Research Article

pubs.acs.org/acscatalysis

Cobalt-Based Fischer−Tropsch Activity and Selectivity as a Function

of Crystallite Size and Water Partial Pressure
N. Fischer, B. Clapham, T. Feltes, and M. Claeys*
Centre for Catalysis Research and c*change (DST-NRF Centre of Excellence in Catalysis), Department of Chemical Engineering,
University of Cape Town, Cape Town 7701, South Africa

ABSTRACT: Using well-defined supported cobalt nanocrystallites in a novel in situ sample

presentation device for laboratory X-ray diffractometers, we investigated the extensively
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

studied structure sensitivity of Fischer−Tropsch (FT) catalysts under simulated high

Downloaded via KING ABDULLAH UNIV SCI TECHLGY on April 16, 2023 at 13:07:18 (UTC).

conversion conditions, that is, at high water to synthesis gas ratios. This study has to be
regarded as a further small step toward a full understanding of the various processes
governing FT activity and selectivity. We were able to show, for two different crystallite sizes,
that water has an overall enhancing effect on carbon monoxide conversion and surface
specific turnover frequency on metallic surfaces and improves the overall product selectivity
with a decrease of methane selectivity. For small crystallites oxidation was observed at
elevated water partial pressures, which caused a decrease of activity. The selectivity to the
undesired product methane is suppressed in favor of chain growth. This influence on the
selectivity might originate from water-induced changes on the active sites responsible for
chain growth or by an inhibiting effect of water on methanation sites. Due to a stronger effect
of water on smaller crystallites, the impact on the methane selectivity reverses previously described trends of increasing methane
selectivity with decreasing crystallite size. Secondary olefin reactions, clearly more pronounced on smaller crystallites and under
“dry” conditions, are severely suppressed via the addition of water, resulting in a pseudo structure insensitivity of this class of
reactions.
KEYWORDS: Fischer−Tropsch synthesis, co-feeding of water, cobalt nanoparticles, activity, selectivity

1. INTRODUCTION work. Water has been reported to enhance sintering at high

To design the most active and selective catalyst at the lowest concentrations.5 However, water has also been reported to
possible raw-material cost is the most important target of all influence the activity6 as well as selectivity6d−f,7 of a Fischer−
catalyst-producing companies.1 For metal-based heterogeneous Tropsch catalyst. Therefore, some of the previously observed
catalysts, the main cost factor is usually the active metal phase. activity and selectivity trends reported for different crystallite
To minimize its content without losing the desired activity, sizes under “dry” conditions might not hold under realistic
metals are often deposited as small crystallites, down to the industrial conversion conditions. It is desired to test the effect
single-digit nanometer range, on an inert carrier material so as of water on the catalytic performance with simulated
to maximize the mass specific surface area. Some reaction conversions at and beyond currently industrially employed
mechanisms, however, have been reported to require a certain conditions. Although not relevant today, one has to be aware
arrangement of surface atoms, a so termed minimum required that it is of great interest to industry to intensify the cobalt-
ensemble, to provide the desired performance.2 In the absence based FT process by increasing the per-pass conversion.
of these ensembles the catalyst may show no or undesired Currently set conversion is limited mainly through the high
activity/selectivity. Two prominent reactions showing such a water partial pressures, which are detrimental to the catalyst’s
catalyst structure sensitivity are the CO oxidation over gold lifetime and, as a result, render the current process
catalysts3 and the here-studied Fischer−Tropsch (FT) syn- uneconomical.8 To achieve this goal it is imperative to exactly
thesis.4 The majority of the latter studies were carried out under understand the effect of water on stability9 as well as on the
mild reaction conditions to avoid crystallite growth of the activity and selectivity of the cobalt catalyst system, especially as
employed model catalyst systems due to reaction temperature currently used rate equations for cobalt-based FT to not
and product water. The study presented here has to be contain a term accounting for the water partial pressure in
regarded as a first step to elucidate the crystallite size effect contrast to iron-based FT catalysts.8,10
under industrial conditions, focusing solely on the intrinsic In this study we combine the approach to use well-defined
crystallite size effect. Variables such as the presence of supported cobalt crystallites as model catalyst systems and first
hydrocarbons, in both gas and liquid phases, and the CO2
partial pressure as well as the presence of product water will Received: July 1, 2014
have to be added piece by piece to build a concise Revised: November 9, 2014
understanding of the different effects and their interplay at Published: November 14, 2014

© 2014 American Chemical Society 113 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121
ACS Catalysis Research Article

Table 1. Reverse Micelle Composition for the Preparation of the Studied Cobalt Model Catalysts
mass (g)
catalyst n-hexane Berol 050 water Co(NO3)2·6H2O NH4(OH) solution (mL) ωa
A 999.90 136.65 52.76 5.12 5.40 0.4
B 2000.01 273.59 54.24 1.32 1.35 0.2
a
Molar water to surfactant ratio.

expose them to mild Fischer−Tropsch conditions, that is, low conditions, the cobalt oxide phase (Co3O4) was reduced in a
temperature and high space velocity resulting in low hydrogen gas flow of 80 mLNPT·min−1·g−1 at atmospheric
conversion. In addition, we co-feed defined amounts of water pressure at 450 °C for 6 h using a ramping rate of 1 °C·min−1.
to the synthesis gas mixture simulating high conversion The reduced catalyst was cooled under hydrogen to 220 °C and
conditions without generating concentration gradients over exposed to a flow of the synthesis gas mixture consisting of
the catalyst bed. To correlate the observed conversion and hydrogen, carbon monoxide, and argon in a ratio of 6:3:1 at 80
selectivities with the actual working catalyst system, all mLNPT·min−1·g−1. The synthesis gas partial pressure was
experiments were conducted in an in-house developed in situ maintained throughout the experiments at 1 bar. In conjunction
X-ray diffraction cell assembly.9,11 with the high space velocity, the resulting low conversions
(<10% CO conversion) guarantee a negligible partial pressure
2. EXPERIMENTAL SECTION of product water and the absence of significant reactant
concentration profiles along the catalyst bed. Higher con-
2.1. Catalyst Preparation. Two alumina-supported cobalt versions were simulated via the stepwise addition of water
model catalysts with different cobalt crystallite sizes were vapor via a saturator, the temperature of which determined the
prepared using a reverse micelle method as previously resulting water vapor or partial pressures, respectively (see
described,12 with the exact compositions given in Table 1. In Table 2). The total pressure was increased according to the
short, an aqueous solution of cobalt nitrate hexahydrate
(Sigma-Aldrich) was added to a mixture of n-hexane (Kimix, Table 2. Water Partial Pressures (PPH2O) and
RSA) and the nonionic surfactant Berol 050 (pentaethylene Corresponding Simulated Conversions
glycol dodecyl ether, AkzoNobel), forming a stable reverse
micelle or microemulsion. Aqueous ammonium hydroxide PPH2O PPH2O/PPH2 Tsaturator XCO‑simulated Preactor
(bar) (bar/bar) (°C) (%) (bar)
solution (25 wt %, Sigma-Aldrich) was added dropwise to form
cobalt hydroxide, entrapped in the reverse micelle structure. 0 0 n/a n/a 1.0
The latter is destabilized via the addition of acetone. 0.5 0.75 82 63 1.5
Precipitated hydroxide is collected, washed in acetone, dried 1 1.5 100 77 2
in air at 120 °C, and subsequently calcined at 200 °C, yielding 2 3 120 87 3
3 4.5 134 91 4
nanocrystalline Co3O4. The desired amount of the cobalt oxide
is resuspended in water under ultrasonication, mixed with the
alumina support (Al2O3, Puralox SCCa 5-150 series, SBET = 162 amount of water fed, so that the water partial pressure was the
m2/g, Vpore = 0.47 cm3/g, dpore = 11.5 nm, dparticle = 150−200 only parameter varied. Each condition was maintained for 6 h
μm; Sasol, Germany), and dried under reduced pressure in a to allow the system to stabilize. It is important to note that the
rotary evaporator. Crystallite size variation was achieved via thermodynamic feasibility of the oxidation of small crystallites is
variation of the composition of the reverse micelle systems determined by the ratio of water to hydrogen and not total
used, the higher water to surfactant ratios and the higher cobalt pressure levels.13 The approach taken in this study, where the
salt concentrations yielding the larger crystallites (see also syngas pressure was kept constant and the ratio of water to
below). syngas or hydrogen (see Table 2), respectively, was varied
2.2. Catalyst Characterization and Testing. Trans- systematically, therefore allows us to study the effect of
mission electron microscopy (TEM, LEO 912 operated at conversion at a syngas pressure level of 1 bar. The reactor
120 kV) analysis of the unsupported Co3O4 crystallites was outlet gas was directly analyzed with an online GC-TCD
used to determine the crystallite size distributions obtained. (Varian CP4900). Additionally gas samples were taken with the
The targeted loading of cobalt (approximately 7 wt %) on the ampule method14 and analyzed in an offline FID (Varian
alumina support was confirmed by means of atomic absorption CP3900). No kinetic data were recorded at water partial
spectroscopy (AAS, Varian SpectraAA 110). pressures of 3 bar. Under these conditions only possible phase
The identification of the crystallite phases and corresponding changes were studied with the X-ray diffractometer.
sizes of the model catalysts as well as their activity and Both model catalyst were studied with TEM in the oxidic
selectivity under Fischer−Tropsch conditions was performed unsupported and in the supported stage before and after
using an in-house developed in situ XRD cell attached to a reduction (catalyst was cooled to room temperature and
conventional X-ray diffractometer (Bruker D8 Advance). passivated in a flow of 1% O2 in N2). After the experiments, the
Quantitative data regarding size and concentration were spent catalysts were again characterized with TEM.
obtained from Rietveld refinement using Topas 4.2. The in
situ XRD cell, which employs a capillary reactor (loaded with 3. RESULTS AND DISCUSSION
10.3 mg of catalyst A and 13.7 mg of catalyst B, respectively) The cobalt oxide catalyst precursor was characterized with
and a thermocouple placed inside the catalyst bed for accurate standard offline techniques such as TEM and AAS. The
temperature control, is described in detail elsewhere.9,11 Prior reduction and all following reaction steps were analyzed in situ
to the exposure of the model catalysts to Fischer−Tropsch using X-ray diffraction. Reactor outlet gases were identified with
114 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121
ACS Catalysis Research Article

Figure 1. Crystallite size distribution obtained from the analysis of more than 300 unsupported Co3O4 crystallites of catalysts A (left) and B (right)
in TEM micrographs. Size distributions are number based.

gas chromatographic techniques. After passivation, the spent catalyst systems with distinct sizes of metallic nanoparticles are
catalyst was further analyzed with TEM to confirm trends obtained. TEM analyses of the supported model catalyst show a
observed in the in situ XRD studies. slight agglomeration of the Co3O4 crystallites (see Figure 2,
3.1. Model Catalyst System. The results of XRD top). This is more pronounced in the case of catalyst A.
characterization confirmed the formation of a pure cobalt Possible reasons for this difference are the higher loading and
oxide spinel phase, Co3O4, in the unsupported product of the the larger average crystallite size (i.e., larger than the pore size
catalyst calcination. The crystallite sizes obtained via Rietveld of the support, 11.5 nm), rendering the Co3O4 crystallites more
refinement of the XRD data as well as from the analysis of TEM difficult to resuspend under ultrasonication. After reduction at
micrographs are in good agreement (see Figure 1 and Table 3). 450 °C in hydrogen (see section Catalyst Characterization and
Testing for experimental details), TEM micrographs of catalyst
Table 3. Characterization Results of the Oxidic Catalyst A show a severe increase in crystallite size (see Figure 2,
Precursor bottom) to an average of 21.3 ± 5.6 nm, whereas catalyst B
retains its average crystallite size (5.6 ± 1.3 nm).
Co3O4 crystallite size
3.2. Fischer−Tropsch Synthesis. 3.2.1. Phase Stability
Co(NO3)2·6H2O XRDb loading ASS and Fischer−Tropsch Activity. As previously reported9 catalyst
catalyst ωa concn (mol/L) (nm) TEMc (nm) (wt %)
A does not fully reduce under the applied conditions and a
A 0.4 0.32 11.1 10.3 ± 1.9 6.4 residual content of Co2+ oxide is still detectable, whereas with
B 0.2 0.08 6.7 6.1 ± 1.4 4.0 catalyst B full reduction was obtained (see Figure 3 for
a
Molar water to surfactant ratio. bCrystallite sizes (volume-based crystallite sizes obtained during reduction and reaction work).
average) determined from XRD analyses; calculated using Rietveld The only measurable phase of metallic cobalt was the face-
refinement. cCrystallite sizes (average and standard deviation) centered cubic (fcc) modification. Average crystallite sizes in
determined from TEM analyses; note that the sizes given are volume the reduction process from Co3O4 via CoO to fcc-Co should, in
based (recalculated from number-based distributions) to allow for
direct comparison with XRD data.
the absence of any secondary effects, decrease by approximately
25% due to the loss of oxygen.16 In the case of catalyst A, a
crystallite size increase from the oxide precursor to the metallic
Indeed, the variations in the reverse micelle composition result phase was measured (from 11.1 to 20.4 nm). Although the
in two distinctly different size fractions, namely, 10−11 nm for sizing of the latter phase is affected by a relatively large error
catalyst A and 6−7 nm for catalyst B. Both catalyst precursors resulting from overlapping peaks of the diffraction pattern of
display a fairly narrow particle size distribution (see Figure 1 the metallic phase and the support material, a crystallite growth
and standard deviations in Table 3), making them ideally process, that is, sintering, is evident. For catalyst B the expected
suitable for model catalyst studies aiming at unravelling size decrease in crystallite size (from 6.7 to 4 nm), clouded as to its
dependencies/effects. With AAS the final loading of the Co3O4 absolute extent by the mentioned uncertainties, is observed.
crystallites on the alumina support was determined as being 6.4 Possibly, small crystallites undergo a more intimate contact
and 4.0 wt % cobalt for catalysts A and B, respectively. The with the support, forming anchoring sites in pores of the
cobalt loading chosen in the present study is significantly higher support (average pore diameter = 11.3 nm) and hence reducing
than previously realized with the reported preparation the risk of crystallite growth. The larger crystallites of catalyst A,
method4a,12 to allow for better signal-to-noise ratio in the in on the other hand, might be too large for the pores of the
situ XRD measurements. As the prepared cobalt oxide spinel support and therefore more susceptible for sintering. A similar
nanocrystallites are dispersed on the alumina support by effect was previously observed by comparing different variations
ultrasonication, that is, in the absence of anchoring chemical of the reverse micelle approach to prepare alumina-supported
bonds which would lead to metal support interactions,15 a cobalt model catalysts. When larger particles are deposited
higher loading results in a decreased average distance between (diameter influenced through surfactant layer present around a
single crystallites, higher probability of cluster formation, and Co(OH)2 precipitate compared to calcined Co3O4), the effect
therefore higher probability of crystallite growth during a of sintering was more pronounced.15 Upon exposure to
thermal treatment such as reduction. These risks are accepted synthesis gas and upon the addition of elevated partial
in the context of this study as long as after reduction two model pressures of water, to simulate higher conversions, the
115 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121
ACS Catalysis Research Article

Figure 2. TEM micrographs of the supported catalyst A (left) and catalyst B (right) in the oxidic (top) and reduced (bottom) state. Identified areas
of crystallite agglomeration and single crystallites are indicated.

crystallite sizes and phases present in catalyst A do not change X-ray diffraction under the presented experimental conditions,
significantly. In contrast, catalyst B undergoes partial oxidation need to be present to explain the observed activity. Indeed, in
upon the addition of 0.5 bar PPH2O (70% of metallic cobalt the complementary study focusing on the oxidation potential of
oxidized to CoO) and full oxidation to Co2+ oxide at higher water,9 magnetic measurements of a very similar model catalyst
water contents. The observed size effect has been previously system showed a severe loss in magnetization due to the
reported by us in detail9 and is in agreement with theoretical oxidation of Co to CoO, however retaining approximately a
studies.13 Previous studies on crystallite size dependencies have fourth of the initial magnetization, that is, reduced cobalt
hypothesized that oxidation of cobalt nanocrystallites might species, after 5 h under a simulated CO conversion of 66.7%.
occur13,17 but that under industrially relevant conditions the The exact position and structure of the metallic cobalt fraction
water partial pressure is too low.5,17b Work by the Davis could not be determined.
group18 showed that cobalt on silica and alumina catalysts is Under “dry” conditions, the overall CO conversion is within
influenced by the addition of water. They concluded the 1.2% the same for both catalyst systems. Due to the differences
formation of cobalt oxide species upon the initial exposure to in cobalt loading (see Table 3) as well as catalyst loading in the
elevated water pressures, a species that upon prolonged capillary cell, this translates to a by nearly 2-fold higher cobalt
exposure can either undergo sintering and reduction or form mass specific reaction rate for catalyst B (see Figure 4, left).
cobalt aluminates. Interestingly, in the present study, whereas This observation is in line with previous studies reporting an
CoO is the only detectable X-ray visible phase, catalyst B still ideal crystallite size regarding the mass specific activity for
shows significant Fischer−Tropsch activity under all studied cobalt-based Fischer−Tropsch synthesis of 6−8 nm.4a,19 Upon
water partial pressures (see Figure 4). As CoO is not active for the addition of water vapor to the reactor inlet gas, the
the hydrogenation of carbon monoxide and subsequent chain Fischer−Tropsch rate increases by a factor of 4 for catalyst A
growth, very small crystallites, thin layers or amorphous while remaining relatively constant for catalyst B. The
metallic cobalt, or very small concentrations of larger difference in the observed effects of water on the mass specific
crystallites, all being below the sensitivity limit of laboratory activity can be correlated to the different changes in the
116 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121
ACS Catalysis Research Article

active material due to the oxidizing effects of co-fed water and

an activating effect of water on the remaining active sites as can
be observed for catalyst A. The extent of the latter cannot be
quantified in this study but is hypothesized to be crystallite size
dependent. Previously the addition of water to an alumina-
supported cobalt Fischer−Tropsch catalyst has been reported
to cause the catalyst to deactivate, which was correlated to the
oxidizing potential of water or a reversible competitive
adsorption on the catalytic surface.18a,b,d,20 In contrast, other
studies on silica and titania as well as alumina-supported cobalt
and ruthenium catalysts report an enhanced Fischer−Tropsch
activity upon the addition of water.6,18c The observed increase
in activity was explained by a facilitated CO dissociation step by
coadsorbed water.6a In addition, it was noted that an increase in
water partial pressure would result in a decrease of surface
carbon species, which have been considered to inhibit the
Fischer−Tropsch reaction.6d Schulz et al. further noted that
water could influence the elemental reaction steps affecting not
only activity but also selectivity.6e,f,7 Water has also been
described as a hydrogen shuttle, further lowering the
dissociation energy required for the hydrogen-assisted CO
dissociation, a process that is believed to play a role in relatively
unreactive surfaces such as terrace sites of cobalt crystallites21
or in ruthenium surfaces.22 For large cobalt crystallites, where
the terrace sites constitute a higher percentage of the surface
atoms, one would therefore expect a higher impact of co-fed
Figure 3. Crystallite sizes of Co3O4 (open squares), CoO (gray water than on small crystallites where edge and kink sites
circles), and fcc-Co (black triangles) for catalysts A (top) and B dominate. In the present study this trend could not be
(bottom) as a function of reaction conditions. Values determined by confirmed as the smaller crystallites show a decreased stability
Rietveld refinement in TOPAS 4.2. Reproduced with permission from against the oxidative pressure of water. Plotting the turnover
ref 9. Copyright 2014 Wiley-VCH. frequency (TOF), that is, the number of moles of carbon
monoxide converted per available surface cobalt atom and time
oxidation states of the model catalysts. Catalyst A retains its (under the assumption of ideal cubooctahedric cobalt particles,
predominantly metallic character, that is, phase composition, as as described in the literature,23 and no unavailability of cobalt
well as crystallite size independent of the elevated levels of surface area due to the interface with the support material),
water. The change in activity can therefore directly be shows 2- and 3.2-fold increases in activity for catalysts A and B,
correlated to the increased water partial pressures. In the case respectively (see Figure 4, right). Again, the proposed role of
of catalyst B, the metallic content is reduced upon the water as a hydrogen shuttle on terrace sites21 could therefore
introduction of 0.5 bar water vapor by 70% in favor of CoO. not be confirmed under the present conditions due to two
This severe loss in active phase content should result in a clear different superimposing effects. The lower TOF of catalyst B
loss in activity. The observed stable activity, seemingly can be explained by the smaller crystallite size, and this has
independent of water partial pressures, can therefore be been reported extensively in the literature for cobalt-based
identified as the sum of two opposing effects: a decrease in catalysts4a−c,f,24 and other FT active materials.4e,25

Figure 4. Rate of Fischer−Tropsch product formation per gram of cobalt (left) and turnover frequency (right) as a function of water partial pressure
for catalyst A (black squares) and catalyst B (gray squares). Note that due to the absence of an XRD visible metallic cobalt phase for catalyst B above
water partial pressures of 0.5 barn no turnover frequency could be calculated.

117 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121

ACS Catalysis Research Article

3.2.2. Chain Growth Probability and Product Selectivity. growth enhancement upon the addition of water vapor. This
Upon the addition of water to the Fischer−Tropsch reaction trend is so pronounced that under the effect of additional water
conditions, the chain growth probabilities (calculated from the partial pressure the smaller crystallites in catalyst B even
straight-chained hydrocarbon products with a carbon number outperform the larger crystallites with respect to low methane
from 3 to 6) for catalysts A and B increased 2-fold, with catalyst and high C5+ selectivity. This shows that the addition of a
B, that is, the smaller cobalt crystallites, displaying a larger second activity and selectivity steering effect can revert
enhancement (see Figure 5). This general trend has been previously identified intrinsic effects of the crystallite size in
the absence of water.4a−c,f,25a,26 Further studies, stepwise
including additional variables such as the presence of liquid
hydrocarbons, will be necessary to fully deconvolute con-
tributions of these various effects on activity and selectivity
observed under industrial conditions.
The two main classes of hydrocarbons formed in the
Fischer−Tropsch synthesis are n-paraffins and α-olefins.6e,28 In
the absence of re-adsorption of olefins and subsequent
hydrogenation or incorporation, a primary olefin selectivity of
70−90 mol % can be expected per carbon number.28d,f,29
Carbon number dependent differences in olefin content have
also been described. With ethylene being highly reactive and
rapidly undergoing secondary reactions, it is detected in the
Fischer−Tropsch product spectrum in lower quantities than
expected from an ideal Anderson−Schulz−Flory distribution.
Figure 5. Chain growth probability α as a function of water partial Propene is therefore often the olefin found in the highest
pressure for catalyst A (black squares) and catalyst B (gray squares). abundance per carbon number fraction as with increasing
carbon numbers the tendency to undergo secondary reaction
observed by other research groups6a,b,e,f,7,17a,18c and has been increases due to the changes in the carbon number dependent
attributed to effects of water inhibiting product desorption.7 diffusion rates,28e solubility,29b and/or physisorption,30 which
Interlinked with the chain growth probability are the result in longer contact times of the adsorbed olefin species on
selectivities to the different product fractions, such as methane the catalytic surface and increased reactor residence times.
and the C5+ fraction (see Figure 6). Under “dry” conditions, Catalyst A shows, regardless of the reaction conditions, a
both model catalysts show very similar selectivities to methane relatively stable and primary olefin content in the total linear
and higher hydrocarbons, irrespective of the cobalt crystallite hydrocarbon fraction (see Figure 7, left). In comparison, the
size. Although this seems to be in contrast to previously olefinic product with catalyst B clearly undergoes severe
reported increases in methane selectivity with decreasing secondary reactions at “dry” conditions (see Figure 7, right).
crystallite size,4a−c,f,25a,26 previous studies have shown that the Upon the addition of water vapor secondary olefin reactions are
extent of the size dependency of selectivities is a function of inhibited and the product spectrum turns into a more primary
reaction conditions; that is, at milder conditions and lower composition. In a very comparable study by us with focus on
conversions fewer differences in selectivity with crystallite size the crystallite size dependency in the size range from
can be expected.4a,d Upon the addition of water, chain growth, approximately 2 to 11 nm,4a where the same model catalysts
that is, the insertion of C1 species such as a methylene species27 were used, no differences in the olefin content were observed.
into a growing chain, is favored over the hydrogenation and In our previous study a higher synthesis gas pressure of 9.9 bar
desorption as methane.6b,c,e,f,7,18c As such, the selectivity to was applied and a 10-fold higher Fischer−Tropsch product
methane drops significantly with a simultaneous increase of the formation rate was observed, although also “dry” conditions
selectivity of the C5+ fraction (see Figure 6). Overall, the with low conversions were applied. Possibly the resulting higher
smaller crystallites in catalyst B show a higher degree of chain water partial pressure is already sufficient to suppress secondary

Figure 6. Methane (squares) and C5+ selectivities (triangles) as a function of water partial pressure for catalyst A (left) and catalyst B (right).

118 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121

ACS Catalysis Research Article

Figure 7. Fraction of olefin content in total linear hydrocarbon content for C2 (squares), C4 (triangles), and C6 (circles) fractions as a function of
water partial pressure for catalyst A (left) and catalyst B (right).

Figure 8. Fraction of α-olefins in the total fraction of linear olefinic product for the representative C7 fraction (left) and ratio of branched to linear
hydrocarbon contents for the representative C5 fraction (right) as a function of water partial pressure for catalyst A (black squares) and catalyst B
(gray squares).

olefin reactions. Alternatively, the higher CO partial pressures branched products. A primary pathway, based on the reaction
in our former study suppressed secondary olefin hydro- of a surface alkylidene and a surface methyl species, has also
genation.31 been proposed.35 It is often argued that steric hindrance on the
Besides the hydrogenation and incorporation into a growing densely populated catalyst surfaces result in the generally low
hydrocarbon chain, re-adsorbed α-olefins can undergo a yield of branched compounds. In the presented study a possible
double-bond shift reaction yielding internal olefins. This decrease in branching with the addition of water can be
reaction has been shown to occur on metal sites in the observed (see Figure 8, right). Due to the low overall
presence of hydrogen as well as on solid acidic sites.32 Whereas concentrations, the observed trend needs to be viewed with
the total olefin content in catalyst A is not affected by the water caution. Nevertheless, it indicates that indeed a secondary
addition, the fraction of internal olefins, here representatively mechanism, via the re-adsorption sites for α-olefins, to
shown by the content of α-heptene in the fraction of linear C7 branched hydrocarbons could be present. Interestingly, both
olefins, clearly increases by approximately 30% (see Figure 8,
catalysts are affected to the same extent.
left). Even more pronounced is the effect in the case of catalyst
3.2.3. Spent Catalyst. Although the catalysts were studied in
B, where an increase from 35 to >80 mol % was observed.
situ with X-ray diffractometry, the spent and subsequently
Again, smaller crystallites seem to facilitate significantly more
passivated catalysts were studied under TEM. Although it was
secondary reactions either due to their intrinsic higher surface
area or due to the different nature and composition of the not possible to distinguish between the different cobalt
surface (higher concentration of kink and edge sites). Only oxidation states present, observed crystallite sizes correlate
small amounts of water are sufficient to cloud/compensate this well with the calculated values from the XRD measurements. In
crystallite size effect. Water inhibits the described secondary addition, a subtle change of the support morphology was
reactions, which might occur not only on active sites relevant observed. In some sections of the observed spent catalysts,
for chain growth but also on non-FT sites responsible for needle-like configurations of potential alumina phases were
hydrogenation and isomerization reactions only.29b,33 observed (see Figure 9). These possibly indicate the formation
A last type of secondary α-olefin reactions involves the of an AlO(OH) phase Boehmite.36 The presence of this phase
formation of branched hydrocarbons. This mechanism could not be confirmed from the diffraction pattern, most likely
developed by Schulz and co-workers and evidenced via co- due to its low concentration. Effects on catalyst activity and/or
feeding experiments34 is not the sole proposed route to selectivity could also not be determined.
119 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121
ACS Catalysis

■
Research Article

REFERENCES
(1) Gallei, E. F.; Hesse, M.; Schwab, E. Development of Industrial
Catalysts. In Handbook of Heterogeneous Catalysis; Wiley-VCH Verlag:
Weinheim, Germany, 2008.
(2) Boudart, M.; McDonald, M. A. J. Phys. Chem. 1984, 88, 2185−
2195.
(3) Haruta, M. Catal. Today 1997, 36, 153−166.
(4) (a) Fischer, N.; van Steen, E.; Claeys, M. J. Catal. 2013, 299, 67−
80. (b) Bezemer, G. L.; Bitter, J. H.; Kuipers Herman, P. C. E.;
Oosterbeek, H.; Holewijn, J. E.; Xu, X.; Kapteijn, F.; van Dillen, A. J.;
de Jong, K. P. J. Am. Chem. Soc. 2006, 128, 3956−3964. (c) den
Breejen, J. P.; Radstake, P. B.; Bezemer, G. L.; Bitter, J. H.; Froseth, V.;
Holmen, A.; Jong, K. P. d. J. Am. Chem. Soc. 2009, 131, 7197−7203.
(d) Welker, C. A. Ruthenium based Fischer−Tropsch Synthesis on
Crystallites and Clusters of Different Sizes. Ph.D. thesis, University of
Cape Town, 2007. (e) Ojeda, M.; Rojas, S.; Boutonnet, M.; Pérez-
Alonso, F. J.; Javier García-García, F.; Fierro, J. L. G. Appl. Catal., A
2004, 274, 33−41. (f) Prieto, G.; Martínez, A.; Concepción, P.;
Moreno-Tost, R. J. Catal. 2009, 266, 129−144. (g) Ho, S. W.; Houalla,
M.; Hercules, D. M. J. Phys. Chem. 1990, 94, 6396−6399. (h) Reuel, R.
C.; Bartholomew, C. H. J. Catal. 1984, 85, 78−88.
(5) Bezemer, G. L.; Remans, T. J.; van, B. A. P.; Dugulan, A. I. J. Am.
Figure 9. Representative TEM micrograph of alumina support of Chem. Soc. 2010, 132, 8540−8541.
spent catalyst B with highlighted needle-like structure not observed (6) (a) Bertole, C. J.; Mims, C. A.; Kiss, G. J. Catal. 2002, 210, 84−
prior to exposure to FT conditions. 96. (b) Lögdberg, S.; Boutonnet, M.; Walmsley, J. C.; Järås, S.;
Holmen, A.; Blekkan, E. A. Appl. Catal., A 2011, 393, 109−121.
(c) Storsæter, S.; Borg, Ø.; Blekkan, E. A.; Holmen, A. J. Catal. 2005,
231, 405−419. (d) Claeys, M.; van Steen, E. Catal. Today 2002, 71,
419−427. (e) Schulz, H.; vein Steen, E.; Claeys, M. Stud. Surf. Sci.
4. CONCLUSION Catal. 1994, 81, 455−460 DOI: 10.1016/S0167-2991(08)63911-7.
Using two model catalyst systems with well-defined crystallite (f) van Steen, E.; Schulz, H. Appl. Catal., A 1999, 186, 309−320.
sizes, we could show that besides the previously reported (7) Schulz, H.; Claeys, M.; Harms, S. Stud. Surf. Sci. Catal. 1997, 107,
193−200 DOI: 10.1016/S0167-2991(97)80334-5.
tendency to undergo oxidation,9 the presence of water, that is, a (8) Dry, M. E. Catal. Today 2002, 71, 227−241.
simulated high conversion condition, has a severe crystallite size (9) Fischer, N.; Clapham, B.; Feltes, T.; van Steen, E.; Claeys, M.
dependent impact on the selectivity of cobalt-based catalytic Angew. Chem., Int. Ed. 2014, 53, 1342−1345.
systems in the Fischer−Tropsch synthesis. We could also shed (10) Yates, I. C.; Satterfield, C. N. Energy Fuels 1991, 5, 168−173.
light on the previously contradictory reports of the effect of (11) Claeys, M.; Fischer, N. U.S Patent 8,597,598, 2013.
water on the Fischer−Tropsch activity.6,20 An increased water (12) Fischer, N.; van Steen, E.; Claeys, M. Catal. Today 2011, 171,
partial pressure results in not only a significantly increased 174−179.
surface specific activity but also an increased chain growth and (13) van Steen, E.; Claeys, M.; Dry, M. E.; Van de Loosdrecht, J.;
lower methane and higher C5+ selectivity. Both effects are more Viljoen, E. L.; Visagie, J. L. J. Phys. Chem. B 2005, 109, 3575−3577.
pronounced for the smaller crystallites. The opposing effect of (14) Schulz, H.; Boehringer, W.; Kohl, C.; Rahman, N.; Will, A.
DGMK Forschungsbericht 1984, 3, 320−329.
the oxidation of smaller crystallites under the oxidative stress of (15) Fischer, N.; Minnermann, M.; Bäumer, M.; van Steen, E.;
water could result in an overall stagnation or even deactivation Claeys, M. Catal. Lett. 2012, 142, 830−837.
as previously reported.20 At high water partial pressures the (16) Nakhaei Pour, A.; Taheri, S. A.; Anahid, S.; Hatami, B.; Tavasoli,
selectivity to desired products (long-chained hydrocarbons at a A. J. Nat. Gas Sci. Eng. 2014, 18, 104−111.
low methane selectivity) over the smaller crystallites actually (17) (a) Iglesia, E. Appl. Catal., A 1997, 161, 59−78. (b) van de
surpasses those of larger ones, contradicting previous studies on Loosdrecht, J.; Balzhinimaev, B.; Dalmon, J. A.; Niemantsverdriet, J.
crystallite size dependency, most of which were carried out at W.; Tsybulya, S. V.; Saib, A. M.; van Berge, P. J.; Visagie, J. L. Catal.
rather water lean conditions.4a−c,f,25a,26 Secondary olefin Today 2007, 123, 293−302. (c) Swart, J. C. W. A Theoretical View on
reactions, the hydrogenation to paraffins as well as the Deactivation of Cobalt-Based Fischer−Tropsch Catalysts. Ph.D. thesis,
double-bond isomerization resulting in the formation of University of Cape Town, 2008.
(18) (a) Li, J.; Zhan, X.; Zhang, Y.; Jacobs, G.; Das, T.; Davis, B. H.
internal olefins, are more pronounced under dry conditions Appl. Catal., A 2002, 228, 203−212. (b) Jacobs, G.; Patterson, P. M.;
in the case of smaller crystallites. This class of reactions might Das, T. K.; Luo, M.; Davis, B. H. Appl. Catal., A 2004, 270, 65−76.
be supported by active sites independent of the Fischer− (c) Li, J.; Jacobs, G.; Das, T.; Zhang, Y.; Davis, B. Appl. Catal., A 2002,
Tropsch chain growth ensemble, which seem to be more 236, 67−76. (d) Jacobs, G.; Das, T. K.; Patterson, P. M.; Li, J.;
prevalent on the small crystallites. However, only a small Sanchez, L.; Davis, B. H. Appl. Catal., A 2003, 247, 335−343.
amount of water vapor seems to be sufficient to nearly (e) Jermwongratanachai, T.; Jacobs, G.; Shafer, W.; Ma, W.; Pendyala,
selectively suppress their activity. V.; Davis, B.; Kitiyanan, B.; Khalid, S.; Cronauer, D.; Kropf, A. J.;

■ AUTHOR INFORMATION
Corresponding Author
Marshall, C. Top. Catal. 2014, 57, 479−490.
(19) Bezemer, G. L.; Radstake, P. B.; Koot, V.; van Dillen, A. J.; Geus,
J. W.; de Jong, K. P. J. Catal. 2006, 237, 291−302.
(20) Hilmen, A. M.; Schanke, D.; Hanssen, K. F.; Holmen, A. Appl.
*(M.C.) E-mail: [email protected]. Catal., A 1999, 186, 169−188.
Notes (21) (a) van Santen, R. A.; Ghouri, M.; Hensen, E. M. J. Phys. Chem.
The authors declare no competing financial interest. Chem. Phys. 2014, 16, 10041−10058. (b) van Santen, R. A.;

120 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121

ACS Catalysis Research Article

Markvoort, A. J.; Filot, I. A. W.; Ghouri, M. M.; Hensen, E. J. M. Phys.

Chem. Chem. Phys. 2013, 15, 17038−17063. (c) van Santen, R. A.;
Markvoort, A. J.; Ghouri, M. M.; Hilbers, P. A. J.; Hensen, E. J. M. J.
Phys. Chem. C 2013, 117, 4488−4504.
(22) Hibbitts, D. D.; Loveless, B. T.; Neurock, M.; Iglesia, E. Angew.
Chem., Int. Ed. 2013, 52, 12273−12278.
(23) Van Hardeveld, R.; Hartog, F. Surf. Sci. 1969, 15, 189−230.
(24) (a) Fischer, N.; van Steen, E.; Claeys, M. ChemCatChem. 2014,
6, 1707−1713. (b) Bezemer, L.Cobalt supported on carbon nanofibres as
catalysts for the Fischer−Tropsch synthesis. Ph.D. thesis, University
Ultrecht, 2006.
(25) (a) Cheang, V. Effect of Crystallite Size and Water Partial Pressure
on thr Activity and Selectivity of Low Temperature Iron-based Fischer−
Tropsch Catalysts. Ph.D. thesis, University of Cape Town, 2009.
(b) Mabaso, E. I.; van Steen, E.; Claeys, M. DGMK/SCI-Conference
“Synthesis Gas Chemistry”; DGMK-Tagungsbericht, 2006; pp 93−100
(c) Carballo, J. M. G.; Yang, J.; Holmen, A.; García-Rodríguez, S.;
Rojas, S.; Ojeda, M.; Fierro, J. L. G. J. Catal. 2011, 284, 102−108.
(d) Welker, C.; Moss, J. R.; van Steen, E.; Claeys, M. DGMK/SCI-
Conference “Synthesis Gas Chemistry”; DGMK-Tagungsbericht, 2006;
pp 223−230 (e) Kang, J.; Deng, W.; Zhang, Q.; Wang, Y. J. Energy
Chem. 2013, 22, 321−328.
(26) Mabaso, E. I. Nanosized Iron Crystallites for Fischer−Tropsch
Synthesis. Ph.D. thesis, University of Cape Town, 2005.
(27) Claeys, M.; van Steen, E. Surf. Sci. Catal. 2004, 152, 601−680
DOI: 10.1016/S0167-2991(04)80465-8.
(28) (a) Iglesia, E.; Reyes, S. C.; Madon, R. J. J. Catal. 1991, 129,
238−256. (b) Madon, R. J.; Reyes, S. C.; Iglesia, E. J. Phys. Chem.
1991, 95, 7795−7804. (c) Gaube, J.; Klein, H. F. J. Mol. Catal. A 2008,
283, 60−68. (d) Gökcebay, H.; Schulz, H. EP Patent 0,071,770, 1983.
(e) Schulz, H. Top. Catal 2003, 26, 73−85. (f) Schulz, H.; Roesch, S.;
Gökcebay, H. Selectivity of the Fischer−Tropsch Synthesis; Verlag
Glueckauf: Essen, Germany, 1981; pp 522−529.
(29) (a) Schulz, H.; Beck, K.; Erich, E. The Fischer-Tropsch carbon
monoxide-hydrogenation, a nontrivial surface polymerization selectiv-
ity of chain branching. In Proceedings of the 9th International Congress
on Catalysis, Calgary; Chemical Institute of Canada: Ottawa, Canada,
1988; Vol. 2. (b) Schulz, H.; Claeys, M. Appl. Catal., A 1999, 186, 71−
90.
(30) Kuipers, E. W.; Scheper, C.; Wilson, J. H.; Vinkenburg, I. H.;
Oosterbeek, H. J. Catal. 1996, 158, 288−300.
(31) Schulz, H.; van, S. E.; Claeys, M. Top. Catal. 1995, 2, 223−34.
(32) Germain, J. E.Catalytic Conversion of Hydrocarbons; Academic:
New York, 1969; pp 322.
(33) Schulz, H.; Claeys, M. Appl. Catal., A 1999, 186, 91−107.
(34) (a) Schulz, H.; Beck, K.; Erich, E. Chem. Inst. Can. 1988, 829−
836. (b) Schulz, H.; Erich, E.; Gorre, H.; Van, S. E. Catal. Lett. 1990, 7,
157−167. (c) Schulz, H.; Rao, B. R.; Elstner, M. Erdoel und Kohle
1970, 23, 651−655.
(35) Lee, C. B.; Anderson, R. B.Chain Grown in the Fischer−Tropsch
Synthesis; Verlag Chemie: Weinheim, Germany, 1985; pp II15−II22.
(36) Tsakoumis, N. E.; Rønning, M.; Borg, Ø.; Rytter, E.; Holmen, A.
Catal. Today 2010, 154, 162−182.

121 dx.doi.org/10.1021/cs500936t | ACS Catal. 2015, 5, 113−121