Structure sensitivity in the ruthenium nanoparticle catalyzed aqueous-phase Fischer-Tropsch reaction

Article abstract of DOI:10.1039/c4cy00709c

Low-temperature Fischer-Tropsch reaction data are reported for Ru nanoparticles suspended in the water phase. Their activity and selectivity strongly depends on particle size, when varied between 1 to 5 nm. Small particles display high oxygenates selectivity. The Anderson-Schulz-Flory (ASF) chain-growth probability for oxygenates is significantly lower than that observed for hydrocarbons. The chain growth parameter for hydrocarbon formation is independent of particle size. For oxygenates it is constant only for particles larger than 3 nm. Oxygenate and hydrocarbon formation occur on different sites. The ASF chain-growth probability for oxygenate formation increases with temperature. For very small 1.2 nm particles it shows a maximum as a function of temperature. This unusual temperature dependence is due to relatively slow CO dissociation compared to the rate of C-C bond formation. This journal is

Full text of DOI:10.1039/c4cy00709c

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Structure sensitivity in the ruthenium nanoparticle
catalyzed aqueous-phase Fischer–Tropsch
reaction
Cite this: Catal. Sci. Technol., 2014,
4, 3510
*
Xian-Yang Quek, Robert Pestman, Rutger A. van Santen and Emiel J. M. Hensen
Low-temperature Fischer–Tropsch reaction data are reported for Ru nanoparticles suspended in the
water phase. Their activity and selectivity strongly depends on particle size, when varied between 1 to
5 nm. Small particles display high oxygenates selectivity. The Anderson–Schulz–Flory (ASF) chain-growth
probability for oxygenates is significantly lower than that observed for hydrocarbons. The chain growth
parameter for hydrocarbon formation is independent of particle size. For oxygenates it is constant only
for particles larger than 3 nm. Oxygenate and hydrocarbon formation occur on different sites. The ASF
chain-growth probability for oxygenate formation increases with temperature. For very small 1.2 nm
particles it shows a maximum as a function of temperature. This unusual temperature dependence is due
to relatively slow CO dissociation compared to the rate of C–C bond formation.
Received 31st May 2014,
Accepted 27th June 2014
DOI: 10.1039/c4cy00709c
www.rsc.org/catalysis
then comprises three main surface reaction steps: (i) CO
dissociation and formation of CHx intermediates, (ii) chain
1. Introduction
Alternative carbon resources such as biomass, natural gas,
and coal are increasingly considered as possible feedstock for
the synthesis of liquid transportation fuels and chemicals
with the aim to reduce our dependence on crude oil reserves.
A versatile and increasingly popular catalytic route is their
conversion by steam reforming or gasification into synthesis
gas (a mixture of CO and H2) followed by the Fischer–
Tropsch (FT) reaction.1–5 Co, Fe and Ru are the typical transi-
tion metals that catalyze the reaction of synthesis gas into
long-chain hydrocarbons.2–4 The rate of CO dissociation on
less reactive transition metals such as Ni is too low and
methane is the main product.6 On metals such as Ir and Pt
the rate of CO dissociation is also slow. Rh has intermediate
reactivity and shows interesting oxygenate selectivity.1 Here,
we will present low-temperature catalytic performance data of
small unsupported Ru nanoparticles, which exhibit high
selectivity to long-chain oxygenates next to long-chain
hydrocarbons.
growth by CHx insertion and subsequent hydrogen transfer
steps, and (iii) termination to form the final product. On
surfaces of low reactivity, CO dissociation may proceed via
the H-assisted pathway9,10 instead of direct CO dissociation.11
For the chain-growth step involving CHx surface intermediates,
different routes have been proposed such as Brady–Pettit
alkyl12 and the Gaube–Maitlis alkelydene/alkenyl13 mecha-
nisms. There are three different termination steps possible,
namely H addition, β-H elimination and CO insertion respec-
tively giving rise to paraffins, olefins and oxygenates.
Within the carbide mechanism, a prerequisite for for-
mation of long-chain products is that the rate of methane
formation is low in order to provide sufficient CHx interme-
diates to maintain a high rate of chain growth.7 Also, for
production of long-chain products the rate of termination
should be relatively low compared to the rate of chain
growth. This implies that increased rate of chain-growth
termination by CO insertion or hydrogen transfer reactions
would decrease the rate of chain growth, and, accordingly,
the chain growth probability (α). When oxygenate and hydro-
carbon formation is the result of competition for the same
growing adsorbed hydrocarbon chain, increased selectivity
towards oxygenate formation implies that the intrinsic rate
for chain-growth termination to oxygenates is faster than that
for chain-growth termination to hydrocarbon. The chain
growth probabilities of both products would then be the
same. The main alternative proposal to the carbide mecha-
nism for the FT reaction is the one formulated by Pichler and
Among the various mechanisms proposed to account for
the formation of long-chain hydrocarbons,7 the carbide
mechanism has found most support in literature. According
to this mechanism, the building blocks for chain growth are
CHx surface intermediates.7,8 The Fischer–Tropsch reaction
Schuit Institute of Catalysis, Laboratory of Inorganic Materials Chemistry,
Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven,
The Netherlands. E-mail: e.j.m.hensen@tue.nl; Fax: +31 40 2455054;
Tel: +31 40 2475178
3510 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Schulz,14,15 who postulated that CO inserts into the growing
chain followed by C–O dissociation (CO insertion mechanism).
Then the rate of CO activation may be slow compared to the
rate of chain growth.7 In this case, high selectivity towards
oxygenates reduces chain-growth probability strongly com-
pared to formation of hydrocarbons.
these catalysts by synchrotron XRD-PDF and their overall per-
formance in aqueous phase FTS have been reported before.29
Here, we expand on this initial study by providing detailed
insight into the influence of particle size and reaction tem-
perature on the activity and selectivity in the aqueous-phase
FT reaction. The results will be discussed in the frame of the
carbide and CO insertion mechanisms.
Whereas Fischer–Tropsch catalysis can be considered
structure insensitive for relatively large transition metal
particles,16 a strong size dependence has been observed when
the particles become smaller than 10 nm.17–22 Recent gas-
phase SSITKA data for hydrocarbon production indicate that
with decreasing Ru nanoparticle size the relative fraction of
reactive sites decreases without change of the intrinsic
activity.22 For Co particles, there are strong indications that
also the intrinsic reactivity of the active centers changes with
particle size.20 The critical particle size below which the
CO consumption rate becomes structure sensitive is in the
7–10 nm range for Co17–20 and 6–10 nm for Ru.21,22 This
structure sensitive behavior has been attributed to the
decreasing stability of step-edge sites on the surface of
smaller particles. These sites are essential for low-barrier CO
dissociation.7,8 An alternative explanation is that CO or
C atoms adsorb too strongly on small particles, thus poison-
ing part of the catalytic surface.17–20 Salmeron et al. recently
suggested that particle size effects in the FT reaction are due
to differences in the rate of H2 dissociation, thus affecting
the rate of H-assisted CO dissociation.23 It is also known that
smaller particles produce more methane and have a lower
C5+ selectivity.17,22 Apart from particle size, the crystal struc-
ture of nanoparticles also affects their catalytic performance.
Recent studies indicate that hcp-Co is more active than
fcc-Co.24,25
Similar to commercial practice, most laboratory studies on
the FT reaction are carried out in fixed-bed or slurry-phase
reactors. An interesting alternative approach involving FT
reaction in the aqueous phase in a batch reactor was recently
demonstrated by Xiao et al.26 In this pioneering work, the
dominant reaction products were reported to be hydro-
carbons. We have recently performed similar experiments
focusing on the effect of the reaction temperature.27 We
found that Ru nanoparticle catalysts produce a mixture of
long-chain hydrocarbons and oxygenates under these anoma-
lous conditions (water solvent, temperatures below 200 °C).
Exceptionally high selectivity towards long-chain aldehydes
and alcohols was observed at temperatures below 200 °C.27
We have proposed that hydrocarbons form on sites with low
barrier for CO dissociation, likely involving step-edge sites,
whereas oxygenates are thought to form on sites with a
higher barrier for CO dissociation. Recently, Wang et al.
investigated the effect of ionic promoters on the Ru-catalyzed
FT reaction and also reported increased oxygenate selectivity.28
In this contribution, we report about the effect of Ru
nanoparticle size and reaction conditions on the activity and
selectivity in aqueous-phase FT. Polyvinylpyrrolidone (PVP)
stabilized Ru nanoparticles in the 1 to 6 nm range were pre-
pared and characterized. Bulk and surface characterization of
2. Experimental and computational
methods
2.1 Synthesis of materials
Polyvinylpyrrolidone (PVP, Mn = 10 000 g mol−1, Sigma
Aldrich) stabilized Ru nanoparticles were synthesized using
1,4-butanediol or H2 as the reducing agent. Ruthenium(III)
chloride hydrate (RuCl3·nH2O, Alfa Aesar) and ruthenium(III)
acetylacetonate (Ru(acac)3, Aldrich) were used as Ru pre-
cursors. The reduction by 1,4-butanediol was carried out in a
similar manner as earlier described for Rh.29 In a typical
synthesis (PVP/Ru molar ratio of 20), 30 mg of Ru(acac)3 and
0.17 g of PVP were dissolved in 2 mL of tetrahydrofuran
(THF, Sigma Aldrich) and 3 mL of 1,4-butanediol (Sigma-Aldrich).
The mixture was then added to 27 mL of 1,4-butanediol,
which was preheated at 225 °C, followed by refluxing in a N2
atmosphere for 2 h. The resulting black suspension was
thoroughly washed with acetone and diethyl ether. After
collecting the Ru nanoparticles by centrifugation, they were
redispersed in 1.5 mL distilled water. In cases where reduc-
tion was carried out in hydrogen, first 40 mg of RuCl3·xH2O
and 0.22 g of PVP were dissolved in 1 mL of distilled water
in a 10 mL autoclave. The autoclave was then pressurized
with 20 bar H2 followed by heating to 150 °C for 2 h under
vigorous stirring. The resultant black suspension was washed,
collected by centrifugation, and redispersed in 3 mL distilled
water. Catalysts are denoted by Ru-x where x stands for the
average particle size in nm.
2.2 Characterization
The concentration of the Ru nanoparticles dispersed in water
was determined by inductively coupled plasma atomic emis-
sion spectroscopy (ICP-AES) analysis performed on a Goffin
Meyvis SpectroCirus apparatus. Transmission electron micro-
graphs (TEM) were made with a FEI Tecnai 20 electron
microscope equipped with a LaB6 filament and operating at
an acceleration voltage of 200 kV. A small amount of the
nanoparticles was mixed with ethanol, dispersed over a
carbon-coated Cu grid and finally dried in air. The average
particle size and the standard deviation were determined by
analyzing at least 150 particles. FTIR spectra of adsorbed CO
were collected using a heated attenuated total reflectance
(HATR) flow cell from Spectra-Tech ARK with a Si 45° crystal
(cutoff at 1500 cm−1) in a Nicolet Protégé 460 FTIR spectrom-
eter equipped with a liquid-nitrogen cooled MCT detector.
Prior to the IR measurement, the Ru nanoparticles dispersed
in water were re-reduced in an autoclave at 150 °C with
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3511

View Article Online
Paper
Catalysis Science & Technology
20 bar H2 for 2 h under vigorous stirring. The Ru particles
were then spin-coated onto the Si crystal under N2 atmo-
sphere. The coated Si crystal was then mounted and sealed
in the ATR cell under N2 atmosphere. The sample was
dried by flowing 5 mL min−1 of He at 80 °C for 30 min. Sub-
sequently, CO was introduced in the cell at a flow of
2.5 mL min−1. After an isothermal period at 50 °C for 0.5 h,
the cell was heated to 150 °C. FTIR spectra were recorded
5 min after reaching 150 °C in absorbance mode at a resolu-
tion of 4 cm−1. A total of 32 scans were measured in this way,
followed by subtraction of the background spectrum of the Si
crystal. X-ray Absorption Spectra (XAS) were recorded in a
home-built transmission cell for liquid samples at the Dutch
Belgium Beamline (DUBBLE) of the European Synchrotron
Radiation Facility (ESRF, Grenoble, France). A droplet of the
liquid containing the Ru nanoparticles was placed between
Kapton windows. XAS spectra were collected in fluorescence
mode at the Ru K-edge with a 9-channel solid-state detector.
Energy selection was done by a double crystal Si(111) mono-
chromator. EXAFS analysis was performed with EXCURVE931
on k3-weighted unfiltered raw data using the curved wave
theory. Phase shifts were derived from ab initio calculations
using Hedin–Lundqvist exchange potentials and Von Barth
ground states as implemented in EXCUREVE98. In a typical
experiment, 28 μmol of Ru dispersed in 200 μL of H2O was
transferred into the XAS cell. The cells were then transferred
to the beamline and the EXAFS spectra were recorded at
room temperature.
column (30 m × 0.25 mm × 0.5 μm) and a flame ionization
detector (FID). Identification and quantification of linear
alkanes (C6 to C16), alcohols (C4 to C12) and aldehydes
(C4 to C12) was established with reference compounds and
p-cymene as the internal standard. GC response factors for
other hydrocarbons (C17+) and oxygenates (C13+) were esti-
mated by extrapolation. The identity of the products was also
verified with a GC-MS equipped (GC-MS, QP5050, Shimadzu)
with a Rxi-5 ms capillary column.
The rate of the FT reaction (r), turnover frequency (TOF),
and selectivity were calculated based on the number of moles
of carbon being formed in the products according to the
following formula:
Total mol C formed in all products
(1)
(2)
r 
Mass of Ru  reaction time
Total mol C formed in all products
TOF 
mol Ru
 reaction time

surface atoms 
mol of C HC/oxy  n



n
Selectivity to hydrocarbon or oxygenate 
Total mol C formed as products
(3)
with the mol Rusurface atoms = mol Ru × D.
The dispersion (D) was calculated assuming spherical
shapes by using the formula given by Scholten et al.30
6 M  
dp  NAv  metal
D 1021 
site
(4)
2.3 Catalytic activity measurements
Prior to reaction, an amount of catalyst (50 μmol Ru) was
dispersed in 3 mL of deionized water and re-reduced in a
10 mL stainless steel autoclave at 150 °C with 20 bar H2 for
2 h under vigorous stirring. After reduction, the autoclave
was cooled to room temperature in an ice bath before releasing
the pressure. The liquid phase FT reaction was carried out at
30 bar using reaction times between 3 and 24 h. The autoclave
was flushed 3 times with CO, before being pressurized with
CO followed by H2 to 30 bar (molar ratio H2/CO = 2). The
autoclave was sealed and the autoclave body was heated to
the reaction temperature with a band heater controlled by a
temperature controller. The reaction was terminated by
immersing the autoclave into an ice bath. The reaction time
varied between 3 to 24 h so as to maintain a maximum pres-
sure drop of 10 bar in the reactor. The gas-phase products
were analyzed with an Interscience Compact GC system,
equipped with an Al2O3/KCl column and a flame ionization
detector (FID). For analysis, the gas cap was flushed through
a 6-way valve allowing injection onto a gas chromatograph.
Methane, ethane, ethylene, propane and propylene were ana-
lyzed against a standard gas mixture. For analysis of the liq-
uid phase, an extraction was carried out with diethyl ether
containing p-cymene as an internal standard. The organic
phase containing the products was then analyzed with a
GC (QP5050, Shimadzu) equipped with a Rxi-5ms capillary
with M being the atomic weight of Ru, ρsite the surface
density (16.3 Ru atom nm−2), ρmetal the density of metal
(12.3 g cm−3), dp the Ru particle size and NAv the Avogadro's
number. Hydrocarbon selectivity includes alkanes and alkenes
and oxygenate selectivity aldehydes and alcohols.
3. Results
3.1 Catalyst characterization
Table 1 summarizes the conditions used to synthesize Ru
nanoparticles. The reduction method, reduction temperature
and the Ru source were varied to control the size of Ru nano-
particles between 1.2 nm to 5.2 nm (Fig. 1). In a standard
synthesis using polyol reduction of RuCl3 at 195 °C for 2 h,
we obtained on average 2.3 nm Ru particles (Ru-2.3). By using
Ru(acac)3 instead of RuCl3 under otherwise similar con-
ditions we obtained larger particles (Ru-3.7). We found that
increasing the temperature of reduction of RuCl3 by the
polyol was not an effective method to control the Ru particle
size. For instance, the average Ru particle size only increased
slightly from 2.3 nm to 2.5 nm when the reduction tempera-
ture was increased from 195 to 225 °C. In contrast, by polyol
reduction of the Ru(acac)3 precursor at different tempera-
tures we were able to obtain nanoparticle catalysts with
3512 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Table 1 Synthesis conditions and properties of Ru nanoparticlesa
Catalyst
Reduction methodb
T (°C)
Ru source
dpc (nm)
Fraction Ru0 (%)d
cRue (μmol mL−1
)
Ru-1.2
Ru-2.3
Ru-2.7
Ru-3.4
Ru-3.7
Ru-4.3
Ru-5.2
H2
145
195
165
175
195
215
225
RuCl3
RuCl3
1.2 0.4
2.3 0.3
2.7 0.3
3.4 0.5
3.7 0.5
4.3 0.6
5.2 0.8
90
85
—
80
90
—
92
35.6
35.8
35.9
34.1
33.2
29.0
32.6
Polyol
Polyol
Polyol
Polyol
Polyol
Polyol
Ru(acac)3
Ru(acac)3
Ru(acac)3
Ru(acac)3
Ru(acac)3
a
b
Ru nanoparticles synthesis parameters: PVP/Ru molar ratio = 20, Mn (PVP) = 10 000, re-dispersed in 1.5 mL H2O. Polyol: 1,4-butanediol as
c
d
reducing agent, H2: 20 bar H2 as reducing agent. Average particle size and standard deviation determined by TEM analysis. Determined
e
from XANES after second reduction with 20 bar H2 at 150 °C. Determined from ICP-AES elemental analysis for the synthesized Ru nano-
particles dispersed in 1.5 mL H2O (theoretical concentration: 50 μmol mL−1).
Fig. 1 Electron micrographs and particle size distributions for (a) Ru-1.2, (b) Ru-2.3, (c) Ru-2.7, (d) Ru-3.4, (e) Ru-3.7, (f) Ru-4.3, and (g) Ru-5.2.
average sizes in the range between 2.7 and 5.2 nm. Transmis-
sion electron micrographs (Fig. 1) show that Ru particles
between 1.2 nm and 3.4 nm are mostly spherical. Larger
particles tend to be more facetted. For the large particle
size catalysts, a small fraction of tetrahedrally-shaped parti-
cles are observed, which were included in the dispersion
determination.
Fig. 2a shows the Ru K-edge near-edge spectra for Ru-2.3
synthesized by polyol reduction (Ru-2.3-P) and after a second
reduction with 20 bar H2 (Ru-2.3-P-H2). The spectra show that
Ru-2.3-P and Ru-2.3-P-H2 contain both metallic and oxidic Ru.
The Ru0 content was estimated by fitting the experimental
near-edge spectra with a linear combination of reference spec-
tra of a metallic Ru foil and RuO2 powder. Separate X-ray
Photoelectron Spectroscopy (XPS) analyses did not reveal the
Fig. 2 Ru K-edge XANES spectra for (a) Ru-2.3 reduced in different
presence of chlorine, implying that the cationic Ru relates
to small amounts of RuO2. The Ru0 content for Ru-2.3-P
ways and (b) Ru nanoparticles of different particle size (black lines:
experimental data; circles: linear combination fitted model).
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3513

View Article Online
Paper
Catalysis Science & Technology
Table 2 Fit parameters of k3-weighted EXAFS data of a Ru foil and Ru
nanoparticle catalysts
and Ru-2.3-P-H2 were 65 and 85%, respectively. Consistent
with our previous studies using NaBH4 for reduction of
PVP-protected nanoparticles,27 Ru nanoparticles reduced by
polyol remain partially oxidic. The reduction degree can be
significantly increased by a second reduction step at 20 bar
H2 at 150 °C. The Ru K-edge near-edge spectra for differently
sized Ru particles after the second reduction step (Fig. 2b)
show that this difference in oxidation degree occurs for
all samples. The quality of the fitting procedure is good,
although one notes the small deviations in the whiteline
which is expected for small nanoparticles. The estimated Ru0
contents of the various nanoparticle catalysts are listed in
Table 1. Although small metal particles seem to be more
prone to oxidation than larger ones,31 our results indicate
that, after the second reduction step, the initial reduction
method and particle size had only minor influence on the Ru0
content. The average Ru0 content was found to be 87% with a
standard deviation of 5%.
EXAFS fit parametersa
Catalyst
Shell
N
R (Å)
Δσ2 (Å2)
ΔE0 (eV)
Ru foil
Ru-1.2
Ru-2.3
Ru–Ru
Ru–Ru
Ru–O
Ru–Ru
Ru–O
Ru–Ru
Ru–O
Ru–Ru
12
2.68
2.64
1.99
2.67
1.97
2.67
1.93
2.67
0.004
0.005
0.002
0.007
0.002
0.006
0.004
0.007
−7.9
−6.1
−0.5
6.1
1.4
6.9
1.1
7.5
1.0
8.7
Ru-3.7
Ru-5.2
−3.3
−2.6
a
Only first Ru–O and Ru–Ru shells fitted: Δk = 2.5–13.4 Å−1
;
estimated errors in R: 0.01 Å, N: 20%, and Δσ2: 10%.
Fig. 3 shows the experimental and fitted Fourier trans-
formed k3-weighted EXAFS and χ(k)·k3 functions for four Ru
nanoparticle catalysts of increasing size (fit parameters in
Table 2) as well as of the Ru foil. The Fourier transforms are
dominated by a Ru–Ru shell with a small contribution of a
Ru–O shell. The small Ru–O contribution indicates that
the Ru particles are not completely reduced in agreement
with the XANES analysis. The Ru–Ru coordination number
decreases when the particles become smaller. All Ru particles
exhibit Ru–Ru bond lengths comparable with the bulk Ru–Ru
distance (2.67 Å) with the exception of Ru-1.2 (2.64 Å).
Fig. 4 CO adsorption FTIR spectra measured at 150 °C for reduced
Fig. 4 presents the FTIR spectra of four Ru nanoparticle
catalysts collected at 150 °C after drying and CO adsorption.
Two main features can be observed in these spectra: a strong
band centered at 1990–2015 cm−1 region with a shoulder at
2050 cm−1 and a weak broad band in the 1900–1950 cm−1
region. The IR spectrum for Ru-1.2 shows an additional band
centered around 2116 cm−1. We assign these bands to the
Ru nanoparticles.
following species:32 bridge-bonded CO at 1900 to 1950 cm−1
,
linearly bonded CO at 1990–2015 cm−1, dicarbonyl CO species
adsorbed on Ru0 at 2050 cm−1 and CO adsorbed on Run+ at
2116 cm−1. A notable finding is that the band assigned to
linearly adsorbed CO sites is red-shifted when the nano-
particle size is decreased from 4.3 to 1.2 nm. The shift is
greatest when the average particle size decreases from 2.3 nm
(2009 cm−1) to 1.2 nm (1990 cm−1). CO is still linearly bonded
when the band shift from 2009 to 1990 cm−1, within the
range usually observed for linearly bonded CO.32 Therefore,
the shift is best explained by stronger adsorption of CO. It is
also seen that a shoulder appears at 2050 cm−1 with decreas-
ing particle size, which is due to Ru-dicarbonyl species. These
surface complexes are likely due to CO adsorption on very
low coordinated Ru surface atoms such as corner sites. The
increase of this band is greatest when the particle size is
decreased from 2.3 to 1.2 nm. This is consistent with the
decrease in coordination number for smaller particles as
determined by EXAFS. The band at 2116 cm−1 is due to the
presence of a small amount of oxidic Ru species on Ru-1.2. It
is also seen that the band due to bridge bonded CO is maxi-
mum for 3.4 nm Ru nanoparticles. The reason for this trend
is unclear, but signifies the different surface structure of the
nanoparticles depending on their size.
Fig. 3 Experimental (○) and fitted model (red lines) for (a) FT EXAFS
functions and (b) k3-weighted EXAFS oscillations of the Ru foil, Ru-1.2,
Ru-2.3, Ru-3.7 and Ru-5.2 after polyol and H2 reduction steps.
3514 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
3.2 Catalytic activity measurements
comparable to the trend earlier reported by us for 2 nm
Ru nanoparticles:27 increasing the temperature results in
decrease of the oxygenate selectivity. Oxygenates are the
product of termination by inserting CO into the growing
chain. Simulations indicate that high CO coverage is consis-
tent with hydrocarbon as well as oxygenate formation.33 The
temperature dependence of oxygenate selectivity can relate to
either (i) different CO activation barriers of the respective
sites that form oxygenates or hydrocarbons or (ii) the activa-
tion energy for termination to hydrocarbons being higher
than that to oxygenates.
It has been shown before that secondary reactions of
the reaction products of the FT reaction may significantly
affect the product distribution.35–37 Relevant to the present
work, Chen et al. reported the degradation of oxygenates to
hydrocarbons under a H2 pressure of 98 bar using a Ru
catalyst.38,39 This reaction involves acid-catalyzed dehydration
of alcohols and subsequent hydrogenation to paraffins.40 The
reaction mixture in the work of Chen et al. was slightly acidic
due to the presence of organic acids.38,39 To exclude the pos-
sibility that the selectivity changes as a function of tempera-
ture are caused by secondary reactions of aldehydes and
alcohols in the present work, we carried out several addi-
tional FT reactions in the presence of decanal. Shorter chain
aldehydes could not be used for this purpose because it was
not possible to analyze them nor their products in our
aqueous-phase FT approach. We verified that the pH of the
solution was neutral before and after the reaction. Table 3
shows that the addition of decanal did not affect either the
FT rate nor the selectivities and chain-growth probabilities.
The main product of decanal conversion is decanol. This
result shows that no preferential cracking of the oxygenated
products occurs during aqueous-phase FT. It also provides a
strong indication that chain-growth reactions do not proceed
via the Pichler–Schulz mechanism. Weststrate et al. have
recently shown that ethanol adsorbed on a Co(0001) single
crystal surface can undergo C–O bond cleavage to produce
3.2.1 Effect of temperature. Fig. 5 shows the effect of
reaction temperature on the activity of aqueous phase FT
synthesis for Ru-1.2 and Ru-5.2. With an increase of the reaction
temperature from 130 to 230 °C, the CO consumption rate
increases from 0.2 to 44.4 and from 0.2 to 12.7 mmol gRu−1 h−1
for Ru-1.2 and Ru-5.2, respectively (Fig. 5a). For these spent
catalysts, TEM images shown in Fig. 6 reveal that the average
size of the particles was nearly similar to that of the initial
samples at 1.4 and 5.4 nm, respectively, implying that no
severe sintering takes place during the FT reaction. The
apparent activation energies (Eaacptp) calculated from these data
(Fig. 5b) are 91 (Ru-1.2) and 75 (Ru-5.2) kJ mol−1. Within the
carbide mechanism the activation energies relate to the
apparent activation energies of the CHx formation reaction
from adsorbed CO.33 The changes in apparent activation
energies (Eaacptp) with particle size are possibly due to the
changes in the selectivity of oxygenates versus hydrocarbons
to be discussed in the next section. These values for E
app
act
are
in the same range as values (67–109 kJ mol−1) reported for
the gas-phase FT reaction using Ru catalysts.34 Fig. 7a shows
the hydrocarbon and oxygenate selectivities for Ru-1.2 and
Ru-5.2. Their general trend as a function of temperature is
Fig. 5 Effect of reaction temperature on activity for FT reaction
catalyzed by Ru-1.2 and Ru-5.2: (a) rate, and (b) Arrhenius plot.
Fig. 7 Effect of reaction temperature on FT reaction catalyzed by
Ru-1.2 and Ru-5.2, (a) hydrocarbons and oxygenates selectivity
Fig. 6 Electron micrographs and particle size distributions for spent
(hydrocarbons:
,
, oxygenates:
,
, Ru-1.2:
,
and Ru-5.2:
,
),
(left) Ru-1.2 and (right) Ru-5.2.
and (b) methane selectivity.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3515

View Article Online
Paper
Catalysis Science & Technology
Table 3 Reactivity data for aqueous-phase Fischer–Tropsch reaction catalyzed by Ru-1.2 in the absence and presence of C10 hydrocarbon
componentsa
Xb (%)
FT rate (mmol gRu h−1
)
SHC (%)
SOXY (%)
αHC
αOXY
−1
Component
—
—
93
77
0.54
0.51
0.58
45
50
50
55
50
50
0.90
0.88
0.87
0.55
Decanalc
1-Decenec
0.57
0.59/0.72d
a
Reaction conditions: Ru nanoparticles (50 μmol), H2O (3 mL), P = 30 bar (H2/CO molar ratio = 2), T = 150 °C, t = 24 h, C10 component
b
c
(1 μmol). Conversion of decanal/1-decene. The contribution of decanal and 1-decene conversion was excluded from the calculation of rate,
selectivity and α. d αOXY is 0.59 from C4 to C14 and changes to 0.72 above C14
.
C2Hx and O surface intermediates.41 This dissociation reac-
tion of adsorbed ethoxy bears resemblance to the proposed
C–O dissociation elementary reaction step after CO insertion
during chain growth in the Pichler–Schulz mechanism. In
any case, the present results show that under our reaction
conditions adsorbed decanal and its hydrogenation product
decanol do not undergo C–O bond cleavage to form an inter-
mediate which can be involved in chain growth.
Another potential reaction pathway for oxygenate forma-
tion involves cationic Ru species, which are present on all of
the Ru nanoparticle catalysts. We discard the possibility that
cationic Ru on the nanoparticles would catalyze oxygenate
formation based on a systematic previous study,27 showing
that the concentration of such cationic Ru species has no
effect on the product selectivity. Related to this, we verified
that oxygenates are not formed via the secondary hydro-
formylation of product olefins. This reaction is typically cata-
lyzed by homogeneous complexes and Ru is a very active
catalyst for this reaction.42 The possibility to carry out hydro-
formylation with metallic nanoparticles has also been
reported recently.43 To further test the possible role of hydro-
formylation for oxygenates formation, we carried out an addi-
tional experiment by adding 1-decene to the reaction
mixture. Table 3 shows that its addition did not significantly
affect the FT rate. 1-Decene was mainly converted to
n-decane. The selectivity of C11-oxygenates was 4%, which is
only marginally higher than the selectivity of 2% for the
experiment without addition of 1-decene. It is interesting to
observe that addition of 1-decene resulted in an increase in
αOXY for oxygenated products longer than C11, whilst the
chain-growth probability for shorter oxygenates was unaf-
fected (Table 3 and Fig. 8). The chain-growth probability for
hydrocarbons was also unaffected. This result clearly shows
that 1-decene is re-adsorbed on the catalytic surface and used
in further chain growth of oxygenates and not of hydro-
carbons. The main conclusion from these results is that
oxygenate formation does not occur via hydroformylation of
intermediate olefins. The finding that 1-decene participates
in further chain growth of oxygenates provides an additional
indication that the surface intermediate for oxygenate chain
growth is different from the one for hydrocarbons chain
growth. This would decouple the mechanisms for formation
of hydrocarbons and oxygenates. We have earlier excluded
the effect of organic capping agent and solvent as the main
factor in the formation of oxygenates.29c
Fig. 7b shows that methane selectivity for both Ru-1.2
and Ru-5.2 are below 12% in the temperature regime 130 to
230 °C. Ru-1.2 shows substantially higher methane selectivity.
It is independent of temperature. For Ru-5.2, the initially
lower methane selectivity increases from 5 to 10% when tem-
perature increases. This difference in methane selectivity
indicates that, for different particle sizes, there is a change in
the competition for CHx to terminate as methane versus
being incorporated into the growing chain. The constant
Fig. 8 Anderson–Schulz–Flory plot for oxygenates produced in FT reaction catalysed by Ru-1.2 (a) in the absence and presence of decanal, and
(b) in the presence of decene.
3516 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
selectivity for methane on the small particle catalyst suggests
a common intermediate with the growing hydrocarbon
chains. Higher methane selectivity is consistent with the acti-
vation energy of CO activation being higher than that of
methane formation.33
The effect of temperature on the chain-growth probability
for Ru-1.2 and Ru-5.2 is shown in Fig. 9. As the short-chain
oxygenates could not been measured in our experiments, the
chain-growth probabilities are calculated based on C4+ prod-
ucts only. It is important to note that similar to our earlier
temperature of 150 °C. The particle size dependence of the
FT activity is shown in Fig. 10. The weight-based rate of the
Ru nanoparticle catalysts shows a maximum around 2.3 nm
for particles between 1.2 and 3.7 nm. With a further increase
of the Ru nanoparticle size the rate increases strongly. When
the surface-atom normalized rates (TOF) are plotted against
particle size (Fig. 10b), it is seen that the activity increases
with particle size with a plateau between 2.3 and 3.7 nm.
These activity trends are in very good qualitative agreement
with literature data for supported Ru nanoparticle cata-
lysts.21,22 Both Kang et al.21 and Carballo et al.22 reported
that the FT activity of Ru nanoparticles decreased below a
critical size of 7–10 nm. Kang et al. investigated Ru nano-
particles sized between 3 and 10 nm supported on carbon
nanotubes.21 Carballo et al. studied Ru/Al2O3 with Ru nano-
particle sizes between 4 and 23 nm.22 Neither literature
report included Ru particles smaller than 2 nm. Similar to
our observations, the surface-atom normalized TOF reported
by Kang et al. shows a region of structure insensitivity in
the FT reaction between 2.5 and 4 nm and an increase in
TOF for particles larger than 4 nm. Consistent with this,
Carballo et al. found that TOF increases for particles larger
than 4 nm.22
Although the origin of the particle size dependence has
not been elucidated yet, our data support the assumption
that a minimum size of the nanoparticles is required to
sustain step-edge sites needed for facile CO dissociation.7,17
In line with the early suggestions of Van Hardeveld and
Montfoort,46 Honkala et al. reported that particles smaller
than 2 nm do not contain step-edge sites.47 On highly regular
cuboctahedron particle highest density of step-edge sites is
predicted for a particle size of about 1.5–2 nm.47 It has been
argued that in the absence of step-edge sites, CO dissociation
may preferentially occur via H-assisted pathways.9,10,48,49
Recent quantum-chemical calculations for Ru surfaces show
that direct CO dissociation on step-edge sites is preferred
over H-assisted CO dissociation.11 In the absence of a clear
method to determine the step-edge site density, we cannot
report the chain growth probabilities for hydrocarbons (αHC
)
and oxygenates (αOXY) are different.27 This is explained in
terms of different reaction centers for formation of these two
product classes on the Ru nanoparticle surface.27 Upon
increasing the reaction temperature from 130 to 230 °C, the
hydrocarbon chain growth probability (αHC) is found to
decrease for both considered nanoparticle catalysts. This
behavior is typical44,45 and consistent with the higher activa-
tion energy for chain-growth termination compared to the
overall rate of chain growth.7,8
The chain-growth probabilities for Ru-1.2 and Ru-5.2 show
anomalous temperature dependence. The initial increase of
αOXY for Ru-5.2 with temperature is comparable to what was
found previously for 2 nm Ru nanoparticles.27 For Ru-1.2,
αOXY goes through a maximum at 210 °C and decreases with
further temperature increase. The increase in αOXY is much
more pronounced for Ru-1.2 compared to Ru-5.2. Trends for
α will be further elaborated in the discussion section using
kinetic considerations.
3.2.2 Effect of particle size. As apparent from several
recent reports,17–22 the FT reaction is structure sensitive for
small metal nanoparticles. We investigated the effect of
particle size in aqueous FT synthesis in more detail at a
Fig. 9 Effect of reaction temperature on the chain growth probability
for FT reaction catalyzed by Ru-1.2 and Ru-5.2 (hydrocarbons:
oxygenates: , Ru-1.2: and Ru-5.2: ).
,
,
Fig. 10 Effect of particle size on the FT reaction: (a) rate and
(b) surface normalized turnover frequency (TOF).
,
,
,
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3517

View Article Online
Paper
Catalysis Science & Technology
firmly conclude on the origin for the activity trends as a func-
tion of particle size. It should be noted that the Honkala data
are based on idealized ‘Wulff’ nanoparticles. Chorkendorff's
group recently employed CO isotope scrambling on differ-
ently sized Ru nanoparticles (3–12 nm) and found that the
scrambling activity increases with particle size.50 This trend
was correlated to the increased density of step-edge sites for
larger particles. Indeed, several further indications from liter-
ature strengthen the supposition that step-edge sites are also
formed on terraces of larger nanoparticles.51–56 For instance,
Nielsen et al. reported that large Ru particles have a different
morphology with rougher surfaces than relatively small parti-
cles.51 In another study, Karim et al. found that supported
Ru with larger particle size form unusual shapes.52 By com-
bining EXAFS structure analysis with a theoretical model,
they concluded that larger particles contain more step sites.
Adsorbates can also cause roughening of smooth single crys-
tal surfaces, giving rise to the formation of defects such as
step-edge sites. Surface roughening of smooth single crystal
surface induced by CO and H2 has been reported for Pt and
Rh53 as well as Co.54–56 Hence, we speculate that the surface
planes on Ru nanoparticles greater than 4 nm are large
enough to accommodate further layers of Ru giving rise to
increased density of step-edge sites. Although the present
data clearly confirm that the performance most strongly
depends on particle size and not on the Ru0 content, it
remains to be determined how the exact synthesis conditions
affects the surface composition of the nanoparticles. We
expect that the surface structure of nanoparticles will criti-
cally depend on the method of preparation and pretreatment.
Fig. 11a shows that the oxygenate selectivity decreases
when the particle size increases from 2.3 to 5.2 nm but
increases when size increases from 1.2 to 2.3 nm. The lower
oxygenate selectivity for the Ru-1.2 nm compared to the
2.3 nm particles is due to the higher selectivity towards meth-
ane as shown in Fig. 12. The higher methane selectivity on
the 1.2 nm is probably due the higher surface density of sites
with higher barrier for CO dissociation, which favors meth-
ane formation.7 We hypothesize that methane formation
occurs on the oxygenate growth sites with high barrier for CO
dissociation. With a higher concentration of sites with high
barrier for CO dissociation, one would expect more methanol
being formed. Oxygenates with less than four carbon atoms
could not be analyzed due to experimental limitations. In
order to understand the effect of higher methane selectivity,
we recalculated the selectivity by taking into account only C4+
products (denoted as selectivity (C4+)). Fig. 11b shows that
the selectivity (C4+) for oxygenates remained constant from
1 to 3 nm and decreased with further increase in particle
size. This indicates that the ratio of oxygenate to hydrocarbon
formation sites remains constant when the particle size increases
Fig. 11 Effect of particle size on the FT reaction: (a) selectivity,
(b) selectivity for C4+ products, (c) hydrocarbons and oxygenates chain
growth probability, and (d) activities for hydrocarbons and oxygenates
formation.
Fig. 12 Effect of particle size on methane selectivity in FT reaction.
residence time for CHx species is independent of Ru particle
size.22 If CHx residence time remains unchanged, we would
expect that the change in the intrinsic rate of oxygenate or
hydrocarbon formation affects α. However, as Fig. 11c shows
that, above 3 nm, both αOXY and αHC are not influenced by
the increase in Ru particle size, we conclude that there is
no change in the intrinsic rate of oxygenate or hydrocarbon
formation. Hence, the decrease in the selectivity (C4+) for oxy-
genates with particle size is primarily due to the decrease in
the relative ratio of oxygenates to hydrocarbons growth sites.
from 1 to 3 nm. The decrease in oxygenate selectivity (C4+
)
above 3 nm could either imply that (i) the ratio of oxygenates
to hydrocarbons growth sites decreases, or (ii) the intrinsic rate
of oxygenate formation decreases with respect to hydrocarbon
formation. Recent SSITKA experiments have shown that the
3518 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Fig. 11c shows that αHC is independent of particle size,
while αOXY increases from 1 to 3 nm but remains constant
above 3 nm. Hydrocarbons are shown to exhibit higher
chain-growth probability compared to oxygenates. This sup-
ports our earlier proposal that oxygenates and hydrocarbons
are formed on two different catalytic centers.27 If oxygenates
and hydrocarbons would be formed on similar catalytic site
with only a different termination step, one would expect
trends in α with Ru particle size to be the same for oxygen-
ates and hydrocarbons. In addition, the lower αOXY compared
to αHC is consistent with CO dissociation at oxygenates
growth sites having a higher activation barrier, and the termi-
nation step to oxygenates having a lower activation barrier
compared hydrocarbons as follows from eqn (8) in the dis-
cussion section.
parallel with an increasing αOXY. This means that more and
longer oxygenate chains are formed. This trend cannot be
explained by an increase in the number of oxygenates growth
sites only. If one also considers the trend for methane selec-
tivity (Fig. 12), the increase in αOXY (Fig. 11c) is complemented
by a decrease in the methane selectivity when the Ru particles
size increases from 1 to 3 nm. We have earlier proposed that
the oxygenate formation sites are also favorable for methane
formation due to the higher barrier for CO dissociation. The
decrease in methane selectivity will lead to higher CHx
coverage near the oxygenates growth site and, accordingly,
more CHx species are available for oxygenate chain growth.
Consequently, this leads to higher chain growth rate at the
oxygenates growth sites and αOXY increases.
The surface-atom normalized rates for hydrocarbon 4. Discussion
(TOFHC) and oxygenate (TOFOXY) formation as a function of
4.1 Effect of temperature on FT reaction
the particle size are given in Fig. 11d. It can be seen that both
rates increase with the particle size. The increase of TOFHC
with particle size is consistent with the proposed increasing
density of step-edge sites. Fig. 11c shows that, despite the
increasing density of these sites for facile CO dissociation,
the chain-growth probability for hydrocarbons remains con-
stant. A rationalization for this behavior is that each site for
CO dissociation is kinetically coupled to a chain growth site.
Such a dual reaction center model has been first proposed by
Shetty et al.57 and its kinetic implications have been recently
developed.6 This insensitivity of reactivity of the reaction cen-
ters to particle size is in line with SSITKA experiments in
which the residence time of CHx species on the surface of a
Ru catalyst was found to be independent of particle size.22
Note that these latter measurements were carried out at tem-
peratures where hydrocarbon formation is dominant. Since
The anomalous trend of increasing α for oxygenate formation
with temperature (Fig. 9) is a consequence of the decreasing
apparent activation barrier for CO dissociation.7,8 It results in
an increase of the apparent activation energy of chain growth
versus the rate of chain termination. This can be deduced in
the following way. The chain-growth probability can be
expressed as
r
C
prop  i 
 r C
 C
 i 

; i 1
(5)
r
C
prop  i  term  i 
where rprop(Ci) is the rate of propagation of a carbon chain
Ci with length i to a carbon chain Ci+1 with length i + 1
and rterm(Ci) the rate of termination of a carbon chain with
length i. An expression for the steady-state coverage of the
building blocks, which are CHx surface species in the carbide
chain growth is predicted to be structure insensitive,58
a
mechanism, θCH , is
X
constant residence time for CHx intermediates will lead to
similar chain length and, hence constant, αHC. Previous
studies have also shown that αHC is independent of particle
size for supported Ru and Co catalysts.44,45

dCH
x
(6)
 rCO diss  rmethane

rprop (Ci )

dt
i1
The kinetic observations for oxygenate formation are quite
different from those for hydrocarbon formation. Two regions
can be discerned. One with Ru nanoparticles in the range
from 1 to 3 nm, and a second with Ru particles ranging from
3 to 5 nm. When the particle size is increased from 3 to
5 nm, TOFOXY increases and αOXY remains constant. This
implies that more dissociated CO is converted into oxygen-
ates, while the chain length remains constant. This can be
explained by an increase in the number of oxygenate growth
sites. Similar to the hydrocarbon growth sites, each site for
CO dissociation is coupled to an oxygenate growth sites for
Ru particles larger than 3 nm. Since we have hypothesized
that oxygenates are formed on planar sites with higher bar-
rier for CO dissociation,27 the increase in TOFOXY for Ru
larger than 3 nm can be explained by the Ru particles having
larger planar facets which can accommodate more oxygenates
growth sites. In contrast, increasing the Ru particle size from
1 to 3 nm (the first range) results in an increasing TOFOXY in
in which the terms on the right hand side reflect the rates
of CO dissociation, methanation (CHx hydrogenation) and
propagation by C–C coupling of CHx to all growing hydro-
carbon chains Ci forming Ci+1, respectively. One deduces
from eqn (6) that in order to generate a high surface coverage
of CHx, the rate of CO dissociation has to be high and the
rate of methane formation low. Assuming that (i) the rate of
water formation is fast, and (ii) the rate of propagation is
much faster than CHx formation and the rate of termination,
the following expression for α can be derived.28


1 1
3 

2


kterm
kprop  kCO diss CO  1
(7)
  1







CO 


 

According to the carbide mechanism, the formation of
oxygenates occurs via CO insertion into the growing chain as
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3519

View Article Online
Paper
Catalysis Science & Technology
ads
ads
a terminating step. We modify eqn (7) to describe the chain-
growth probability for oxygenates by replacing kterm with
kterm oxy·θCO. Then, the following equation can be derived:
significantly affect αOXY. An increase in E
and/or E
CO diss prop
ads
would decrease αOXY, while an increase in E
would
term
ads
CO diss
ads
increase αOXY. Fig. 13a shows that when E
< E
<
prop
ads
term
E
, the rate of CO dissociation is very fast and termination


1 1
3 

occurs very slowly. Hence, αOXY approaches 1. Furthermore,
when the rate of CO dissociation is high, the surface coverage
of CO will be low and the probability to terminate by CO
kterm oxy2 CO


(8)
OXY  1






kprop  kCO diss  1

CO 


 
ads
prop
ads
term
ads

insertion is low. In contrast, for E
> E
≫ E
,
CO diss
there is insufficient CHx for chain growth so that termination
The change in αOXY with temperature was then modeled
using eqn (8) and the values used in the Arrhenius equation
are given in Table 4. The surface CO coverage was estimated
is slow and, henceforth, αOXY approaches 0. This implies that
only methane and methanol will be formed. Based on this
ads
CO diss
analysis, for the formation of long chain oxygenates E
by a simple method using the Van't Hoff equation and
should not be too low to maintain sufficient CO on the sur-
face near the oxygenate growth site for termination by CO
ads
CO
assuming θCO to be 0.8 at 130 °C with E
of 130 kJ mol−1
.
ads
ads
Fig. 13 shows that αOXY increases with temperature under
insertion. High E
has to be accompanied by low E
prop
CO diss
ads
our specified conditions, which supports the experimental
and high E
to allow chain growth.
term
ads
CO
ads
prop
ads
findings in Fig. 9. E
diss, E
and E
are shown to
term
The change in αHC is modeled using eqn (7). For easier
comparison, the values used to model αOXY in Table 4 will be
ads
CO diss
used, except that a value of E
of 50 kJ mol−1 was used
Table 4 Kinetic parameters used in simplified model to predict chain
growth probabilities for oxygenates (OXY) and hydrocarbons (HC)
for the base case, the low value being consistent with the pro-
posal that hydrocarbons are formed on sites with relatively
low barrier for CO dissociation. From Fig. 14, the conven-
tional temperature dependence of decreasing α with tempera-
ture is always observed. This is in agreement with our results
Rate constant
Ao (s−1
)
Eact (kJ mol−1
)
kCO diss OXY
kprop OXY
kterm OXY
kCO diss HC
kprop HC
1013
1013
1017
1013
1013
1017
110
60
80
50
60
80
for αHC in Fig. 9 and also those reported in literature.44,45
ads
Similar to the findings for αOXY, an increase in E
and/or
CO diss
ads
kterm HC
E
E
would lead to a decrease in αHC, and an increase in
would increase αHC. Hence, to maintain high chain
prop
act
term
ads
ads
growth for hydrocarbon, E
as possible and E
and E
should be as low
prop
CO diss
ads
term
should be high.
ads
CO
Fig. 13d and 14d show that the change in E affects αOXY
ads
and αHC very differently. Higher E leads to higher αOXY but
CO
lower αHC. However, consistent to both αOXY and αHC, higher
ads
CO
E
leads to more rapid changes in α. This relates to the
dependence of the overall rate of CO dissociation on the
desorption barrier of CO as observed from eqn (9).
act
act
CO diss
ads
CO
E
= E
− (1 − 2θCO)·E
(9)
app,CO diss
On the oxygenate formation sites, the activation barrier
for CO dissociation is high and θCO is also high. This means
that in order for CO dissociation to occur a vacancy needs to
be available requiring CO desorption. This will lead to
act
act
E
> E
when θCO approaches 1. Hence, the
CO diss
app,CO diss
act
higher CO adsorption energy will lead to higher E
,
app,CO diss
and the rate of CO dissociation will increase more rapidly
with temperature resulting in higher αOXY. In contrast, on the
hydrocarbon growth site where barrier for CO dissociation is
act
low and θCO is low, the adsorption of CO will lower E
.
app,CO diss
act
Accordingly, Eaapctp,CO diss > E
when θCO approaches 0. Also,
CO diss
act
higher CO adsorption energy will lead to lower E
, and
app,CO diss
the rate for CO dissociation will increase much slower
compared the rate of termination with temperature, resulting
in lower αHC
.
Fig. 13 Simplified kinetics model for oxygenates chain growth
probability with varying activation energy for (a) CO dissociation,
(b) propagation, (c) termination, and (d) CO adsorption energy.
The anomalous temperature dependence of αOXY is incon-
sistent with the alternative Pichler–Schulz CO insertion
3520 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
relative rate of methanation remains unchanged. When more
CHx species are being formed, they are predominately being
utilized for chain growth. Consequently, this leads to a more
pronounced increase in αOXY for Ru-1.2 compared to Ru-5.2.
These results also imply that methane formation for Ru-1.2 is
due the low rate of CO dissociation, but for Ru-5.2 methane
formation is controlled by the activation energy for methana-
tion. Another contributing factor to the faster increase in
αOXY could be the strength of CO adsorption. As observed
from CO ATR-IR studies and the dependence of the surface
coordination number on Ru nanoparticle size as determined
from EXAFS, smaller particles will bind CO more strongly
than the larger particles. As discussed earlier and also shown
ads
CO
Fig. 13d, higher E would increase αOXY more rapidly.
5. Conclusions
We have shown that in aqueous-phase FT the chain growth
probabilities of oxygenates and hydrocarbons exhibit differ-
ent trends with respect to temperature and particle size. This
indicates that two different sites are responsible for their for-
mation. With increasing temperature, αHC decreases as usu-
ally observed in conventional FT synthesis. Conversely,
increasing temperature leads to an increase in αOXY. We pro-
pose that the increase in αOXY is caused by the high apparent
activation energy of CO dissociation, which is due to a high
CO coverage. This essentially hinders CO dissociation, since
the rate of chain termination by CO insertion is relatively
fast. Results were analyzed with a kinetic model and are con-
sistent with the carbide mechanism. The increasing αOXY and
decreasing oxygenate selectivity with temperature cannot be
explained using the Pichler–Schulz mechanism. An increase
in Ru particle size leads to a decrease in the oxygenate selec-
tivity, indicating that the relative ratio of oxygenates to hydro-
carbons growth sites decreases. The kinetic parameters point
to an increasing density of active sites for hydrocarbons and
oxygenates formation with increasing particle size. The rela-
tive ratio of oxygenates to hydrocarbons growth sites becomes
smaller for larger particles.
Fig. 14 Simplified kinetics model for hydrocarbons chain growth
probability with varying activation energy for (a) CO dissociation,
(b) propagation, (c) termination, and (d) CO adsorption energy.
mechanism for chain growth. The expression for α according
to the Pichler–Schulz mechanism is
.
kprop CO
PS

(10)
kpropCO  kterm
In contrast to expression (7), kterm will always have a higher
activation energy for high α.
From the methane selectivity (Fig. 7b) and αOXY (Fig. 9),
Ru-1.2 and Ru-5.2 are shown to exhibit very different trends.
This indicates that the chemical nature of the reactive sites
are different for these two catalysts. From the trends in
Fig. 7b and 9, two important observations are made, namely Acknowledgements
(i) constant methane selectivity for Ru-1.2 is accompanied
The authors thank the National Research School Combina-
by a steep increase in αOXY, and (ii) increasing methane selec-
tivity for Ru-5.2 is accompanied by gradual increase in αOXY
tion on Catalysis for financial support, NWO-Dubble for
access to X-ray absorption spectroscopy facilities at ESRF,
ESRF staff for their support and the Soft Matter Cryo-TEM
Research Unit for access to the TEM facility. We thank
Adelheid Elemans for the elemental analysis, Dr. Peter Thüne
and Ajin Cheruvathur for providing assistance with ATR-IR
measurements.
.
These trends can be explained based on eqn (6) to (8).
Increasing methane selectivity with temperature for Ru-5.2
indicates methane formation has a higher barrier than for
Ru-1.2, which becomes more important at high temperature
for Ru-5.2. Based on eqn (6), the consumption of CHx for
methanation would result in less CHx available for the C–C
coupling chain growth step. Hence, when more CO dis-
sociates for Ru-5.2 with increasing temperature, there would
be an increase in competition for CHx species between
methanation and oxygenates chain growth. In comparison,
for Ru-1.2, constant methane selectivity indicates that the
References
1 J. J. Spivey and A. Egbebi, Chem. Soc. Rev., 2007, 36,
1514–1528.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3521

View Article Online
Paper
Catalysis Science & Technology
2 E. van Steen and M. Claeys, Chem. Eng. Technol., 2008, 31,
655–666.
3 A. Y. Khodakov, W. Chu and P. Fongarland, Chem. Rev.,
2007, 107, 1692–1744.
4 Q. Zhang, J. Kang and Y. Wang, ChemCatChem, 2010, 2, 1030.
5 B. H. Davis, Ind. Eng. Chem. Res., 2007, 46, 8938–8945.
6 A. J. Markvoort, R. A. van Santen, P. A. J. Hilbers and
E. J. M. Hensen, Angew. Chem., Int. Ed., 2012, 51, 9015–9019.
7 R. A. van Santen, I. M. Ciobica, E. van Steen and
M. M. Ghouri, Adv. Catal., 2011, 54, 127–187.
C. Wang, J. Yang, C. Xiao, H. Wang, D. Ma and
Y. Kou, Catal. Today, 2012, 183, 136; (c) H. Wang,
W. Zhou, J.-X. Liu, R. Si, G. Sun, M.-Q. Zhong, H.-Y. Su,
H.-B. Zhao, J. A. Rodriguez, S. J. Pennycook, J.-C. Idrobo,
W.-X. Li, Y. Kou and D. Ma, J. Am. Chem. Soc., 2013,
135, 4149.
27 X. Y. Quek, Y. Guan, R. A. van Santen and E. J. M. Hensen,
ChemCatChem, 2011, 3, 1735–1738.
28 C. Wang, H. Zhao, H. Wang, L. Liu, C. Xiao and D. Ma,
Catal. Today, 2012, 183, 143–153.
8 R. A. van Santen, M. M. Ghouri, S. Shetty and
E. J. M. Hensen, Catal. Sci. Technol., 2011, 1, 891–911;
R. A. van Santen, M. Ghouri and E. J. M. Hensen, Phys.
Chem. Chem. Phys., 2014, 16, 10041.
9 I. M. Ciobica and R. A. van Santen, J. Phys. Chem. B,
2003, 107, 3808–3812.
29 (a) X.-Y. Quek, I. A. W. Filot, R. Pestman, R. A. van Santen,
V. Petkov and E. J. M. Hensen, Chem. Commun., 2014, 50,
6005–6008; (b) X.-Y. Quek, Y. Guan and E. J. M. Hensen,
Catal. Today, 2012, 183, 72–78; (c) X.-Y. Quek, R. Pestman,
R. A. van Santen and E. J. M. Hensen, ChemCatChem.,
2013, 5, 3148–3155.
10 M. Ojeda, R. Nabar, A. U. Nilekar, A. Ishikawa,
M. Mavrikakis and E. Iglesia, J. Catal., 2010, 272, 287–297.
11 S. Shetty, A. P. J. Jansen and R. A. van Santen, J. Am. Chem.
Soc., 2009, 131, 12874–12875.
30 J. J. F. Scholten, A. P. Pijpers and A. M. L. Hustings, Catal.
Rev.: Sci. Eng., 1985, 27, 151–206.
31 D. A. J. M. Ligthart, R. A. van Santen and E. J. M. Hensen,
Angew. Chem., Int. Ed., 2011, 50, 5306–5310.
12 R. C. Brady and R. Pettit, J. Am. Chem. Soc., 1981, 103,
1287–1289.
32 S. Y. Chin, C. T. Williams and M. D. Amiridis, J. Phys. Chem.
B, 2006, 110, 871–882.
13 J. Gaube and H. F. Klein, J. Mol. Catal. A: Chem., 2008, 283, 60–68.
14 H. Pichler and H. Schulz, Chem. Ing. Tech., 1970, 42,
1162–1174.
15 B. H. Davis, Catal. Today, 2009, 141, 25–33.
16 E. Iglesia, S. L. Soled and R. A. Fiato, J. Catal., 1992, 137,
212–224.
33 R. A. van Santen and A. Markvoort, ChemCatChem, 2013, 5,
3384–3397.
34 Y.-W. Chen and J. Goodwin Jr., React. Kinet. Catal. Lett.,
1984, 26, 453–459.
35 E. Iglesia, S. C. Reyes and R. J. Madon, J. Catal., 1991, 129,
238–256.
17 G. L. Bezemer, J. H. Bitter, H. Kuipers, H. Oosterbeek,
J. E. Holewijn, X. D. Xu, F. Kapteijn, A. J. van Dillen and
K. P. de Jong, J. Am. Chem. Soc., 2006, 128, 3956–3964.
18 O. Borg, P. D. C. Dietzel, A. I. Spjelkavik, E. Z. Tveten,
J. C. Walmsley, S. Diplas, S. Eri, A. Holmen and E. Ryttera,
J. Catal., 2008, 259, 161–164.
19 G. Prieto, A. Martínez, P. Concepción and R. Moreno-Tost,
J. Catal., 2009, 266, 129–144.
20 J. P. den Breejen, P. B. Radstake, G. L. Bezemer, J. H. Bitter,
V. Froseth, A. Holmen and K. P. de Jong, J. Am. Chem. Soc.,
2009, 131, 7197–7203.
21 J. Kang, S. Zhang, Q. Zhang and Y. Wang, Angew. Chem., Int.
Ed., 2009, 48, 2565–2568.
22 J. M. G. Carballo, J. Yang, A. Holmen, S. Garcia-Rodriguez,
S. Rojas, M. Ojeda and J. L. G. Fierro, J. Catal., 2011, 284,
102–108.
36 R. J. Madon, S. C. Reyes and E. Iglesia, J. Phys. Chem.,
1991, 95, 7795–7804.
37 J. G. Ekerdt and A. T. Bell, J. Catal., 1980, 62, 19–25.
38 L. Chen, Y. Zhu, H. Zheng, C. Zhang and Y. Li, J. Chem.
Technol. Biotechnol., 2012, 87, 1089–1097.
39 L. Chen, Y. Zhu, H. Zheng, C. Zhang, B. Zhang and Y. Li,
J. Chem. Technol. Biotechnol., 2012, 87, 112–122.
40 X. Y. Quek, Y. Guan, R. van Santen and E. J. M. Hensen,
ChemSusChem, 2010, 3, 1264–1267.
41 C. J. Weststrate, H. J. Gericke, M. W. G. M. Verhoeven,
I. M. Ciobîcă, A. M. Saib and J. W. Niemantsverdriet, J. Phys.
Chem. Lett., 2010, 1, 1767–1770.
42 R. Franke, D. Selent and A. Börner, Chem. Rev., 2012, 112,
5675–5732.
43 A. J. Bruss, M. A. Gelesky, G. Machado and J. Dupont, J. Mol.
Catal. A: Chem., 2006, 252, 212–218.
23 A. Tuxen, S. Carenco, M. Chintapalli, C.-H. Chuang,
C. Escudero, E. Pach, P. Jiang, F. Borondics, B. J. Beberwyck,
A. P. Alivisatos, G. Thornton, W.-F. Pong, J. Guo, R. Perez,
F. Besenbacher and M. Salmeron, J. Am. Chem. Soc.,
2013, 135, 2273–2278.
44 G. P. Van der Laan and A. Beenackers, Catal. Rev.: Sci. Eng.,
1999, 41, 255–318.
45 R. A. Dictor and A. T. Bell, J. Catal., 1986, 97, 121–136.
46 R. van Hardeveld and A. van Montfoort, Surf. Sci., 1966, 4,
396–430.
24 D. I. Enache, B. Rebours, M. Roy-Auberger and R. Revel,
J. Catal., 2002, 205, 346–353.
25 H. L. Karaca, O. V. Safonova, S. P. Chambrey, P. Fongarland,
P. Roussel, A. Griboval-Constant, M. Lacroix and
A. Y. Khodakov, J. Catal., 2011, 277, 14–26.
47 K. Honkala, A. Hellman, I. N. Remediakis, A. Logadottir,
A. Carlsson, S. Dahl, C. H. Christensen and J. K. Norskov,
Science, 2005, 307, 555–558.
48 M. P. Andersson, E. Abild-Pedersen, I. N. Remediakis,
T. Bligaard, G. Jones, J. Engbwk, O. Lytken, S. Horch,
J. H. Nielsen, J. Sehested, J. R. Rostrup-Nielsen, J. K. Norskov
and I. Chorkendorff, J. Catal., 2008, 255, 6–19.
26 (a) C. X. Xiao, Z. P. Cai, T. Wang, Y. Kou and N. Yan,
Angew. Chem., Int. Ed., 2008, 47, 746–749; (b) L. Liu, G. Sun,
3522 | Catal. Sci. Technol., 2014, 4, 3510–3523
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
49 G. A. Morgan, D. C. Sorescu, T. Zubkov and J. T. Yates,
J. Phys. Chem. B, 2004, 108, 3614–3624.
53 G. A. Somorjai, K. S. Hwang and J. S. Parker, Top. Catal.,
2003, 26, 87–99.
50 C. Strebel, S. Murphy, R. M. Nielsen, J. H. Nielsen and
I. Chorkendorff, Phys. Chem. Chem. Phys., 2012, 14,
8005–8012.
51 R. M. Nielsen, S. Murphy, C. Strebel, M. Johansson,
I. Chorkendorff and J. H. Nielsen, J. Nanopart. Res., 2010, 12,
1249–1262.
52 A. M. Karim, V. Prasad, G. Mpourmpakis, W. W. Lonergan,
A. I. Frenkel, J. G. Chen and D. G. Vlachos, J. Am. Chem.
Soc., 2009, 131, 12230–12239.
54 J. J. C. Geerlings, J. H. Wilson, G. J. Kramer, H. Kuipers,
A. Hoek and H. M. Huisman, Appl. Catal., A, 1999, 186, 27–40.
55 J. Wilson and C. Degroot, J. Phys. Chem., 1995, 99,
7860–7866.
56 G. A. Beitel, C. P. M. de Groot, H. Oosterbeek and
J. H. Wilson, J. Phys. Chem. B, 1997, 101, 4035–4043.
57 S. G. Shetty, I. M. Ciobica, E. J. M. Hensen and R. A. van Santen,
Chem. Commun., 2011, 47, 9822–9824.
58 R. A. van Santen, Acc. Chem. Res., 2009, 42, 57–66.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 3510–3523 | 3523