Catalysis Today 261 (2016) 17–27

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

Hydrogen spillover in the Fischer–Tropsch synthesis: An analysis of

platinum as a promoter for cobalt–alumina catalysts
Doreen Nabaho a , J.W. (Hans) Niemantsverdriet b , Michael Claeys a , Eric van Steen a,∗
a
DST-NRF Centre of Excellence in Catalysis (c*change), Department of Chemical Engineering, University of Cape Town, Private Bag X3, Rondebosch 7701,
South Africa
b
Syngaschem BV and Laboratory for Physical Chemistry of Surfaces, Eindhoven University of Technology, P.O. Box 513, Eindhoven 5600 MB, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Hydrogen spillover has been invoked to explain the functioning of platinum as a promoter in cobalt-
Received 18 May 2015 based Fischer–Tropsch catalysts. In this study, the operation of Pt was investigated using a model hybrid
Received in revised form 14 August 2015 catalyst (i.e. mixture of Pt/Al2 O3 + Co/Al2 O3 ) which allowed for decoupling of hydrogen spillover effects
Accepted 24 August 2015
from those requiring direct Pt–Co contact. Pt improved the reducibility of the hybrid catalyst despite
Available online 9 October 2015
physical separation of Pt from Co.
TPR, TGA, in situ XRD and quasi in situ XPS confirmed that supported Co3 O4 reduced via formation of
Keywords:
CoO as a stable intermediate. Pt had only a slight effect on Co3 O4 → CoO reduction, but greatly improved
Hydrogen spillover
Hybrid catalysts
the CoO → Co0 reduction which was severely hindered by interaction with the alumina support. The
Noble metal promoter catalysing effect of Pt on the reduction was attributed to H2 dissociation and its subsequent spillover
TPR occurring more readily than direct H2 activation by the cobalt oxides. During the Fischer–Tropsch reac-
TGA tion, the role of spillover hydrogen was again invoked to explain the higher TOF and enhanced selectivity
In situ XRD towards CH4 and paraffins. Spillover from Pt was proposed to induce a hydrogen-rich microenviron-
In situ XPS ment that resulted in a ‘cleansing’ effect on the catalyst surface and an increase in the selectivity of
hydrogenated products.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
Cobalt-based catalysts for the commercial low-temperature
Fischer–Tropsch reaction require activation in a reducing atmo-
sphere to form the metallic active phase. A high extent of reduction
of the cobalt oxide to the metal is desired to maximise the cata-
lyst activity but is seldom achieved especially when using highly
interacting supports like Al2 O3 . Some interaction between the
support and the cobalt crystallites is crucial because it provides
the anchorage required to prevent crystallites from migrating and
sintering, and therefore helps to minimise the associated loss in
catalytic activity. Unfortunately, strong metal support interactions
(SMSIs) favour cobalt dispersion at the expense of reducibility,
which results in the need for reduction promoters such as Pt in
the commercial Al2 O3 -supported catalyst [1–14].
Platinum’s ability to catalyse the reduction of cobalt oxides has
been attributed to a high affinity for H2 activation (i.e. adsorp-
tion and dissociation) although electronic/ligand effects arising
∗ Corresponding author.
E-mail address: eric.vansteen@uct.ac.za (E. van Steen).
from Pt–Co coordination may also occur. The proposed promotion
mechanism(s) is therefore inadvertently linked to the promoter’s
location relative to the cobalt crystallites: Pt must coordinate
directly with Co to induce an electronic effect, while it must be
exposed to the gas phase to activate H2 molecules that may sub-
sequently spillover to the Co for reduction. Studies have shown
that Pt–Co coordination occurs readily [5,9,10,12,15] and this could
have non-trivial effects on the activity, deactivation, regeneration
and the reduction of the oxide phase. On the other hand, reduction
of cobalt oxides via a hydrogen spillover mechanism has been sug-
gested for noble metal promoters [16–18]. The effects of hydrogen
spillover from promoters are not limited to reduction, but have also
been proposed to influence the CH4 selectivity and cobalt activity
during the Fischer–Tropsch reaction [13,19–21].
Scheme 1 illustrates the intrinsic steps that may occur during
the reduction of a metal oxide in the presence of a noble metal
promoter according to a hydrogen spillover mechanism. The pro-
cess is initiated by dissociative adsorption of H2 on the promoter
surface (Step 1). This is followed by the spillover step during
which hydrogen atoms cross the promoter–oxide interface (Step
2). Reduction may then occur when the spillover hydrogen atoms
react with oxygen in the structure of the metal oxide to form water,

http://dx.doi.org/10.1016/j.cattod.2015.08.050
0920-5861/© 2015 Elsevier B.V. All rights reserved.
18 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27

Scheme 1. Proposed pathway of spillover hydrogen during the promoter-aided reduction of a metal oxide.
Adapted from Conner and Falconer [30] and Luo and Epling [29].

leaving the metallic phase behind (Step 3). Alternatively, spillover

can occur from an isolated promoter crystallite to the surface of
the support (Step II). During this process, a strong hydrogen–metal
bond is broken in favour of a bond/interaction with the receiving
support surface. Surface migration of hydrogen atoms may then
occur via the surface hydroxyl (OH) groups of the oxidic support
(Step III). Finally, spillover from the support to the metal oxide
may occur (Step IV), followed by reduction (Step 3). Reduction
via spillover competes with direct H2 adsorption at active sites of
the cobalt oxides which are depicted as two adjacent metal atoms
with an O vacancy (i.e. a defect) between them (Step A).
Migration distances of a few nanometres to several centimetres
have been reported for spillover hydrogen species [17,18,22–27].
On oxidic surfaces such as Al2 O3 , surface diffusion of hydrogen
species is reported to involve interaction with surface oxygen
atoms or hydroxyl groups [17,25,28]. It has been suggested that
hydrogen migrates by continuously forming and breaking adjacent
OH bonds, but the energy barrier for this process is quite high [29].
A viable alternative explanation is the ‘bucket-brigade’ model in
which surface OH groups do not migrate, but rather act as vehicles
that pass along spillover hydrogen species. The interactions with
Scheme 2. Illustration of the surface migration of hydrogen according to the
hydrogen are proposed to be weak H· · ·OH bonds and enable a pro- bucket-brigade model/Grotthuss-type mechanism. , covalent bonds and other
cess that has a low energy barrier and fast rates of surface diffusion. interactions; - - - , hydrogen bonds.
The bucket-brigade model has been directly referred to or inferred Adapted from Carley et al. [31].
in a number of studies [24,29–31] and is illustrated in Scheme 2.
The movement of hydrogen is analogous to the Grotthuss mecha-
nism that governs the unusually high proton mobility in water via
a network of hydrogen bonds [32]. Spillover hydrogen (H␣ ) from
the dissociating metal (MeI ) migrates to a secondary surface (MeII ) a solvated proton when a co-catalyst with a high proton affin-
for reaction via a hydrogen-bonded network of OH groups on the ity such as water abstracted an adsorbed hydrogen atom from
surface of an oxidic support. H␣ transfers onto an O atom in its vicin- the dissociating metal surface to form a proton. The proton could
ity from where it is passed and received in an on-going exchange then transfer easily to the site where it was required through a
of covalent and hydrogen bonds between neighbouring OH groups molecularly adsorbed co-catalyst monolayer. Formation of the pro-
[30]. When H␣ crosses the metal–support interface to the closest ton was determined to be exothermic and compensated for the
surface OH group (see red arrow in Scheme 2), the effect is that energy required to break the hydrogen–metal bond. Alternative
H at the extreme end of the chain is ‘pushed’ out of the network pathways that have been suggested to govern interfacial spillover
towards MeII . Consequently, the spilt over H␣ does not physically include: transfer of only weakly chemisorbed hydrogen atoms [34],
migrate but rather triggers a knock-on effect that results in a net ‘jumpover’ of activated hydrogen atoms via the gas phase [35] and
displacement of hydrogen. Mobility in this manner is high because ‘bridges’ such as carbon and specific interfacial OH groups at the
it is the charge of the H␣ that is transferred, rather than its mass. metal–support interface [36–39].
The ‘bucket-brigade’ model accounts for the migration of hydro- The objective of this study was to explicate the effect of the
gen over oxidic surfaces such as catalyst supports. However, before Pt promoter on the alumina supported Co-based catalyst during
this transportation can occur, a strong hydrogen–metal (H–M) sur- reduction and at commercially relevant Fischer–Tropsch condi-
face bond must be broken to allow hydrogen species to spillover tions using a model ‘hybrid’ catalyst. The hybrid catalyst was a
from the dissociating metal onto the support. This interfacial trans- mechanical mixture of Pt/Al2 O3 + Co/Al2 O3 catalysts which pro-
fer is energetically hindered because of the high binding energy vided a means to isolate hydrogen spillover as a remote-control
between hydrogen and metal surfaces [24,33]. Levy and Boudart effect, from the ligand/electronic promoter effects that require
[33] proposed that interfacial spillover occurred via formation of physical contact between Pt and Co.
 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
2. Experimental
2.1. Catalyst preparation
Two sets of catalysts were prepared using a ␥-Al2 O3 support
(Puralox SCCa 5-150). The first set was prepared using the unmodi-
fied ␥-Al2 O3 support (d90,Al2 O3 = 128 ␮m) while the second set was
prepared using a batch of the support that had been milled to a fine
powder in a bench top McCrone Micronising Mill® to obtain a sam-
ple with d90,Al2 O3 = 15.5 ␮m (i.e. 90% of the milled particles were
smaller than 15.5 ␮m).
For each catalyst set, a 20% Co/Al2 O3 base sample was prepared
using a slurry impregnation method with Co(NO3 )2 ·6H2 O as a pre-
cursor. The support was characterised using BET prior to catalyst
preparation (SBET = 162 m2 /g, Vpore = 0.5 cm3 /g, and dpore = 10.1 nm).
Two sequential impregnations using aqueous solutions of 1.27 and
1.58 M were required to attain the desired cobalt metal loading
of 20 g per 100 g of catalyst. Each solution was prepared using
a volume of deionised water equal to three times the pore vol-
ume (Vpore ), to which the support powder was added to obtain
the slurry. After each impregnation, the catalyst was transferred
to a BUCHI® rotary evaporator to remove the liquid portion of the
slurry (85 ◦ C, 50–100 mbar, ca. 40 min). Thereafter, the moist sam-
ple was recovered, dried in an oven for 12 h at 120 ◦ C, and then
calcined at 350 ◦ C in a fluidised bed reactor. Calcination was carried
out in a continuous air flow of 50 ml min−1 gAl2 O3 −1 for 6 h. The
Pt–Co/Al2 O3 sample was subsequently prepared by slurry impreg-
nation of a portion of calcined Co/Al2 O3 with an aqueous solution
of 0.0171 M Pt[(NH3 )4 ]Cl2 ·H2 O. The same procedure was followed
to prepare the monometallic Pt/Al2 O3 sample at the same platinum
salt concentration but using a slurry of the ␥-Al2 O3 support instead.
The sequence of drying and calcination steps carried out on the
Co/Al2 O3 sample was then followed to obtain calcined 0.5% Pt–20%
Co/Al2 O3 and 0.5% Pt/Al2 O3 . Finally, the model hybrid Pt–Co sample
was prepared by physically mixing Pt/Al2 O3 and Co/Al2 O3 in a 1:1
mass ratio to maintain the same Co/Pt ratio of 40 in the Pt–Co/Al2 O3
sample. No further treatment was carried out on the hybrid catalyst.
The elemental composition of all calcined catalyst samples
was confirmed by ICP-OES using a Varian Vista-Pro analyser, and
the crystallites size distributions in the reduced catalysts were
determined using a Zeiss LEO 912-Omega Transmission Electron
Microscope (TEM). The particle sizes in the TEM micrographs were
measured using ImageJ® software, and the cumulative lognormal
distribution of a sample size of ca. 100 particles was plotted to
determine the particle size distributions in the range of 4–17 nm
for Co and 0.5–2 nm for Pt.
2.2. Temperature-programmed reduction (TPR)
An AutoChem 2910 Chemisorption Analyzer was used to study
the temperature-programmed reduction of the calcined catalysts.
The analyses were carried out at atmospheric pressure using 5%
H2 –Ar (GHSV = 2500 ml min−1 gCo−1 ) and the sample temperature
was ramped at 10 ◦ C/min to 950 ◦ C. The hydrogen concentration
was continuously monitored using a thermal conductivity detector
(TCD) and a bleed from the exhaust line of the analyzer was fed
to a Pfeiffer Omnistar GSD 300 Mass Spectrometer for qualitative
analyses.
2.3. Isothermal thermogravimetric analysis (TGA)
A Mettler-Toledo TGA/SDTA 851e was used for the thermogravi-
metric analyses during reduction. Each analysis commenced with a
drying step in N2 , which involved a 10 ◦ C/min temperature ramp to
350 ◦ C. Thereafter, the catalyst was reduced at the same activation
conditions used prior to the Fischer–Tropsch reaction, i.e. 350 ◦ C
19
Scheme 3. Reduction sequence of a single ideal Co3 O4 crystallite showing the for-
mation of CoO as a stable intermediate. Ototal refers to the total mass of O (oxygen)
in the structure of Co3 O4 .
in H2 (1000 ml min−1 gCo−1 ), 1 bar, for 12 h. The reduction was fol-
lowed by an hour-long re-oxidation step in synthetic air after which
the analysis was terminated. The mass loss during the N2 drying
step was assigned to physisorbed water while the mass loss dur-
ing the H2 stage and gain during the air stage were attributed to
the removal and re-incorporation of O into the cobalt structure,
respectively. The degree of reduction (DOR) could be determined
from mass loss during the H2 stage according to Eq. (1), but could
also be determined from the mass gain during the reoxidation step,
which gave similar values
 mass during reduction
Degree of reduction (DOR) = × 100
mass of O in Co3 O4 loaded
(1)
Co3 O4 reduces via formation of CoO as a stable intermediate
according to Scheme 3, which illustrates the two-step reduction
sequence of a single crystallite. Not all the O removed from the
Co3 O4 structure (i.e.  mass during reduction in Eq. (1)) directly
results in the formation of the metallic Co0 phase as would be
expected in the case of a single reduction step. According to an
ideal 2-step reduction, nucleation of a pure CoO phase occurs after
removal of 25% of the O from the structure of Co3 O4 (Scheme 3) and
so formation of a metallic phase requires a minimum threshold DOR
of 25% as described by Eq. (1). It must be cautioned that a value of
the DOR calculated using Eq. (1) is not necessarily representative of
each individual cobalt crystallite as implied by the ideal reduction
sequence in Scheme 3. Rather, the value of the DOR is cumulative
over the multiple crystallites contained in each catalyst sample.
2.4. X-ray diffraction
A Bruker D8 Advance Diffractometer with a Co K␣ radiation
source ( = 1.78897 Å) was used for the XRD analyses. In situ anal-
yses were carried out during catalyst activation (350 ◦ C, pure H2
at 1440 ml min−1 gCo−1 , 1 bar, 12 h) in an Anton-Paar XRK-900
reaction chamber. Approximately 25 diffractograms were collected
(during the 12-h reduction) over a 2 range of 40–95◦ at a scan rate
of 2 = 0.0123◦ /step and scan time of 0.2 s/step. Rietveld refinement
was carried out using TOPAS 4.2® software to determine crystallite
sizes and composition of the cobalt phase. A ‘peaks phase’ group
model was defined to fit the Al2 O3 peaks because the Puralox SCCa
5-150 support is a mixture of ␥-Al2 O3 and a smaller amount of
␦-Al2 O3 whose crystal structure is not well known [40].
2.5. X-ray photoelectron spectroscopy (XPS)
A VG Escalab 200 MKII Spectrometer with an Al K␣ radiation
source (Al K␣ = 1486.3 eV) was used for the XPS analyses of the cal-
cined samples. Quasi-in situ XPS analyses of the reduced catalyst
samples were carried out in the Kratos AXIS Ultra Spectrome-
ter equipped with a monochromatic Al K␣ radiation source (Al
K␣ = 1486.6 eV). The catalysts were activated in a tubular quartz
reactor (H2 , 350 ◦ C, 12 h) which was subsequently pressurised and
the inlet and outlet sealed. Thereafter, the reactor was transferred
to an Argon-flushed glove box to minimise the risk of reoxidation
during sample preparation before the samples were transferred
to the spectrometer for analysis. The tabulated values of the
20 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
Fig. 1. (a) TPR-TCD profile with time of Co/Al2 O3 (prepared on milled Al2 O3 ). Gaus-
sian peaks were used to qualitatively fit the profile and so absolute peak areas do
not necessarily correspond to the percentage (%) of H2 consumed in each specific
reaction. (b) The TPR-MS profiles with time. The evolution of ion with m/z of 46
(NO2 ) was monitored but not detected.
electron binding energies of various elements were available from
the ‘Handbook of X-ray Photoelectron Spectroscopy’ [41] and
CasaXPS® software was used for peak deconvolution.
2.6. Low temperature Fischer–Tropsch testing
Fischer–Tropsch tests were carried out in a 300 mm-long
packed bed reactor. The selected catalyst activation treatment
was reduction in pure H2 (720 ml min−1 gCo−1 ) at 350 ◦ C and
atmospheric pressure for 12 h. The Fischer–Tropsch reaction con-
ditions used were 20 barg, 220 ◦ C, H2 /CO = 2, and a GHSV(H2 +CO)
of 2400 ml (min gCo)−1 . A GC-TCD was used for on-line analysis of
the permanent gases (H2 , CO, CH4 ). An ampoule technique devel-
oped by Schulz et al. [42–44] was used for off-line sampling of the
gaseous hydrocarbon product, which was subsequently analysed
using a Varian 3900 GC-FID.
3. Results and discussion
3.1. Temperature-programmed reduction (TPR)
The TPR-TCD profile of unpromoted Co/Al2 O3 is shown in Fig. 1a.
Five distinct reduction peaks were identified in the profile which
has been deconvoluted using Gaussian peaks to indicate the pro-
posed peak positions. The effluent of the reduction process was also
monitored using a GC–MS and the obtained profiles are shown in
Fig. 1b. The MS profile of H2 was consistent with the TCD signal,
while that of H2 O mirrored the TCD signal, indicating that water
was a co-product of the various reduction processes.
Peak 1 in the TPR profile coincided with evolution of NO in
the MS profile, and was attributed to the reduction of nitrate
species. Nitrates were remnants of the Co(NO3 )2 precursor, and
were also observed using XPS as low-intensity N 1s peaks at
408 eV. Nitrates may have persisted as cobalt compounds or as ions
adsorbed on the support without necessarily forming a crystalline
cobalt phase. Co(NO3 )2 peaks were not observed using XRD so the
low nitrate quantities (as suggested by XPS) were likely to have
been present as small, finely dispersed and/or amorphous particles
that lacked long-range ordering. Following the TPR – MS analy-
sis, deductions from other studies and techniques including XPS,
in situ TPR-XRD and a series of H2 TPR-TCD analyses of various
reference samples were used to identify the reduction peaks as
follows: Peak 1: Co(NO3 )2 , NO3 − , and potentially CoO(OH) [45];
Peak 2: Co3 O4 ; Peak 3: CoO; Peak 4: CoO in close interaction
with Al2 O3 ; Peak 5: non-stoichiometric cobalt alumina complexes
and/or CoAl2 O4 .
Fig. 2 shows the TPR profiles of the first set of catalysts prepared
using unmodified Al2 O3 . Pt/Al2 O3 did not exhibit any reduction
peaks (i.e. it was fully reduced by 60 ◦ C – the start of the TPR analy-
sis). The reduction peaks of hybrid Pt–Co appeared at near identical
temperatures to those of unpromoted Co/Al2 O3 , which implied that
the reduction had followed a similar process. However, the profile
of Pt–Co/Al2 O3 (in which Pt and Co were in contact/close proximity)
shifted to lower temperatures, which signified an improvement in
the overall cobalt reducibility compared to the unpromoted sam-
ple. The profile of hybrid Pt–Co was contrary to expectations of
a reduction mechanism involving spillover hydrogen. The reason
for the poor reducibility of hybrid Pt–Co compared to Pt–Co/Al2 O3
was hypothesized to involve the physicochemical dissimilarities
between the samples. It was likely that besides the longer distances
between Pt and Co on separate supports, the poor intimate contact
between the fairly large support particles contributed to surface
discontinuity. The latter case was detrimental to the surface diffu-
sion of hydrogen species which now had to ‘jump’ between adjacent
support particles.
The importance of intimate contact between the support par-
ticles (and thus surface continuity) to the successful migration
of spillover hydrogen has been reported in other studies [16,25].
Hilmen et al. [16] only observed an improvement in the cobalt
reducibility in hybrid Re–Co (Re/Al2 O3 + Co/Al2 O3 ) after crushing
and pressing the sample to obtain intimately mixed particles of
425–850 ␮m. In a study of CoMo catalysts, Baeza et al. [25] demon-
strated that monolithic support particles were highly conducive
for spillover compared to samples where Co and Mo were on sepa-
rate supports. Spillover was enhanced when using Al2 O3 monoliths,
which were cylindrical extrudates with Co and Mo crystallites at
each extremity. As a single continuous particle, the Al2 O3 mono-
lith was proposed to behave like a very fine powder with a very
high surface area and thus multiple contact points between adja-
cent particles to allow for smooth migration of surface hydrogen
[25].
In light of these deductions, the second set of catalysts was pre-
pared using milled Al2 O3 as the support. In these catalysts, intimate
contact between adjacent support particles was enhanced because
the smaller particle size resulted in a higher contact area per unit
mass. The TPR profiles of the new catalysts supported on milled
Al2 O3 are shown in Fig. 3. The reducibility increased as follows:
Co/Al2 O3 < hybrid Pt–Co < Pt–Co/Al2 O3 . The unpromoted sample
continued to reduce even after 900 ◦ C, while the reduction of hybrid
Pt–Co was complete at 650 ◦ C, which was only 70 ◦ C after that of
Pt–Co/Al2 O3 . Fig. 3 indicates that direct Pt–Co contact or coordina-
tion was not required to observe the reduction promotion effect
in promoted Co-based Fischer–Tropsch catalysts. Furthermore,
hydrogen could migrate over adjacent Al2 O3 particles, perhaps
via surface OH groups according to the ‘bucket-brigade’ model to
 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
Fig. 2. TPR profiles of the catalyst samples prepared using unmodified Al2 O3.
ultimately reduce cobalt oxides at faster rates that in the unpro-
moted sample.
A noteworthy observation was the fact that the high temper-
ature peak (Peak 5 in Fig. 2) that was assigned to the reduction
of non-stoichiometric cobalt aluminates completely disappeared
in the hybrid Pt–Co sample. Non-stoichiometric cobalt aluminates
form more easily when Co ions are exposed to a high partial
pressure of water and high temperatures such as during cal-
cination [16,46,47]. It has been proposed that some promoters
may prevent the formation of mixed support oxides by inti-
mate interaction (physical and/or chemical) with the cobalt or
by preferentially depositing as a layer between cobalt crystal-
lites and the support [17,48]. Based on hybrid Pt–Co, direct Pt/Co
contact was not required to prevent the formation of mixed
oxides since the peak of non-stoichiometric cobalt aluminates
was absent. In hybrid Pt–Co, Pt facilitated the reduction of cobalt
species via remote-control, which allowed Co ions to reduce at
sufficiently low temperatures and consequently prevented the for-
mation of mixed cobalt–alumina complexes. The latter deduction
is significant, because it suggests that the mixed cobalt–alumina
complexes often observed in the reduction profile of unpro-
moted Co/Al2 O3 do not actually form during calcination, but rather
during the TPR analyses themselves. A similar conclusion was
21
reached by Hilmen et al. [16] who studied the Re–Co–alumina
system.
3.2. Isothermal thermogravimetric analyses (TGA) during
catalyst activation
Fig. 4 shows the reduction curves (DOR as a function of time)
determined using the TGA results. The horizontal line at the thresh-
old DOR of 25% is derived from the 2-step reduction of an ideal
Co3 O4 crystallite (Scheme 3) where formation of the Co0 phase
requires removal of more than 25% of the O from the structure of
Co3 O4 . After the 12-h activation in H2 , the values of the DOR were
as follows: Co/Al2 O3 (30.9%) < hybrid Pt–Co (71.8%) < Pt–Co/Al2 O3
(89.8%). The trend was consistent with TPR and showed that Pt
improved the reducibility of Co even without Pt and Co contact in
the hybrid sample.
It was interesting that despite a low final DOR of 30.9%, the
unpromoted Co/Al2 O3 sample rapidly attained the threshold DOR
of 25%. In fact, the threshold DOR was attained in the first 15 min,
but yet it took an additional 10.6 h to reach the final DOR of 30.9%.
Hence, the reduction rate from 25% → 30.9% was 200 times slower
than that from 0% → 25%. Based on the two-step reduction of an
ideal Co3 O4 crystallite (Scheme 3), attaining the threshold DOR,
22 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
Fig. 3. TPR profiles of catalyst samples prepared using milled Al2 O3.
Fig. 4. Reduction curves determined from thermogravimetric analyses during
isothermal reduction in H2 at 350 ◦ C. Samples prepared on milled Al2 O3 .
which incidentally corresponds to the reduction of Co3 O4 → CoO
was facile, while the kinetics of CoO → Co0 reduction were severely
hindered.
In order to investigate the threshold DOR, the reduction curve of
Co/Al2 O3 was compared to that of an unsupported Co3 O4 sample
from Batley et al. [49] shown in Fig. 5. Batley et al. [49] used TGA to
analyse Co3 O4 powder during isothermal reduction at 256 ◦ C. Even
at this lower reduction temperature (i.e. 256 versus 350 ◦ C), the
reduction rate of Co3 O4 was significantly faster in the absence of a
support. This indicated that the reduction kinetics of Co3 O4 → CoO
versus CoO → Co0 did not differ as greatly in unsupported Co3 O4
as they did in the presence of Al2 O3 . It is proposed that the rapid
decline in the reduction rate of Co/Al2 O3 after the threshold DOR
Fig. 5. Reduction curve of unpromoted Co/Al2 O3 (prepared on milled Al2 O3 ) overlaid
with a curve of unsupported Co3 O4 from Batley et al. [49]. The x-axis is shown only
for the first 15 min of the 12-h reduction of Co/Al2 O3 .
was induced by interaction of the Co oxides with the Al2 O3 support.
The stabilising effect of the support appeared to have affected the
reduction of the CoO intermediate oxide to a greater extent than
Co3 O4 . Accordingly, the reduction curve of a hypothetical Co/Al2 O3
sample in which Al2 O3 affected the reduction kinetics of Co3 O4
and CoO to the same extent was imagined free of the rapid decline
in the rate after the threshold DOR was attained, and thus would
mimic that of unsupported Co3 O4 . From the reduction curves in
Fig. 4, it was apparent that the catalysing effect of Pt became more
pronounced after the threshold DOR, when Pt helped overcome the
drastic decline in the rate of reduction. In fact, the presence of Pt
appeared to better align the reduction curves of Al2 O3 -supported
Co3 O4 with that of the unsupported Co3 O4 sample that was devoid
of metal–support interactions.
Fig. 6 shows the curves of the differential reduction rates as a
function of the degree of reduction (DOR). Each curve was charac-
terised by three stages: (i) an initial increase to attain the maximum
rate, followed by (ii) a rapid decline to very low rates until the
threshold DOR, after which (iii) the decline in the rate continued,
albeit very slowly, resulting in extended tails until the maximum
DOR was reached. In all three stages, the reduction rate was great-
est in Pt–Co/Al2 O3 followed by hybrid Pt–Co, which demonstrated
improved reduction kinetics in the presence of Pt. Nevertheless,
the low reduction rates after the threshold DOR meant that com-
plete reduction was not achieved by any of the catalyst samples.
To illustrate the negative effect of the Al2 O3 support, a curve of
the unsupported Co3 O4 sample derived from Batley et al. [49] is
included in Fig. 6. Even at the low reduction temperature of 256 ◦ C,
unsupported Co3 O4 reduced at considerably higher rates and did
not show a decline in the reduction rate at the threshold DOR.
It is proposed that the reduction rates declined greatly after the
threshold DOR because Al3+ ions in the support stabilised the CoO
intermediate oxide, resulting in the slow kinetics of CoO → Co0
reduction. On the other hand, the reduction rate of unsupported
Co3 O4 did not decline to very low values once the threshold DOR
was achieved because the sample was not subject to metal–support
interactions.
 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
Fig. 6. Differential reduction rates as a function of the degree of reduction of cata-
lysts samples prepared on milled Al2 O3 .
Arnoldy and Moulijn [50] used the concept of polarisation
to explain the stabilising effect of Al2 O3 on supported CoO: the
covalent character of the Co–O bond (based on ‘Fajans’ rules’ of
covalency) suggests that it can be polarised by the Al3+ ions in the
support. Polarisation increases the effective charge of Co and thus
the lattice energy, which results in the need for higher reduction
temperatures for supported cobalt oxides. In accordance with this
argument, it is suggested that the polarisation of the Co–O bonds
by Al3+ ions would be even greater in the CoO intermediate oxide,
which has a lower O/Co ratio than Co3 O4 . The much stronger nega-
tive influence of oxidic supports on the reduction of CoO compared
to that of Co3 O4 has indeed been observed in other studies [51–54].
It was noteworthy that the reduction curves in Figs. 4 and 5
had sigmoid shapes (less visible in Fig. 4 due to the time scale)
while the rate curves in Fig. 6 had the bell-shapes, both of which are
characteristic of the nucleation-growth kinetic model (and/or auto-
catalysis) for the reduction of transition metals oxides [9,13,55–57].
The shapes of these curves are consistent with an increase in the
rate as the reduction proceeds, but it is unclear whether the kinetic
models typically presented for unsupported metal oxide particles
directly apply to the reduction of cobalt nano-crystallites supported
on oxides such as Al2 O3 .
3.3. In situ XRD during catalyst activation
The diffractograms of the catalysts after the 12-h reduction at
350 ◦ C are shown in Fig. 7. CoO and fcc Co were the only cobalt
phases observed after the reduction. Rietveld refinement was used
to quantify the Co0 content within the cobalt phase of the various
samples (%Co0 shown inset in Fig. 7), which increased as fol-
lows Co/Al2 O3 (0%) < hybrid Pt–Co (25%) < Pt–Co/Al2 O3 (89%). The
reducibility of the unpromoted Co/Al2 O3 sample was poor and the
only observable cobalt phase after reduction was CoO while the Pt
promoter significantly improved the reducibility of cobalt oxides
in hybrid Pt–Co and even more so in Pt–Co/Al2 O3 .
Fig. 8 shows the intensity at the peak maxima of the charac-
teristic cobalt peaks as a function of time, i.e. Ii = f(t). The figures
were generated from the diffractograms collected during the in situ
reduction analyses. For Co/Al2 O3 , the drop in the intensity of Co3 O4
coincided with an increase in that of CoO in accordance with the 2-
step reduction (Fig. 8a). The intensity of CoO reached a maximum at
approximately 1 h, and then declined by just 4.6% by the end of the
23
Fig. 7. In situ XRD diffractograms of the samples prepared on milled Al2 O3 after
reduction in H2 at 350 ◦ C. The diffractogram of calcined Co/Al2 O3 is overlaid with
that of the reduced sample. Prior to the analysis, Co/Al2 O3 and Pt–Co/Al2 O3 were
diluted with Al2 O3 to obtain the same Al2 O3 /Co ratio in the hybrid sample.
analysis. The intensity at the 2 value of fcc Co (51.7◦ ) increased by
approximately 6.1% in the same period even though distinct fcc Co
peaks were never observed in reduced Co/Al2 O3 (Fig. 7). The curve
of the peak intensity of the CoO phase concurred with the reduction
curve in Fig. 4 and showed that reduction to the threshold DOR (i.e.
to form CoO) was facile and occurred rapidly following exposure to
H2 , but further reduction of CoO → Co0 was very slow.
Pt had a minimal effect on the rate at which the intensity of the
Co3 O4 peak decreased in favour of CoO, but its catalysing effect was
evident in the higher rates at which the intensity of the CoO peak
decreased in favour of Co0 in the Pt–Co/Al2 O3 and hybrid Pt–Co
samples. The curves of the maximum peak intensities as a func-
tion of time, i.e. Ii = f(t) in Fig. 8 were used to quantify the reduction
rates by assuming first order kinetics and assigning rate constants
‘k1 ’ and ‘k2 ’ to the reduction of Co3 O4 → CoO and CoO → Co0 respec-
tively in the unpromoted Co/Al2 O3 sample. A correlation was then
assumed between the peak intensity and the quantity of each reduc-
ing oxide species so that the rate of change of the peak intensity
correlated with the reduction rate, i.e. dIi /dt = k · Ii . The 1st order
rate constants ‘k1 ’ and ‘k2 ’ could then be calculated for Co/Al2 O3
and are shown inset in Fig. 8a. The rate constant for CoO → Co0
reduction (‘k2 ’) was several orders of magnitude smaller than that
for Co3 O4 → CoO (‘k1 ’). The rate constants determined for the Pt
containing samples were calculated relative to those in the unpro-
moted sample and are included inset in Fig. 8b (hybrid Pt–Co) and
Fig. 8c (Pt–Co/Al2 O3 ). The reduction of Co3 O4 → CoO was facile even
in the unpromoted sample and so the corresponding rate constant
k1 was hardly affected by Pt. However, the rate constant k2 , corre-
sponding to the reduction of CoO → Co0 was 15 and 61 times greater
in hybrid Pt–Co and Pt–Co/Al2 O3 , respectively. It is acknowledged
that the kinetics of metal oxide reduction are more complex, and
the assumption of first order kinetics simply provided a tentative
means to quantify the influence of Pt on the two-step/consecutive
reduction using the shapes of the curves in Fig. 8.
3.4. Quasi in situ XPS
The Co 2p regions of the XPS spectra of the activated samples
are shown in Fig. 9. The characteristic Co 2p doublet (Co 2p3/2 and
Co 2p1/2 ) due to spin–orbit splitting was observed in all samples. In
elemental Co, the main photoelectron lines are present at 793 eV
24 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
Fig. 8. Intensity at the peak maxima of CoO (49.7◦ ), Co3 O4 (70.3◦ ) and Co0 /fcc Co (51.7◦ ) peaks as a function of time during in situ XRD analyses of the isothermal reduction
at 350 ◦ C in H2 . Catalysts prepared using milled Al2 O3 ; ‘k1 ’ and ‘k2 ’ are rate constants assuming 1st order kinetics during the reduction of Co3 O4 .
Fig. 9. Co 2p regions of the XPS spectrum of reduced samples (prepared on milled
Al2 O3 ). The spectra have been deconvoluted with Gaussian peaks to highlight con-
tributions from Co2+ , Co0 and shake-up satellites. Main photoelectron lines for
elemental Co indicated by vertical lines at 793 eV (Co 2p1/2 ) and 778 eV (Co 2p3/2 ).
(Co 2p1/2 ) and 778 eV (Co 2p3/2 ) and these are indicated as verti-
cal lines in Fig. 9. The binding energy of the relevant cobalt species
increases as follows: Co0 < Co2+ < Co3+ ; the shake-up satellites are
not present with Co0 , while they are more intense and closer to the
main elemental photoelectron lines with Co2+ compared to Co3+ .
The Co 2p regions in Fig. 9 were deconvoluted using Gaussian peaks
corresponding to Co0 , Co2+ and shake-up satellites but no peaks
were assigned to Co3+ . Following reduction, the main photoelec-
tron lines shifted towards the lower binding energy of elemental
Co from the characteristic positions of Co3 O4 . The Co 2p peaks in the
unpromoted Co/Al2 O3 displayed stronger shake-up satellites com-
pared to the calcined sample, and the peaks were consistent with
Co2+ compounds like CoAl2 O4 and CoO [58,59]. Besides stronger
shake-up satellites compared to the unreduced calcined samples,
the Co 2p3/2 peaks of hybrid Pt–Co and Pt–Co/Al2 O3 also exhibited
shoulders close to the binding energy of elemental Co. The sizes
of the Co0 Gaussian peaks increased as follows: Co/Al2 O3 > hybrid
Pt–Co > Pt–Co/Al2 O3 , and the opposite trend was observed with the
CoO. These trends in the surface specific Co0 content correlated
with the values of the DOR determined from the thermogravimet-
ric analyses (Fig. 4), and the compositions obtained from Rietveld
refinement of in situ XRD data (Fig. 7).
3.5. Fischer–Tropsch testing
3.5.1. Effects of Pt on the carbon monoxide conversion
The values of the steady state CO conversion, cobalt-time
yield, turnover frequency (TOF) and product selectivity are sum-
marised in Table 1, and presented in greater detail in Nabaho
[60]. The CO conversion and cobalt-time yield increased as fol-
lows: Co/Al2 O3 < hybrid Pt–Co < Pt–Co/Al2 O3 . The Pt/Al2 O3 sample
was also tested but had a very low conversion of ca. 0.02%. The
observed trends were consistent with the higher quantity of Co0
per gram of catalyst after catalyst activation in the presence of
Pt. However, catalytic activity is not a function of the quantity of
Co0 as defined by cobalt-time yield, but rather the quantity Co0
on the surface as described by the TOF, which increased as follows:
Co/Al2 O3 < hybrid Pt–Co  Pt–Co/Al2 O3 . Hybrid Pt–Co showed only
a slight improvement in the TOF compared to Co/Al2 O3 . However,
the TOF of Pt–Co/Al2 O3 was twice that of the unpromoted Co/Al2 O3 ,
which suggested that a synergy arose when Pt was present under
Fischer–Tropsch conditions.
 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
Table 1
Cobalt-time yield, turnover frequency (TOF) and product selectivitya of samples prepared on milled Al2 O3 .
Sample XCO at 24 h, % Cobalt-time yield ×103 s−1
Co/Al2 O3 9.5 4.8
Hybrid Pt–Co 13 6.6
Pt–Co/Al2 O3 19 9.2
a
Product selectivity of all samples determined at CO conversion of ca. 12%.
A higher TOF is corroborated by other studies in which high
promoter-to-cobalt mass ratios were used (i.e. >0.1 wt%) [7,61,62].
SSITKA of methanation studies have shown that even though the
apparent TOF increases in the presence of noble metal promoters,
the true intrinsic TOF of Co0 remains unaffected [7,61,63]. A con-
stant intrinsic TOF of Co0 is obtained when the effects of surface
coverage are decoupled from those of the observed TOF. The higher
TOF is instead attributed to a greater coverage of reactive interme-
diates (CHX ) in the presence of reduction promoters. During the
reaction, a low coverage of reactive intermediates might arise when
a portion of the surface is blocked by unreactive species like carbon
and near surface or even subsurface oxygen/oxides that accumulate
on the cobalt surface following C–O bond breaking [17,7,62]. Pro-
moters can spillover hydrogen which may have a ‘cleansing’ effect
by removing unreactive species from the cobalt surface [17,62],
and studies suggest that Pt–Co contact is not required to obtain an
increase in the TOF [17]. Besides a cleansing effect, an increase in
the localised H/CO ratio such as would arise from a hydrogen-rich
microenvironment generated by spillover hydrogen is also well-
known to correlate positively with the reaction rate within a range
of conditions [64,65].
3.5.2. The effects of Pt on the product distribution of Co/Al2 O3
Pt–Co/Al2 O3 : At the same conversion of ca. 12%, Pt was observed
to have a distinct influence on the hydrogenation activity in
Pt–Co/Al2 O3 , which was characterised by higher CH4 selectiv-
ity and lower values of the C5 + selectivity, olefin selectivity and
chain growth probability (˛). These trends were consistent with
expectations of hydrogen spillover contributing to the enhanced
hydrogenation activity in the presence of the Pt. Generally, the high
hydrogenation effect is pronounced at noble metal promoter load-
ings greater than 0.1% [13] and so detrimental effects such as an
increase in the CH4 selectivity are minimised by use of the very
low Pt loadings in commercial catalysts.
Besides spillover, other effects have been suggested to influence
the product distribution such as enhanced WGS activity and smaller
crystallite size distribution. Pt has been cited as the cause of the
higher amount of CO2 obtained with the promoted cobalt catalyst
[5]. The H2 produced from the WGS reaction may have a similar
effect as spillover since it would also increase the localised H/CO
ratio. Unfortunately, low amounts of CO2 could not be detected in
this study due to the limitations of the experimental set-up. Crys-
tallite size has also been observed to influence the CH4 selectivity
[66–68], but Rietveld refinement of the XRD diffractograms of the
reduced samples showed that the crystallite sizes of Co0 /CoO in
Pt–Co/Al2 O3 and Co/Al2 O3 were in the same range (Fig. 7).
Hybrid Pt–Co: The effect of Pt on the product distribution of
Hybrid Pt–Co was less clear and in some cases contradictory. The
C5 + selectivity and chain growth probability decreased, which was
in agreement with increased hydrogenation activity. However, the
CH4 selectivity remained unchanged while the olefin selectivity
increased, which was contrary to expectations of a higher localised
H/CO ratio due to spillover. Even though dramatic evidence to sup-
port a spillover mechanism was obtained during the reduction of
hybrid Pt–Co, it appeared that Pt–Co contact was important to
observe the effects of Pt under Fischer–Tropsch conditions. Pt–Co
25
TOF, ×103 s−1 CH4 selectivity, %C C5 + selectivity, %C ˛ Olefin selectivity, %
C2 C3 C5
45 21 65 0.85 44.8 50.4 41.5
51 21 63 0.81 46.2 59.2 52.6
91 30 57 0.81 26.7 29.5 27.3
co-ordination is known to occur readily [5,9,10,12,15] but SSITKA of
methanation studies indicate that the Pt does not affect the intrin-
sic activity of Co [7,61,63]. It is suggested that electronic/bonding
effects were unlikely to have contributed to the modification of
the product distribution in Pt–Co/Al2 O3 . Nevertheless, direct Pt–Co
contact may play an important role in ensuring the efficiency of the
transfer of spillover species from Pt to Co. This could be seen during
the reduction, when the improvement in Co reducibility attributed
to hydrogen spillover was diminished in the hybrid Pt–Co due to
the increased distance between Pt and Co.
Under Fischer–Tropsch conditions, the diminished effect of Pt
due to separation from Co is further compounded by CO and
hydrocarbons that are not present during reduction. The effect of
CO poisoning to the detriment of H2 adsorption may hinder the
hydrogen spillover process; CO poisoning of Pt to the detriment of
hydrogenation reactions is a well known in PEM fuel cells [69–71]
and in olefin hydrogenation reactions [72,73]. CO poisoning on Pt
occurs because it binds more strongly and has a higher heat of
adsorption than H [69–71]. The barrier for CO dissociation on Pt
is also significantly higher compared to other metals like Co [74].
In the presence of CO, H2 adsorption may be favoured on alternative
sites like the Pt–Al2 O3 interface [72,73] or at other (limited) sites
on the Pt surface since the maximum CO coverage on Pt ranges
between 0.5 and 0.75 ML [75–77].
It is suggested that the inhibition of the Pt surface by
CO has unfavourable consequences for Pt → Co spillover under
Fischer–Tropsch conditions compared to during the reduction. Dur-
ing catalyst activation, most Pt active sites are available to H2 , with
limited competition from other molecules. During the LTFT reac-
tion, the CO in syngas may effectively block a large portion of the
Pt surface, making it unavailable for H2 adsorption and subsequent
spillover. Therefore, even though the detrimental effect of CO on
H2 spillover also arose in Pt–Co/Al2 O3 , it was compounded in the
hybrid sample in which the efficiency of the spillover process was
already low as evidenced from the poor reducibility when hybrid
Pt–Co was prepared using unmilled unmodified Al2 O3 (Fig. 2).
4. Conclusions
A mechanical mixture of 20 wt%Co/Al2 O3 + 0.5 wt%Pt/Al2 O3 ,
which allowed for decoupling of hydrogen spillover effects from
those that require direct Pt–Co contact was analysed in order to
explicate the role of Pt as a promoter.
• The catalysing effect of Pt on the reduction of supported Co3 O4 in
the hybrid Pt–Co sample was not observed when large unmodi-
fied Al2 O3 support particles were used. However, the reducibility
improved after the inter-support contact area per unit mass was
increased by milling. The observed improvement in reduction
was evidence of hydrogen transportation via surface migration on
the Al2 O3 support (e.g. according to the ‘bucket-brigade’ model)
because the transfer of hydrogen from Pt to Co by other means
such as via gas-phase transportation, would not have been depen-
dent on the diameter of the support particles.
26 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
• Dramatic evidence for the improved reducibility of hybrid Pt–Co
prepared on milled Al2 O3 demonstrated that Pt–Co contact was
not required for the catalysing effect of Pt.
• Besides the improved reducibility, the TPR profile of hybrid Pt–Co
did not have the peak of non-stoichiometric cobalt aluminate
complexes, which suggested that these species form during the
TPR analysis rather than during catalyst preparation. Intimate
physical and/or chemical interaction with Pt is not required to
prevent the formation of the mixed cobalt support compounds.
• A multifaceted approach was used to quantify the effect of Pt
during catalyst activation. TGA and in situ XRD were employed
for bulk properties while quasi-in situ XPS was used to analyse
the effect of Pt on the surface composition. Supported Co3 O4 was
deduced to reduce via the formation as CoO as a stable inter-
mediate oxide. Co3 O4 → CoO reduction occurred easily even in
the unpromoted sample, while the CoO → Co0 reduction kinetics
were severely hindered due to the influence of the Al2 O3 support.
Pt slightly improved the already facile Co3 O4 → CoO reduction,
but significantly increased the reduction rate of CoO → Co0 .
During the Fischer–Tropsch synthesis, Pt–Co/Al2 O3 and hybrid
Pt–Co had higher cobalt-time yields in accordance with their
improved reduction.
• The higher TOF and the highly hydrogenated product of
Pt–Co/Al2 O3 were consistent with expectations of hydrogen
spillover. An increase in the H/CO ratio correlates positively with
the reaction rate and may result in a ‘cleansing’ effect on the Co0
surface.
• Despite the improved reducibility, the effects of Pt under reac-
tive conditions were significantly diminished in hybrid Pt–Co.
It is suggested that the presence of CO resulted in a severely
diminished flux of spillover hydrogen, which was compounded
by the larger distance between Pt and Co separation compared to
Pt–Co/Al2 O3 . Therefore, the effect of the Pt promoter during the
Fischer–Tropsch reaction could only be tentatively attributed to
spillover effects.
The systematic investigation of the hydrogen spillover phe-
nomenon using hybrid catalysts at commercially relevant con-
ditions was successfully carried out in this study. Thus far, no
comparable investigation of the hydrogen spillover phenomenon
in the Pt–Co–alumina catalyst system during both reduction and
under real reactive conditions has been encountered in the pub-
lished literature.
Acknowledgment
Funding from c*change (DST-NRF Centre of Excellence in Catal-
ysis) is gratefully acknowledged.
References
[1] E. van Steen, M. Claeys, M.E. Dry, J. van de Loosdrecht, E.L. Viljoen, J.L. Visagie,
J. Phys. Chem. B 109 (2005) 3575–3577.
[2] A.M. Saib, D.J. Moodley, I.M. Ciobîcă, M.M. Hauman, B.H. Sigwebela, C.J.
Weststrate, J.W. Niemantsverdriet, J. van de Loosdrecht, Catal. Today 154
(2010) 271–282.
[3] F. Morales, B.M. Weckhuysen, Catalysis 19 (2006) 1–40.
[4] W. Ma, G. Jacobs, D.E. Sparks, M.K. Gnanamani, V.R.R. Pendyala, C.H. Yen, J.L.S.
Klettlinger, T.M. Tomsik, B.H. Davis, Fuel 90 (2011) 756–765.
[5] T. Jermwongratanachai, G. Jacobs, W.D. Shafer, V.R.R. Pendyala, W. Ma, M.K.
Gnanamani, S. Hopps, G.A. Thomas, B. Kitiyanan, S. Khalid, B.H. Davis, Catal.
Today 228 (2014) 15–21.
[6] T.K. Das, G. Jacobs, P.M. Patterson, W.A. Conner, J. Li, B.H. Davis, Fuel 82 (2003)
805–815.
[7] D. Schanke, S. Vada, E.A. Blekkan, A.M. Hilmen, A. Hoff, A. Holmen, J. Catal. 156
(1995) 85–95.
[8] G. Jacobs, T.K. Das, Y. Zhang, J. Li, G. Racoillet, B.H. Davis, Appl. Catal. A Gen.
233 (2002) 263–281.
[9] G. Jacobs, Y. Ji, B.H. Davis, D. Cronauer, A.J. Kropf, C.L. Marshall, Appl. Catal. A
Gen. 333 (2007) 177–191.
[10] T. Jermwongratanachai, G. Jacobs, W. Ma, W.D. Shafer, M.K. Gnanamani, P.
Gao, B. Kitiyanan, B.H. Davis, J.L.S. Klettlinger, H.Y. Chia, D.C. Cronauer, A.J.
Kropf, C.L. Marshall, Appl. Catal. A: Gen. 464 (2013) 165–180.
[11] T.K. Das, G. Jacobs, B.H. Davis, Catal. Lett. 101 (2005) 187–190.
[12] G. Jacobs, J.A. Chaney, P.M. Patterson, T.K. Das, J.C. Maillot, B.H. Davis, J.
Synchrotron Radiat. 11 (2004) 414–422.
[13] F. Diehl, A.Y. Khodakov, Oil Gas Sci. Technol. – Rev. l’IFP 64 (2008) 11–24.
[14] M. de Beer, A. Kunene, D. Nabaho, M. Claeys, E. van Steen, J.S. Afr, Inst. Min.
Metall. 114 (2014) 157–165.
[15] M.D. Shannon, C.M. Lok, J.L. Casci, J. Catal. 249 (2007) 41–51.
[16] A.M. Hilmen, D. Schanke, A. Holmen, Catal. Lett. 38 (1996) 143–147.
[17] S.K. Beaumont, S. Alayoglu, C. Specht, W.D. Michalak, W.D. Michalak, V.V.
Pushkarev, J. Guo, N. Kruse, G.A. Somorjai, J. Am. Chem. Soc. 136 (2014)
9898–9901.
[18] S.K. Beaumont, S. Alayoglu, C. Specht, N. Kruse, G.A. Somorjai, Nano Lett. 14
(2014) 4792–4796.
[19] N. Tsubaki, S. Sun, K. Fujimoto, J. Catal. 199 (2001) 236–246.
[20] L. Guczi, D. Bazin, I. Kovacs, L. Borko, Z. Schay, J. Lynch, P. Parent, C. Lafon, G.
Stefler, Z. Koppany, I. Sajo, Top. Catal. 20 (2002) 129–139.
[21] L. Guczi, Z. Schay, G. Stefler, F. Mizukami, J. Mol. Catal. A Chem. 141 (1999)
177–185.
[22] F. Roessner, U. Roland, J. Mol. Catal. A Chem. 112 (1996) 401–412.
[23] N. Escalona, R. García, G. Lagos, C. Navarrete, P. Baeza, F.J. Gil-Llambías, Catal.
Commun. 7 (2006) 1053–1056.
[24] J.F. Cevallos- Candau, W.C. Conner, J. Catal. 106 (1987) 378–385.
[25] P. Baeza, M. Villarroel, P. Ávila, A. López Agudo, B. Delmon, F.J. Gil-Llambías,
Appl. Catal. A Gen. 304 (2006) 109–115.
[26] P. Baeza, M.S. Ureta-Zañartu, N. Escalona, J. Ojeda, F.J. Gil-Llambías, B. Delmon,
Appl. Catal. A Gen. 274 (2004) 303–309.
[27] J. Ojeda, N. Escalona, P. Baeza, M. Escudey, F.J. Gil- Llambías, Chem. Commun.
(2003) 1608–1609.
[28] J. Conradie, J. Gracia, J.W. (Hans) Niemantsverdriet, J. Phys. Chem. C 116
(2012) 25362–25367.
[29] J.Y. Luo, W.S. Epling, Appl. Catal. B Environ. 97 (2010) 236–247.
[30] W.C. Conner, J.L. Falconer, Chem. Rev. 95 (1995) 759–788.
[31] A.F. Carley, H.A. Edwards, B. Mile, M.W. Roberts, C.C. Rowlands, J. Chem. Soc.
Faraday Trans. 90 (1994) 3341–3346.
[32] N. Agmon, The Grotthuss mechanism, Chem. Phys. Lett. 244 (1995) 456–462.
[33] R.B. Levy, M. Boudart, J. Catal. 32 (1974) 304–314.
[34] K. Gadgil, R.D. Gonzalez, J. Catal. 40 (1975) 190–196.
[35] G.A. Badun, B.F. Johnson, N.E. Shchepina, Mendeleev Commun. 19 (2009)
235–236.
[36] W.C. Neikam, M.A. Vannice, J. Catal. 27 (1972) 207–214.
[37] J.T. Miller, B.L. Meyers, F.S. Modica, G.S. Lane, M. Vaarkamp, D.C.
Koningsberger, J. Catal. 143 (1993) 395–408.
[38] J.M. Cece, R.D. Gonzalez, J. Catal. 28 (1973) 254–259.
[39] Y. Li, R.T. Yang, J. Am. Chem. Soc. 128 (2006) 8136–8137.
[40] W. Gao, Developments in High Temperature Corrosion and Protection of
Materials, Woodhead Publishing Ltd, Cambridge, 2008.
[41] K.D. Stickle, W.F. Sobol, P.E. Bomben, Handbook of X-Ray Photoelectron
Spectroscopy, Perkin Elmer, Eden Prairie, Minnesota, 1992.
[42] H. Schulz, M. Claeys, Appl. Catal. A Gen. 186 (1999) 91–107.
[43] S. Schulz, H. Nehren, Erdöl Kohle Erdgas Petrochem. 39 (1986) 93–94.
[44] H. Schulz, Appl. Catal. A Gen. 186 (1999) 3–12.
[45] J. van de Loosdrecht, S. Barradas, E.A. Caricato, N.G. Ngwenya, P.S. Nkwanyana,
M.A.S. Rawat, B.H. Sigwebela, P.J. van Berge, J.L. Visagie, Top. Catal. 26 (2003)
121–127.
[46] P.H. Bolt, Transition metal–aluminate formation in alumina-supported model
catalysts (PhD thesis), University of Utrecht, The Netherlands, 1994.
[47] Y. Zhang, D. Wei, S. Hammache, J.G. Goodwin, J. Catal. 188 (1999) 281–290.
[48] K. Jalama, N.J. Coville, D. Hildebrandt, D. Glasser, L.L. Jewell, Top. Catal. 44
(2007) 129–136.
[49] G.E. Batley, A. Ekstrom, D.A. Johnson, J. Catal. 34 (1974) 368–375.
[50] P. Arnoldy, J.C.M. de Jonge, J.A. Moulijn, J. Phys. Chem. 89 (1985) 4517–4526.
[51] A.Y. Khodakov, W. Chu, P. Fongarland, Chem. Rev. 107 (2007) 1692–1744.
[52] B.A. Sexton, A.E. Hughes, T.W. Turney, J. Catal. 97 (1986) 390–406.
[53] W.-J. Wang, Y.-W. Chen, Appl. Catal. 77 (1991) 223–233.
[54] A.Y. Khodakov, J. Lynch, D. Bazin, B. Rebours, N. Zanier, B. Moisson, P.
Chaumette, J. Catal. 168 (1997) 16–25.
[55] G. Jacobs, M.C. Ribeiro, W. Ma, Y. Ji, S. Khalid, P.T. Sumodjo, B.H. Davis, Appl.
Catal. A Gen. 361 (2009) 137–151.
[56] A. Auroux, Calorimetry and Thermal Methods in Catalysis, vol. 154,
Springer-Verlag, Berlin; Heidelberg, 2013.
[57] O.J. Wimmers, P. Arnoldy, J.A. Moulijn, J. Phys. Chem. 90 (1986) 1331–1337.
[58] D.J. Moodley, J. van de Loosdrecht, A.M. Saib, M.J. Overett, A.K. Datye, J.W.
Niemantsverdriet, Appl. Catal. A Gen. 354 (2009) 102–110.
[59] D.J. Moodley, On the deactivation of cobalt-based Fischer–Tropsch synthesis
catalysts (PhD thesis), Eindhoven University of Technology, The Netherlands,
2008.
[60] D. Nabaho, Hydrogen spillover in the Fischer–Tropsch synthesis: the roles of
platinum and gold as promoters in cobalt-based catalysts (PhD thesis),
University of Cape Town, South Africa, 2015.
 D. Nabaho et al. / Catalysis Today 261 (2016) 17–27
[61] S. Vada, A. Hoff, E. Ådnanes, D. Schanke, A. Holmen, Top. Catal. 2 (1995)
155–162.
[62] E. Iglesia, S.L. Soled, R.A. Fiato, G.H. Via, J. Catal. 143 (1993) 345–368.
[63] A. Kogelbauer, J.G. Goodwin Jr., R. Oukaci, J. Catal. 160 (1996) 125–133.
[64] E. van Steen, H. Schulz, Appl. Catal. A Gen. 186 (1999) 309–320.
[65] I.C. Yates, C.N. Satterfield, Energy Fuels 5 (1991) 168–173.
[66] A.Y. Khodakov, Catal. Today 144 (2009) 251–257.
[67] G.L. Bezemer, J.H. Bitter, H.P.C.E. Kuipers, H. Oosterbeek, J.E. Holewijn, X. Xu, F.
Kapteijn, A.J. van Dillen, K.P. de Jong, J. Am. Chem. Soc. 128 (2006) 3956–
3964.
[68] G.L. Bezemer, U. Falke, A.J. van Dillen, K.P. de Jong, Chem. Commun. 6 (2005)
731–733.
[69] G. Postole, A. Auroux, Int. J. Hydrogen Energy 36 (2011) 6817–6825.
[70] J.J. Baschuk, X. Li, Int. J. Energy Res. 25 (2001) 695–713.
27
[71] W. Vogel, L. Lundquist, P. Ross, P. Stonehart, Electrochim. Acta 20 (1975)
79–93.
[72] K.S. Hwang, M. Yang, J. Zhu, J. Grunes, G.A. Somorjai, J. Mol. Catal. A Chem.
204–205 (2003) 499–507.
[73] R. Rioux, R. Komor, H. Song, J.D. Hoefelmeyer, M. Grass, K. Niesz, P. Yang, G.A.
Somorjai, J. Catal. 254 (2008) 1–11.
[74] R.A. van Santen, I.M. Ciobîcă, E. van Steen, M.M. Ghouri, Adv. Catal. 54 (2011)
127–187.
[75] S. Baldelli, A.S. Eppler, E. Anderson, Y.R. Shen, G.A. Somorjai, J. Chem. Phys.
113 (2000) 5432–5438.
[76] P. Inkaew, Electrochemistry of carbon monoxide on platinum single-crystal
surfaces (PhD thesis), Texas Tech University, USA, 2008.
[77] A.D. Allian, K. Takanabe, K.L. Fujdala, X. Hao, T.J. Truex, J. Cai, C. Buda, M.
Neurock, E. Iglesia, J. Am. Chem. Soc. 133 (2011) 4498–4517.