Influence on nickel particle size on the hydrodeoxygenation of phenol over Ni/SiO2

Article abstract of DOI:10.1016/j.cattod.2015.08.022

Hydrodeoxygenation (HDO) of phenol over nickel nano-particles of different size (5-22 nm) supported on SiO₂ has been investigated in a batch reactor at 275 °C and 100 bar. Deoxygenation was only observed as a consecutive step of initial hydroge

Full text of DOI:10.1016/j.cattod.2015.08.022

Catalysis Today 259 (2016) 277–284
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Influence on nickel particle size on the hydrodeoxygenation of phenol
over Ni/SiO2
a
b
a
a,∗
Peter M. Mortensen , Jan-Dierk Grunwaldt , Peter A. Jensen , Anker D. Jensen
a
Department of Chemical and Biochemical Engineering, Technical University of Denmark, Søltofts Plads, 2800 Kgs. Lyngby, Denmark
Institute for Chemical Technology and Polymer Chemistry and Institute of Catalysis Research and Technology, Karlsruhe Institute of Technology,
Engesserstrasse 20, 76131 Karlsruhe, Germany
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Hydrodeoxygenation (HDO) of phenol over nickel nano-particles of different size (5–22 nm) supported
on SiO2 has been investigated in a batch reactor at 275 C and 100 bar. Deoxygenation was only observed
Received 17 December 2014
Received in revised form 24 June 2015
Accepted 11 August 2015
◦
as a consecutive step of initial hydrogenation of phenol at the given conditions. Both the hydrogenation
and deoxygenation reaction were found to be Ni-particle size dependent. Rapid hydrogenation of phenol
to cyclohexanol was achieved over the catalysts with large particles, while the rate of deoxygenation
of cyclohexanol was slow. For the catalysts with small Ni particles, the opposite behavior was observed
Specifically, the turn over frequency (TOF) of hydrogenation was 85 times slower for 5 nm particles
than for 22 nm particles. On the contrary, the TOF of cyclohexanol deoxygenation increased by a factor
of 20 when decreasing the particle size from 20 nm to 5 nm. A simple kinetic model showed that the
rate limiting step for phenol HDO shifted from deoxygenation to hydrogenation when the particle size
was below 9–10 nm. Surface site population theory evidenced that the deoxygenation reactions were
favored on step/corner sites, giving higher deoxygenation rates at small particles. For hydrogenation, the
influence of particle size on the rate could be related to the size of the Ni facets with larger facets thus
being better.
Available online 11 September 2015
Keywords:
Bio-oil
Heterogeneous catalysis
Hydrodeoxygenation (HDO)
Kinetics
Phenol
Structure sensitivity
©
2015 Elsevier B.V. All rights reserved.
1
. Introduction
requires the presence of a catalyst, a temperature in the order of
◦
2
00–400 C, and typically a pressure of 70–200 bar [1].
Flash pyrolysis of biomass followed by hydrodeoxygenation
Bio-oil consists of a vast number of oxygenated species and
(
HDO) of the produced bio-oil has been identified as a prospective
therefore many different reactions occur in HDO. It has been
observed that a part of the oil (simple molecules such as short chain
ketones) can easily be deoxygenated by direct removal of the oxy-
gen functionality [5,6]. However, to deoxygenate more complex
molecules, such as phenolics, an initial hydrogenation of the aro-
matic part of the molecule may be needed to make the oxygen
group more susceptible for reaction with hydrogen [7]. Gener-
ally, phenol and phenolic derivatives are a fairly abundant part of
bio-oil [8], which has been found as one of the more persistent oxy-
compounds to deoxygenate [5,9]. This makes phenol an interesting
model compound when investigating catalysts for HDO.
route for CO2 neutral engine fuel production [1]. Through pyroly-
sis, practically any source of biomass can be converted into bio-oil
[
2]. Although much more energy dense than the original biomass,
bio-oil has a low heating value compared to crude oil, a low shelf
storage life, is viscous, and polar, making it unsuitable as an engine
fuel. These characteristics are all associated with high contents of
water and chemically bound oxygen in the bio-oil. However, as
the oil has a higher volumetric energy density and is easy to han-
dle relative to biomass it is more suitable for transport and further
processing [1,3,4].
It would be advantageous if bio-oil can be upgraded to an oil
similar to conventional crude oil using the HDO route, where hydro-
gen is used to remove the oxygen functionality in the bio-oil. This
One of the major challenges in HDO is to find a suitable cata-
lyst. Recent work has shown that nickel based catalysts have high
activity for HDO reactions [10–17].
HDO of phenol over metallic catalysts and temperatures below
◦
3
00 C preferentially proceeds through a sequential reaction path
as shown in Fig. 1, where hydrogenation of the aromatic ring takes
place initially, producing cyclohexanone. This is rapidly hydro-
genated to cyclohexanol. Deoxygenation of the cyclohexanol then
∗ Corresponding author.
E-mail address: aj@kt.dtu.dk (A.D. Jensen).
http://dx.doi.org/10.1016/j.cattod.2015.08.022
0
920-5861/© 2015 Elsevier B.V. All rights reserved.

2
78
P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
particles, due to the direct reduction of the nickel nitrate catalyst
precursor [28].
Temperature programmed reduction (TPR) data can be found in
◦
the ESI showing that complete reduction is achieved around 380 C.
2.2. Catalyst testing
The experiments were performed in a 300 ml batch reactor from
Parr (type 4566 of Hastelloy C steel). In an experiment 1 g of cata-
lyst and 50 g of phenol (Sigma–Aldrich, ≥99%) were poured into the
reactor. The mixture was stirred with a propeller at 380–390 rpm
Fig. 1. Reaction scheme of HDO of phenol under mild conditions. Solid arrows indi-
cate main pathways; while the dotted arrows indicate the steps of the kinetic model
(
see Section 2).
◦
and heated to 275 C in a hydrogen atmosphere, giving a final pres-
◦
sure of 100 bar. The heating rate was around 15 C/min. During the
experiments, hydrogen was added continuously to maintain the
pressure of 100 bar. The experiments had a total reaction time of
proceeds through either hydrogenolysis to cyclohexane or dehy-
dration forming cyclohexene, which is readily hydrogenated into
the final product cyclohexane [18,19]. This reaction path is fairly
well established in the literature under mild conditions using noble
metal/nickel catalysts. This mechanism has been reported, e.g.,
for HDO of phenol over Pd/C [20,21], Pt/C [22], Ni/HZSM-5 [23],
Ni–MoS /Al O [24], and nickel based catalysts [16,18].
In the current work HDO of phenol over Ni/SiO2 catalysts has
been investigated with special emphasis on understanding the
influence of metal particle size and the relationship between the
hydrogenation and deoxygenation reactions taking place for HDO
of phenol.
5
h. To stop the experiment, the reactor was placed in an ice bath
to cool the reactor within 5 min to room temperature. The start of
the experiment was taken at the time when the heater was turned
on (heating took ca. 20 min) and the end of the experiment was
regarded as the time when the reactor was lowered into the ice
bath. By threefold repetition of an experiment, it was found that
this procedure had an uncertainty in the measured yields within
2
2
3
±
2 mol%, corresponding to less than 5% as relative standard devia-
tion. Overall there was a good repeatability of the experiments.
A blank experiment without catalyst with 10 g of phenol and
◦
4
0 ml of H O at 275 C and 100 bar for 4 h resulted in a conver-
2
sion of only 0.3%, showing that the reactor was not catalytically
active. Calculation of Mears’ and Weisz–Prater criteria [29] indi-
cated absence of mass transfer limitations in this system with the
observed reaction rates. Details on the evaluation can be found in
the ESI.
2
. Experimental
2.1. Catalyst synthesis
In some cases shorter experiments were performed to obtain
conversions well below 100% for determination of kinetic parame-
ters. For this purpose the batch reactor was initially heated without
stirring, which decreased the rate of reaction to practically zero,
until the desired temperature was reached. At this point the stirring
was started at 380–390 rpm (start of experiment) and the reac-
tion could be made at close to isothermal conditions. This allowed
measuring the activity in short experiments of ca. 15 min.
5
wt% Ni/SiO2 was prepared by incipient wetness impregnation
using Ni(NO ) ·6H O (Sigma–Aldrich, ≥97.0%) as precursor. Silica
3
2
2
was supplied by Saint-Gobain NorPro, type SS6*138 with a purity
2
of ≥99.5%, a specific surface area of 250 m /g, and a pore volume
of 1.0 ml/g. Before impregnation, the SiO was grinded to a particle
2
size of 63–125 ␮m. The SiO2 was impregnated with a 0.90 mol/l
solution of Ni(NO ) ·6H O in water in one step. After impregnation,
3
2
2
◦
the sample was dried at 110 C for 12 h.
In order to make catalysts with different particle size but same
composition, the catalysts were pre-treated/calcined and reduced
in different ways prior to the catalytic activity test:
2.3. Product analysis
Analysis of the liquid product was done with a Shimadzu
GCMS/FID-QP2010UltraEi fitted with a Supelco Equity-5 column.
The products were identified using a mass spectrometer (MS) and
quantified with a flame ionization detector (FID). External stan-
dards were prepared for phenol, cyclohexanol, cyclohexanone, and
cyclohexane. The concentrations of the remaining peaks were cal-
culated from the FID on the basis of the effective carbon number
method [30], where the concentration of a compound is found as:
◦
•
•
•
Red. 1: The catalyst was calcined initially at 400 C in an oven and
◦
then reduced in the batch reactor at 395 C (catalyst temperature)
and 7 bar of H in a stagnant gas atmosphere for 2 h. This method
is expected to yield large nickel particles due to the high pressure
and the presence of water [25–27].
Red. 2: The catalyst was calcined initially at 400 C in an oven
and then reduced in the batch reactor with a flow of 1 Nl/min
H2 at 395 C (catalyst temperature) and 5 bar of hydrogen for 2 h.
This method is expected to yield intermediate size nickel particles
since water is continuously removed.
Red. 3: The catalyst was calcined initially at 400 C in an oven
2
◦
◦
Ai
Ci = Cref · A
ꢀeff,ref
·
(1)
ref
ꢀeff,i
◦
Here C is the concentration, A the area of the peak in the FID
spectrum, and ꢀ the effective carbon number. Index i refers to the
compound with the unknown concentration and index ref refers
to a reference compound where the concentration is known. In all
calculations based on this formula cyclohexanol was used as refer-
ence. The effective carbon number was taken from the review by
Schofield [30].
and then reduced in a continuous flow reactor with a flow of
50 Nml/min H2 and 250 Nml/min N2 at 400 C and 1 bar of
◦
2
hydrogen for 2 h and then transferred directly to the batch reac-
tor. This method is expected to yield smaller nickel particles than
Red. 2 due to the milder conditions and more controlled removal
of water.
•
Red. 4: Reduction of un-calcined Ni(NO ) /SiO in a continuous
The conversion, X, was calculated as:
3
2
2
ꢀ
ꢁ
flow reactor with a flow of 250 Nml/min H and 250 Nml/min
2
CPhenol · Vfinal
n0,Phenol
◦
N2 at 400 C and 1 bar for 2 h and then transferred directly to
X = 1 −
· 100%
(2)
the batch reactor. This method is expected to yield small nickel

P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
279
Table 1
Here Vfinal is the final liquid volume and n0,Phenol the moles of
Overview of results from the 5 wt% Ni/SiO2 catalysts with different particle sizes. dNi
is the nickel crystallite size measured by ETEM, k1 is the rate constant for hydrogena-
tion, k2 is the rate constant for deoxygenation, and ꢁC is the deviation in the carbon
balance. The experiments were made with 1 g of catalyst in 50 g phenol. T = 275 C,
P = 100 bar, reaction time = 5 h.
phenol prior to reaction.
The yields (Y ) of relevant products were calculated as:
i
◦
ꢀi · ni
Yi = 6
· 100%
(3)
· n0
,Phenol
Sample
dNi [nm]
k1 [ml/(kgcat min)]
k2 [ml/(kgcat min)]
ꢁC [%]
Here n is the moles of product i after reaction and ꢀ is the
number of carbon atoms in compound i.
i
i
Red. 1
Red. 2
Red. 3
Red. 4
22 ± 3
14 ± 3
10 ± 3
5 ± 2
1834a
8
70
175
645
−0.7
−5.5
−0.2
−1.6
b
898
The selectivity (S ) of a compound (i) was calculated as:
i
224
78
Yi
Si =
· 100%
(4)
a
◦
Determined from a 15 min isothermal experiment at 275 C with 40% phenol
X
conversion.
All calculations of X, Y , and S were corrected for loss of mass
b
◦
i
i
Determined from a 15 min isothermal experiment at 275 C with 21% phenol
conversion.
from transferring and filtration processes, which was quantified as
the average of a number of blank tests.
In our previous work [16] we developed and validated a simple
2.4. Catalyst characterization
kinetic model based on these two reactions to describe the HDO of
phenol quantitatively by the rate constants for hydrogenation (k1)
A Quantachrome iQ2 was used for measurement of the specific
and deoxygenation (k ). This model will be used to quantify the
2
surface area using BET theory [31]. Nitrogen at its boiling point was
experimental data in this work as well.
3. Results and discussion
3.1. Results
used in the p/p range from 0.05–0.3 to construct a seven point BET
0
2
plot. The produced catalyst had a specific surface area of 210 m /g
independent on preparation method.
Environmental transmission electron microscopy (ETEM) was
performed using a Titan E-Cell 80-300ST TEM where images were
obtained at reducing conditions. The samples were crushed, slur-
ried with ethanol, and deposited on a copper grid covered with
Ni/SiO2 was prepared using four different preparation proce-
dures (Red. 1–4, as described in Section 2) to investigate the effect
of the metal particle size on the catalytic performance. As nickel
is very prone to oxidation, the nickel particle size was analyzed in
lacey carbon film. In the microscope these samples were heated
◦
to 450 C and reduced in a flow of 5 Nml/min H at a pressure of
2
◦
2
.5 mbar. The average nickel particle size was determined by mea-
an environmental TEM after re-reduction at 450 C and 2.5 mbar
suring the size of >200 particles on the images and subsequently
calculating the number based average.
Based on the ETEM determination of the average particle size,
the dispersion (D) was calculated as [32]:
hydrogen. Fig. 2 shows selected images of each of the four catalysts
and the particle size distribution. The number based average par-
ticle size of the four samples was 22 ± 3 nm, 14 ± 3 nm, 10 ± 3 nm,
and 5 ± 2 nm for the catalysts Red. 1, Red. 2, Red. 3, and Red. 4,
respectively. The Gaussian function fitted to the size distribution is
shown to the right of the respective ETEM images in Fig. 2. Overall
the four samples represent very different average particle sizes and
relatively narrow particle size distributions.
5
.01 · dM,at
D =
D =
, dM > 6 nm
(5)
(6)
dM
ꢀ
ꢁ
0
.81
3.32 · dM,at
, 1 nm < dM < 6 nm
The different catalysts were tested in the batch reactor setup
dM
◦
with 50 g of phenol as reactant at 275 C and 100 bar for 5 h. Fig. 3
Here dM,at is the atomic diameter of the relevant metal
shows the conversion of phenol and the yields of cyclohexane
and cyclohexanol for these experiments. The conversion of phenol
increased as a function of the particle size. 41% conversion of phe-
nol was obtained for the catalyst with 5 nm Ni particles, while ca.
(
dNi,at = 2.48 A˚ ) and dM is the average particle size of the metal.
From the dispersion the number of surface sites can be deter-
mined as:
1
00% conversion was obtained for both the 14 nm and 22 nm cases.
wNi
NT =
(7)
(8)
It can further be seen that the yield of cyclohexanol increased with
increasing particle size, while the cyclohexane yield had a maxi-
mum for the 10 nm particle size and was lowest in the case of the
largest particles. The case of the Ni/SiO2 sample with the lowest
particle size had a very high selectivity towards cyclohexane, with
practically no cyclohexanol in the products.
Generally, cyclohexane and cyclohexanol constituted the main
products from the reaction with a combined selectivity in the order
of 80–90%. Besides these, cyclohexanone (up to 10% selectivity),
cyclohexene (up to 5% selectivity), and bicyclic compounds such
as dicyclohexyl ether and bicyclohexyl (up to 5% selectivity) were
identified as the major byproducts. These observations support the
reaction scheme in Fig. 1. As presented in Table 1, the carbon bal-
ance closed well in all experiments. Only the experiment using the
catalyst Red. 2 had a loss of more than 6% carbon, but the other
experiments were closed within 2%.
MNi
NS = D · NT
Here NT is the total number of nickel atoms on the catalyst, N is
S
the number of surface atoms, wNi is the mass fraction of nickel on
the catalyst, and MNi is the molar mass of nickel. This is later also
used to calculate the turnover frequency (TOF) of the catalyst (cf.
Section 3.2).
2
.5. Kinetic model
We have previously shown [16] that the reaction scheme of
Fig. 1 can be simplified to a system of only two reactions at the
investigated conditions (development of the reaction scheme for
the catalyst and derivation of the kinetic model is found in the ESI):
The data in Fig. 3 were used to fit values of k1 and k2 in the
kinetic model introduced in the experimental section. For Red. 1
and Red. 2 (where 100% conversion was achieved during the 5 h
experiments presented in Fig. 3), additional experiments of 15 min
duration were performed to obtain lower conversions and allow
k
1
Reaction 1 : Phenol + 3H →Cyclohexanol
2
k
2
Reaction 2 : Cyclohexanol + H →Cyclohexane
2

2
80
P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
Fig. 2. Environmental transmission microscopy images of the 5 wt% Ni/SiO2 catalysts showing particle sizes and particle size distribution and Gaussian fit of Red. 1 (a), Red.
◦
2
(b), Red. 3 (c), and Red. 4 (d). All samples were reduced in the ETEM at 450 C and 2.5 mbar H2 prior to the analysis.
determination of k (cf. Table 1). The rate constants are shown as a
function of the nickel particle size in Fig. 4. The rate of hydrogena-
particle size. Linking these observations to Fig. 3, it follows that the
low conversion of phenol to cyclohexanol for Red. 4 was accompa-
nied by a very rapid deoxygenation, explaining the high selectivity
toward cyclohexane. For this catalyst the rate constant for deoxy-
genation was higher than the rate constant for hydrogenation. Thus,
1
tion (k ) increased as a function of particle size by about two orders
1
of magnitude from the largest to the smallest particle size. On the
contrary, the rate of deoxygenation (k ) decreased with increasing
2
Fig. 3. Conversion of phenol (X) and yields of cyclohexanol and cyclohexane as a function of the nickel particle sizes on 5 wt% Ni/SiO2 catalysts. The experiments were made
◦
with 1 g of catalyst in 50 g phenol. T = 275 C, P = 100 bar, reaction time = 5 h.

P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
281
Fig. 4. k1 and k2 as a function of the nickel particle sizes of the 5 wt% Ni/SiO2 catalysts. k1 in the two cases with 100% conversion (cf. Fig. 3) was determined by additional
shorter 15 min experiments. k1 represents the rate constant for hydrogenation and k2 represents the rate constant for deoxygenation. The experiments were made with 1 g
◦
of catalyst in 50 g phenol. Typical conditions: T = 275 C, P = 100 bar, reaction time = 5 h.
a shift in the rate determining step occurs for Ni/SiO catalysts with
Here k is the overall rate constant in ml/(kgcat min), k is the
2
i
nickel particles around ca. 9–10 nm, as seen from the crossing of the
two curves in Fig. 4. If the particle size is above 9–10 nm, the slowest
step is the deoxygenation, but below this range the hydrogenation
reaction is the slowest.
rate constant at the specific site in ml/(site min), N is the number
of sites of type i in term of sites/kgcat, NS is the total number of
i
exposed surface atoms in sites/kgcat, and x is the fraction of the
i
different sites. This can be linked to the turn over frequency (TOF):
k
◦
3
.2. Theoretical distribution of nickel sites
TOF = C · NA · N
(12)
(13)
S
Further interpretation of the results in Figs. 3 and 4 requires
◦
TOF = C · NA · (kfacet · xfacet + kstep · xstep + kcorner · xcorner)
Here TOF is the turn over frequency in s 1, C is the standard
concentration of 1 mol/l (as recommended by Kozuch and Martin
[33]), and NA is the Avogadro constant in mol . As both a rate
constant for the hydrogenation reaction (k ) and the deoxygenation
analysis of the crystal structure and the availability of different sur-
face sites. Three types of sites are generally considered: facet sites,
step sites, and corner sites, as visualized in Fig. 5. The overall (mea-
sured) rate constant is the result of the contribution from each of
these sites, leading to the following equation:
−
◦
−
1
1
reaction (k ) have been defined, the TOF for hydrogenation (TOFHyd)
and the TOF for deoxygenation (TOFDeox) can now be defined on the
basis of Eq. (12).
Icosahedrons are one of the preferred crystal shapes for
nickel nanoparticles, as shown by Cleveland and Landman by the
embedded atom method [34] and as expected for an inert support
like silica where no strong metal-support interaction is expected.
2
k = kfacet · Nfacet + kstep · Nstep + kcorner · Ncorner
(9)
k
Nfacet
Nstep
Ncorner
=
=
kfacet
·
+ kstep ·
+ kcorner ·
(10)
NS
NS
NS
NS
k
kfacet · xfacet + kstep · xstep + kcorner · xcorner
(11)
NS
Fig. 5. Fraction of step, corner, and facet surface sites (xi) and the total amount of surface sites (NS) as a function of dispersion of icosahedron crystals. Calculations were
made on the basis of the relations derived by Benfield [35].

2
82
P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
Fig. 6. TOFHyd and TOFDeox over Ni/SiO2 catalysts plotted as a function of dispersion. TOFHyd represents the TOF for hydrogenation and TOFDeox represents the TOF for
◦
deoxygenation. The experiments were made with 1 g of catalyst in 50 g phenol. T = 275 C, P = 100 bar, reaction time = 5 h.
The distribution of step, corner, and facet sites on an icosahedron
crystal has been derived by Benfield [35] and is plotted as a function
of dispersion in Fig. 5. The fraction of facet sites decreases with dis-
persion (approximately linearly), the fraction of step sites increases
with dispersion (approximately linearly), and the fraction of cor-
ner sites slowly increases with dispersion (approximately with the
square of the dispersion). Coupling this to the TOF defined in Eq.
others. Thus, structure sensitive reactions are well established.
Additionally, it has also been observed for other combinations of
catalysts and reactants for HDO that the rate of deoxygenation
can be increased by using small particles of the active metal on
supported catalysts [17,19].
Based on the experimental results and by analogy to the obser-
vations from other catalytic systems, we conclude that in the HDO
process on nickel catalysts, the low-coordinated step and corner
site atoms interact more readily with the oxygen group of the oxy-
compound to facilitate the deoxygenation. Thus, the C O bond
breaking reaction will more readily take place on this type of
site. Nevertheless, at the applied temperature, the ring still needs
to be hydrogenated before the C O bond is weak enough to be
broken.
In Fig. 6, TOFHyd decreases as a function of dispersion, but not
linearly. None of the surface site distributions follow a similar trend
(cf. Fig. 5), indicating that this reaction is probably not directly
linked to a specific type of site. At most, some similarity to the
behavior of the fraction of facet sites can be drawn as this decreases
linearly with dispersion.
(
13), shows that the TOF will decrease as a function of the dispersion
if the reaction primarily is taking place on the facet sites. Contrary,
the TOF will increase approximately linearly as a function of disper-
sion if the reaction preferentially occurs on the step sites and if the
reaction preferably takes place on the corner sites it will increase
with the square of the dispersion.
Additionally, the total amount of surface sites available per mass
of catalyst is plotted in Fig. 5 to clearly show that the amount of
available sites scales with the dispersion; at high dispersions more
sites are available.
4
. Discussion
Newman et al. [19] also observed that the surface site normal-
ized rate of hydrogenation of phenol over a range of supported
ruthenium catalysts increased with an increasing particle size of
ruthenium. They suggested that large Ru particles (>ca. 10 nm)
could facilitate interaction with the aromatic ring leading to its
hydrogenation to cyclohexanol and subsequently cyclohexane. In
our previous work [16] we have found, that hydrogenation of
On the basis of Eq. (12) and the kinetic data in Fig. 4, TOFHyd
and TOFDeox can be plotted as a function of dispersion, as shown
in Fig. 6. TOFDeox increases almost linearly as a function of disper-
sion. This is similar to the observed trend of the fraction of step
sites in Fig. 5. As the development in the fraction of corner sites
is limited in the relevant dispersion range (cf. Fig. 5), the contri-
bution from this type of site is difficult to deduce, but most likely
both step and corner sites could catalyze the reaction as both are
low coordinated sites leading to the observed structure-sensitivity
of the reaction. Our previous work [16] evidenced that the deox-
ygenation of cyclohexanol takes place by direct adsorption of the
alcohol on the nickel particle, and from Fig. 6 it therefore follows
that the low coordinated nickel sites facilitate the adsorption and
subsequent hydrogenolysis/deoxygenation.
Structure sensitivity relations have also been observed in:
steam reforming over supported noble metal and nickel cat-
alysts [36–38], the methanation reaction over nickel catalysts
39,40], Fischer–Tropsch synthesis on cobalt and ruthenium cata-
lysts [41–44], CO oxidation over supported gold catalysts [45–48],
and CO and NO oxidation over Pt based catalysts [49], among
◦
phenol over nickel based catalyst at temperatures below 300 C
required an oxide support. For a Ni/C catalyst, practically no conver-
sion took place. We proposed that vacancies in the oxide support
facilitated the adsorption for the phenol followed by hydrogenation
by hydrogen donation from the nickel particles [16]. The present
results further show that the hydrogenation of the ring is faster for
large than for small particles. This may be due to simple geomet-
ric reasons, i.e. that while the phenol is adsorbed on the support
it is beneficial with larger facets from which the hydrogen can be
donated. It is also possible that for Ni particles of sufficient size,
direct adsorption on the Ni particle becomes possible opening up a
parallel path for hydrogenation relative to the small particles that
require the assistance of the oxidic support. However, additional
[

P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
283
analysis of the reaction is required to thoroughly understand the
mechanism.
We thank Thomas Willum Hansen from Center for Electron
Nanoscopy (CEN) at DTU for assistance with the ETEM measure-
ments and interpretation.
The rate of hydrogenation was increased by one order of magni-
tude by increasing the particle size (cf. Table 1), but proper choice of
oxide support can lead to an increase in hydrogenation rate by sev-
eral orders of magnitude [16]. Thus, proper choice of support has a
larger effect when optimizing the hydrogenation rate compared to
particle size.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.cattod.2015.08.
0
22.
5
. Conclusion
The influence of nickel particle size for a Ni/SiO -based catalyst
References
2
on the catalytic HDO of phenol has been investigated for particle
sizes varying from 5 nm to 22 nm. A strong particle size and thus
structure sensitivity was observed. Hydrogenation of the aromatic
ring to form cyclohexanol was required before deoxygenation could
take place. The large particles were observed to be very active for
hydrogenation of phenol, giving 100% conversion of phenol into
[1] P.M. Mortensen, J.-D. Grunwaldt, P.A. Jensen, K.G. Knudsen, A.D. Jensen, Appl.
Catal. A: Gen. 407 (2011) 1–19.
[
[
2] T. Bridgwater, J. Sci. Food Agric. 86 (2006) 1755–1768.
3] Q. Zhang, J. Chang, T. Wang, Y. Xu, Energy Convers. Manag. 48 (2007) 87–92.
[4] A.V. Bridgwater, S. Czernik, J. Diebold, D. Meier, A. Oasmaa, C. Peakocke, J.
Piskorz, D. Radlein, Fast Pyrolysis of Biomass: A Handbook, CPL Press,
Newbury, 1999.
[
[
5] E. Furimsky, Appl. Catal. A: Gen. 199 (2000) 144–190.
6] P. Grange, E. Laurent, R. Maggi, A. Centeno, B. Delmon, Catal. Today 29 (1996)
297–301.
◦
primarily cyclohexanol in 5 h reaction time at 275 C. By decreasing
the particle size the conversion of phenol decreased and the yield
of cyclohexane increased, reaching a maximum yield of 38% for a
catalyst with a nickel particle size of 10 nm.
A simple kinetic model was set up to analyze the observations
from the experiments quantitatively in terms of hydrogenation and
deoxygenation rate constants. The model showed that the rate of
hydrogenation increased with increasing particle size (decreasing
dispersion) while the opposite trend was observed for the deox-
ygenation step. The rate determining step thus changed at about
[
[
[
7] R.H. Venderbosch, A.R. Ardiyanti, J. Wildschut, A. Oasmaa, H.J. Heeres, J. Chem.
Technol. Biotechnol. 85 (2010) 674–686.
8] M.S.A. Moraes, M.V. Migliorini, F.C. Damasceno, F. Georges, S. Almeida, C.A.
Zini, R.A. Jacques, E.B. Caramão, J. Anal. Appl. Pyrolysis 98 (2012) 51–64.
9] Q. Bu, H. Lei, A.H. Zacher, L. Wang, S. Ren, J. Liang, Y. Wei, Y. Liu, J. Tang, Q.
Zhang, R. Ruan, Bioresour. Technol. 124 (2012) 470–477.
[
[
10] V.A. Yakovlev, S.A. Khromova, O.V. Sherstyuk, V.O. Dundich, D.Y. Ermakov,
V.M. Novopashina, M.Y. Lebedev, O. Bulavchenko, V.N. Parmon, Catal. Today
144 (2009) 362–366.
11] A.R. Adriyanti, S.A. Khromova, R.H. Venderbosch, V.A. Yakovlev, H.J. Heeres,
Appl. Catal. B: Environ. 117–118 (2012) 105–117.
[12] M.V. Bykova, O.A. Bulavchenko, D.Y. Ermakov, M.Y. Lebedev, V.A. Yakovlev,
V.N. Parmon, Biocatalysis 3 (2011) 15–22.
13] M.V. Bykova, D.Y. Ermakov, V.V. Kaichev, O. Bulavchenko, A.A. Saraev, M.Y.
Lebedev, V.A. Yakovlev, Appl. Catal. B: Environ. 113–114 (2012) 296–307.
[14] M.V. Bykova, S.G. Zavarukhin, L.I. Trusov, V.A. Yakovlev, Kinet. Catal. 54 (2013)
40–48.
15] V.O. Dundich, S.A. Khromova, D.Y. Ermakov, M.Y. Lebedev, V.M. Novopashina,
V.G. Sister, A.I. Yakimchuk, V.A. Yakovlev, Kinet. Catal. 51 (2010) 728–734.
16] P.M. Mortensen, J.-D. Grunwaldt, P.A. Jensen, A.D. Jensen, ACS Catal. 3 (2013)
1774–1785.
17] P.M. Mortensen, D. Gardini, H.W.P. de Carvalho, C.D. Damsgaard, J.-D.
Grunwaldt, P.A. Jensen, J.B. Wagner, A.D. Jensen, Catal. Sci. Technol. 4 (2014)
9
–10 nm, where hydrogenation was the slower step for particles
below 9–10 nm while the deoxygenation step was rate-limiting
above this size. Thus, to achieve a high yield of the deoxygenated
product cyclohexane, good control of the particle size is needed and
a size of 9–10 nm seems optimal.
Comparison of the TOF for the deoxygenation reaction with that
calculated from surface site population theory for the nickel parti-
cles led to the conclusion that the rate of deoxygenation was linked
to the fraction of available step sites on the nickel particles. This
reaction therefore more readily took place on small nickel parti-
cles. Probably, both the adsorption of the oxy-compound and the
[
[
[
[
3672–3686.
[18] N. Shuikin, L. Erivanskaya, Russ. Chem. Rev. 29 (1960) 309–320.
19] C. Newman, X. Zhou, B. Goundie, I.T. Ghampson, R.A. Pollock, Z. Ross, M.C.
Wheeler, R.W. Meulenberg, R.N. Austin, B.G. Frederick, Appl. Catal. A: Gen.
[
C
O bond cleavage more readily occurs on a low coordinated nickel
atom.
Hydrogenation of the aromatic ring occurred best on large nickel
477 (2014) 64–74.
[20] C. Zhao, J. He, A.A. Lemonidou, X. Li, J.A. Lercher, J. Catal. 280 (2011) 8–16.
[
[
21] C. Zhao, Y. Kuo, A.A. Lemonidou, X. Li, J.A. Lercher, Angew. Chem. Int. Ed. 48
2009) 3987–3990.
particles (>10 nm). This is possibly due to faster hydrogenation of
the benzene ring from the larger facets while the phenol is attached
to the support, or that phenol can adsorp directly and become
hydrogenated on the large particles, opening up a parallel path
for hydrogenation. If nickel is present as small nickel nanoparti-
cles (less than ca. 10 nm), an oxidic support is required to assist in
the adsorption and hydrogenation of phenol.
(
22] H. Ohta, H. Kobayashi, K. Hara, A. Fukuoka, Chem. Commun. 47 (2011)
12209–12211.
23] C. Zhao, S. Kasakov, J. He, J.A. Lercher, J. Catal. 296 (2012) 12–23.
24] E.-M. Ryymin, M.L. Honkela, T.-R. Viljava, A.O.I. Krause, Appl. Catal. A: Gen.
[
[
389 (2010) 114–121.
[25] C.H. Bartholomew, Appl. Catal. A: Gen. 107 (1993) 1–57.
[26] J. Sehested, J. Catal. 217 (2003) 417–426.
27] J. Sehested, Catal. Today 111 (2006) 103–110.
28] C. Louis, Z.X. Cheng, M. Che, J. Phys. Chem. 97 (1993) 5703–5712.
[29] H.S. Fogler, Elements of Chemical Reaction Engineering, Prentice Hall, New
Jersey, 2006.
30] K. Schofield, Prog. Energy Combust. Sci. 34 (2008) 330–350.
31] P.L. Llewellyn, E. Bloch, S. Bourelly, Characterization of Solid Material and
Heterogenous Catalysts, Wiley-VCH, Weinheim, 2012, pp. 853–880.
[
[
Overall it is concluded that an intermediate nickel particle size
is needed for optimal phenol HDO over Ni/SiO2 at temperature
◦
below 300 C. Relatively small particles are required to facilitate
[
[
deoxygenation, but too small particles will have lower activity for
hydrogenation which will hinder the overall degree of deoxygen-
ation.
[32] A. Borodzinski, M. Bonarowska, Langmuir 13 (1997) 5613–5620.
[
[
[
33] S. Kozuch, J.M.L. Martin, ACS Catal. 2 (2012) 2787–2794.
34] C.L. Cleveland, U. Landman, J. Chem. Phys. 94 (1991) 7376–7396.
35] R.E. Benfield, J. Chem. Soc. Faraday Trans. 88 (1992) 1107–1110.
Acknowledgements
[36] G. Jones, J.G. Jakobsen, S.S. Shim, J. Kleis, M.P. Andersson, J. Rossmeisl, F.
Abild-Pedersen, T. Bligaard, S. Helveg, B. Hinnemann, J.R. Rostrup-Nielsen, I.
Chorkendorff, J. Sehested, J.K. Nørskov, J. Catal. 259 (2008) 147–160.
This work is part of the Combustion and Harmful Emission Con-
trol (CHEC) research center at the Department of Chemical and
Biochemical Engineering at the Technical University of Denmark
[
37] D.A.J.M. Lighthart, R.A. van Santen, E.J.M. Hensen, J. Catal. 280 (2011) 206–220.
[38] J. Wei, E. Iglesia, J. Phys. Chem. B 108 (2004) 4094–4103.
[
39] J. Engbæk, O. Lytken, J.H. Nielsen, I. Chorkendorff, Surf. Sci 602 (2008)
33–743.
7
(
DTU). The present work is financed by DTU and the Catalysis for
[
40] M.P. Andersson, F. Abild-Pedersen, I.N. Remediakis, T. Bligaard, G. Jones, J.
Engbæk, O. Lytken, S. Horch, J.H. Nielsen, J. Sehested, J.R. Rostrup-Nielsen, J.K.
Nørskov, I. Chork-endorff, J. Catal. 255 (2008) 6–19.
Sustainable Energy initiative (CASE), funded by the Danish Ministry
of Science, Technology and Innovation.

2
84
P.M. Mortensen et al. / Catalysis Today 259 (2016) 277–284
[
[
41] V. Ragaini, R. Carli, C.L. Bianchi, D. Lorenzetti, G. Predieri, P. Moggi, Appl. Catal.
A: Gen. 139 (1996) 31–42.
42] J.P. den Breejen, P.B. Radstake, G.L. Bezemer, J.H. Bitter, V. Frøseth, A. Holmen,
K.P. de Jong, J. Am. Chem. Soc. 131 (2009) 7197–7203.
[46] B. Hvolbæk, T.V.W. Janssens, B.S. Clausen, H. Falsig, C.H. Christensen, J.K.
Nørskov, Nanotoday 2 (2007) 14–18.
[47] T.V.W. Janssens, A. Carlsson, A. Puig-Molina, B.S. Clausen, J. Catal. 240 (2006)
108–113.
[
[
43] J. Yang, E.Z. Tveten, D. Chen, A. Holmen, Langmuir 26 (2010) 16558–16567.
44] J.M.G. Carballo, J. Yang, A. Holmen, S. García-Rodríguez, S. Rojas, M. Ojeda,
J.L.G. Fierro, J. Catal. 284 (2011) 102–108.
[48] N. Lopez, J.K. Nørskov, T.V.W. Janssens, A. Carlsson, A. Puig-Molina, B.S.
Clausen, J.-D. Grunwaldt, J. Catal. 225 (2004) 86.
[49] A. Boubnov, S. Dahl, E. Johnson, A.P. Molina, S.B. Simonsen, F.M. Cano, S.
Helveg, L.J. Lemus-Yegres, J.-D. Grunwaldt, Appl. Catal. B: Environ. 126 (2012)
315–325.
[
45] J.-D. Grunwaldt, M. Maciejewski, O.S. Becker, P. Fabrizioli, A. Baiker, J. Catal.
1
86 (1999) 458–469.