, B. X. Han, S. G. Liang and Y. X. Zhou,
57 R. Cid and G. Pecchi, Appl. Catal., 1985, 14, 15–21. Science, 2009, 326, 1250–1252.
58 S. D. Gardner, C. S. K. Singamsetty, G. L. Booth, G.-R. He 71 K. Leiva, C. Sepulveda, R. García, D. Laurenti, M. Vrinat, C.
and C. U. Pittman, Carbon, 1995, 33, 587–595. Geantet and N. Escalona, Appl. Catal., A, 2015, 505, 302–308.
59 P. Arnoldy, E. M. van Oers, O. S. L. Bruinsma, V. H. J. de 72 N. Ota, M. Tamura, Y. Nakagawa, K. Okumura and K.
Beer and J. A. Moulijn, J. Catal., 1985, 93, 231–245. Tomishige, Angew. Chem., Int. Ed., 2015, 54, 1897–1900.
60 Y. Yuan and Y. Iwasawa, J. Phys. Chem. B, 2002, 106, 73 E.-J. Shin and M. A. Keane, J. Catal., 1998, 173, 450–459.
4441–4449. 74 V. N. Bui, D. Laurenti, P. Afanasiev and C. Geantet, Appl.
61 J. Okal, W. Tylus and L. Kȩpiński, J. Catal., 2004, 225, 498–509. Catal., B, 2011, 101, 239–245.
62 W. T. Tysoe, F. Zaera and G. A. Somorjai, Surf. Sci., 75 T. Nimmanwudipong, R. Runnebaum, D. Block and B.
1988, 200, 1–14. Gates, Catal. Lett., 2011, 141, 779–783.
63 N. R. Murphy, R. C. Gallagher, L. Sun, J. G. Jones and J. T. 76 T. Prasomsri, M. Shetty, K. Murugappan and Y. Roman-
Grant, Opt. Mater., 2015, 45, 191–196. Leshkov, Energy Environ. Sci., 2014, 7, 2660–2669.
64 A. Naor, N. Eliaz, L. Burstein and E. Gileadi, Electrochem. 77 J. Chang, T. Danuthai, S. Dewiyanti, C. Wang and A. Borgna,
Solid-State Lett., 2010, 13, D91–D93. ChemCatChem, 2013, 5, 3041–3049.
65 S. Lwin, C. Keturakis, J. Handzlik, P. Sautet, Y. Li, A. I. 78 A. Ausavasukhi, Y. Huang, A. T. To, T. Sooknoi and D. E.
Frenkel and I. E. Wachs, ACS Catal., 2015, 5, 1432–1444. Resasco, J. Catal., 2012, 290, 90–100.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2016

You might also like