Catalysis Today 228 (2014) 65–76

Contents lists available at ScienceDirect

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

Molecular structure and localization of carbon species in alumina

supported cobalt Fischer–Tropsch catalysts in a slurry reactor
D. Peña a , A. Griboval-Constant a , C. Lancelot a , M. Quijada b , N. Visez c , O. Stéphan d ,
V. Lecocq e , F. Diehl e , A.Y. Khodakov a,∗
a
Unité de Catalyse et de Chimie du Solide (UCCS), UMR 8181 CNRS, Université Lille 1-USTL-ENSCL-EC Lille, Bât. C3, Cité scientifique, 59655 Villeneuve
d’Ascq, France
b
Laboratoire GEOSYSTEMES, UMR 8217, Université Lille 1, Bât. SN5, Cité scientifique, 59655 Villeneuve d’Ascq, France
c
Laboratoire de Physicochimie des Processus de Combustion et de l’Atmosphère, Université de Lille 1, Bât. C11, Cité scientifique, 59655 Villeneuve d’Ascq,
France
d
Laboratoire de Physique des Solides, Université Paris-Sud, Orsay Cedex 91405, France
e
IFP Energies nouvelles, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France

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

Article history: The paper focusses on the identification of molecular structure of carbon species in 15 wt.% Co/Al2 O3 sup-
Received 11 July 2013 ported catalyst exposed to different conditions of Fischer–Tropsch synthesis in a continuously stirred tank
Received in revised form reactor (CSTR). Characterization of wax-extracted spent catalyst by a combination of techniques showed
29 September 2013
significant carbon deposition under the conditions where the catalyst exhibited a noticeable deactivation.
Accepted 1 October 2013
Carbon deposition was less important when catalyst activity was relatively stable. TPH-MS suggested the
Available online 8 November 2013
presence of different carbon species on surface. GC–MS analysis indicated that the strongly adsorbed car-
bon deposited on wax-extracted spent catalyst was mainly constituted of ␣-olefins, n-paraffins, branched
Keywords:
Fischer–Tropsch synthesis
alkanes/alkanes, 2-methyl-branched fatty aldehydes and 2-ketones. 2-Methyl-branched fatty carboxylic
Slurry reactor acids were mostly detected at high H2 O pressure in the reactor. Branched alkenes/alkenes (2-methyl-
Cobalt catalyst alkenes/alkanes) were favored at high CO conversion. Alpha-alkyl cinnamic aldehydes were detected
Deactivation under hydrogen deficient conditions. STEM-EELS analysis suggests that carbon is located both on cobalt
Oxygenates particles and on alumina support, the carbon species situated on the support being more reluctant to
Olefins hydrogenation.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction a slow longer-term decline [7] suggesting that deactivation could

be a combination of several simultaneous phenomena [4].
The Fischer–Tropsch synthesis (FT) is in the core of Gas-to- Efficient control of catalyst deactivation requires identification
Liquid (GTL), Biomass-to-Liquid (BTL) and Coal-to-Liquid (CTL) of the main reasons of deactivation and suggesting efficient solu-
technologies, which involve conversion of syngas (CO, H2 ) pro- tions to improve catalyst stability and catalyst life-time. Several
duced from natural and shale gas, biomass and coal into ultra-clean recent studies [8,9] suggest that cobalt sintering and deposition
synthetic fuels and chemical feedstocks using iron and cobalt of carbonaceous species on the catalyst surface could contribute
catalysts [1,2]. Alumina supported cobalt catalysts are generally to catalyst deactivation. Cobalt sintering favored by the ther-
preferred for FT synthesis in slurry reactor due to their high activity, modynamics [4], is usually most pronounced during the first
mechanical and chemical stability, high selectivity to linear paraf- hours of reaction hereafter, its levels off. It can be reversed
fins, lower water gas shift activity and oxygenate selectivity [2–5]. by reduction–oxidation–reduction (ROR) [6]. Carbon deposition
Nevertheless, catalytic deactivation is still a crucial drawback for appears to be the main reason of catalyst deactivation in a longer
the industrial use of this technology. The catalyst activity typically term due to the slow accumulation of deleterious carbon species
decreases with time-on-stream affecting carbon monoxide con- on active catalyst surface [4,6]. It seems that sintering can be
version and hydrocarbon productivity [4,6]. Usually, the activity affected by oxidation of small cobalt nanoparticles by water pro-
shows a fast drop during the first days of the reaction followed by duced during the reaction. A quantitative mechanism of cobalt
sintering which involves cobalt surface oxidation has been pro-
posed in our recent paper [10,11]. Cobalt sintering could also lead
∗ Corresponding author. Tel.: +33 3 20 33 54 39; fax: +33 3 20 43 65 61. to microscopic cobalt particles clusters of around 3–5 ␮m, which
E-mail address: andrei.khodakov@univ-lille1.fr (A.Y. Khodakov). were detected using scanning electron microscopy in the catalysts

0920-5861/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cattod.2013.10.005
66 D. Peña et al. / Catalysis Today 228 (2014) 65–76

in slurry reactor under FT reaction conditions favoring deactiva- K-Ru-Co/␥-Al2 O3 catalyst during FT synthesis in a fixed bed reactor
tion [8]. High exothermicity is a major problem of FT synthesis in at 250 ◦ C, 1 bar and a H2 /CO ratio of 2 increased along the catalytic
fixed bed reactors. Formation of microscopic cobalt metal parti- bed. Our previous work using TPH-MS technique was also indica-
cles was also observed in fixed bed reactor when the reactor was tive of the presence of different types of carbon species in the spent
exposed to temperatures much higher than those usually used for catalysts after slurry FT reactor [23]. Polymeric carbon and strongly
low temperature FT synthesis [10,11]. adsorbed hydrocarbons seem to contribute to catalyst deactivation
In addition to cobalt sintering, carbon deposition can produce at lower gas space velocities and lower H2 /CO syngas ratio. Lee
several negative effects on the structure and catalytic performance et al. [15,16] suggested a relation between carbon deposition and
of supported metallic catalysts. First, carbon deposits may block olefin selectivity. A shift in the selectivity of paraffins to olefins was
the catalyst pores resulting in diffusion limitations. Second, car- observed with increasing carbon deposition. Recently, Park et al.
bon can poison the metal surface by binding irreversibly or even [24] suggested that an increase in olefins/paraffins (O/P) ratio of
encapsulating metal particles [12]. Subsurface carbon may also hydrocarbons products in cobalt-based catalyst during FT synthesis
play a role in electronic inhibition of activity [13]. It has also using a slurry reactor could lead to high selectivity to polymeric car-
been shown that carbon bound to a metal surface induces sur- bon. A correlation was observed between an increase of O/P ratio,
face reconstruction, which affects the catalyst performance [13]. In polymeric carbon formation and catalyst deactivation.
extreme cases, strong carbon filaments may build up in the pores Despite a significant number of works indicating carbon depo-
to extent that they could stress and fracture the support, causing sition in FT catalysts under the reaction conditions, much less
disintegration of catalyst pellets [12]. Blocking of catalytic sites by information is available about molecular structures of carbon
chemisorbed hydrocarbons and thin films of relatively reactive car- species particularly about the structure of polymeric carbon. The
bon species could be generally reversible, while formation of bulk term “polymeric carbon” in the case of FT catalysts generally refers
coke could irreversibly affect the catalyst. The above-mentioned to the chains of carbon monomers connected by covalent bonds
deleterious effects depend on the amount and type of deposited [13]. Polymeric carbon has been identified as a less reactive carbon
carbon [12]. which might form from polymerization of CHx and alkyl groups
Menon [14] has suggested that FT synthesis is a carbon insen- [25]. Therefore this term could involve a large number of molecular
sitive reaction, considering that the amount of hydrogen present structures.
during FT synthesis is sufficient for hydrogenation of adsorbed The present work focuses on the identification of the molecular
carbon species. Bartholomew [12] indicated that deactivation rate structures of carbon species formed and adsorbed onto the surface
by carbon deposition is a function of the rates of formation and of a cobalt supported alumina-based catalyst during FT reaction in a
gasification of carbon/coke precursors, i.e. rd = rf − rg . If the rate of continuously stirred tank reactor (CSTR) under different operating
gasification rg is equal to or greater than that of carbon formation conditions. An attempt has been made to identify a possible role of
rf carbon/coke is not deposited and the deactivation is limited. these species in catalyst deactivation.
Lee et al. [15,16] investigated the effect of carbon deposi-
tion on alumina-supported cobalt catalyst under FT conditions.
2. Experimental
Temperature-programmed experiments indicate that higher reac-
tion temperature (250–350 ◦ C) leads to a more significant amount
2.1. Catalyst preparation
of carbon deposition. Carbon deposition increases with the reaction
time. The atomic surface carbon transformed morphologically to
A 15 wt.% Co/Al2 O3 catalyst was prepared via incipient wet-
polymeric and graphite carbon. Moodley et al. [17] have examined
ness impregnation of alumina based commercial ␥-alumina with
the role of carbon deposition in the deactivation of cobalt catalysts
an aqueous cobalt nitrate solution. Two impregnation steps were
exposed to the commercially relevant conditions of FT synthesis
necessary to obtain 15 wt.% of cobalt loading. Between each step the
in a 100-barrel/day-slurry bubble column reactor. Catalyst deacti-
catalytic precursor was dried in an oven under air at 120 ◦ C for 2 h
vation was attributed to deposition of polymeric carbon on cobalt
and calcined at 420 ◦ C under a airflow (1000 cm3 h−1 g−1 ) for 2 h.
particles. In particular, a slow accumulation of polymeric carbon
Finally, the catalyst was reduced at 400 ◦ C under a pure hydrogen
species during the extended FT run was observed on the catalyst.
flow (2000 cm3 h−1 g−1 ) at 1 bar for 16 h. The freshly reduced cata-
The polymeric (non-graphitic) carbon was resistant to hydrogen
lyst (∼3–8 g) was transferred under a helium atmosphere to small
treatments at temperatures well above those used under realistic
vials filled with hot liquid wax (Sasolwax C80® ). The wax protected
conditions of FT synthesis. Recently, the presence of polynuclear
the catalyst from oxidation in air. The first portion of the reduced
aromatic compounds such as graphene layers and surface carbide
catalyst was collected in a vessel without wax, in order to keep a
phases which could contribute to catalyst deactivation was sug-
sample in a passivated form for further characterization.
gested by Tan et al. [18,19] in alumina supported catalysts on the
basis of the results of TPH-MS, X-ray photoelectron spectroscopy
(XPS), high resolution transmission electron microscopy (HRTEM) 2.2. Catalytic tests
and density functional theory (DFT). Four kinds of carbonaceous
compounds in Ru-promoted alumina supported cobalt catalysts The performance of the Co/Al2 O3 catalyst was evaluated in a
were observed by Pinard et al. [20]: atomic carbon (cobalt carbide), 100 cm3 continuously stirred tank reactor (CSTR) at fixed pressure
alcohols, carboxylic acids and polymeric carbon. Carboxylic acids of 20 bar and a stirring rate of 1000 rpm using Durasyn 170® as
and polymeric carbon which were resistant to hydrogen treatment a start-up solvent. Initially, the reactor was loaded with reduced
could be related to catalyst deactivation. catalyst embedded in Sasolwax C80® and then around 70 mL of
It has been also reported that the formation of different car- Durasyn 170® was added to the reactor. Then the reactor temper-
bon species greatly depends on operating conditions, especially ature increased to 100 ◦ C. The reactor pressure was increased in
temperature and chemical composition (e.g. H2 /CO ratio) [12]. It argon (Ar, 99%) to around 25 bar overnight with a stirring rate of
seems that hydrogen deficient conditions may lead to the forma- 300 rpm. After checking the reactor system for leaks, the pressure
tion of more stable carbon species. It is known that H2 /CO ratio and was set to 20 bar and the temperature was further increased using a
amount of heavier hydrocarbons would vary along the catalyst bed ramp of 20 ◦ C h−1 either to 220 or 230 ◦ C. Once the reaction temper-
radially or in the pellets [21]. Tehrani et al. [22] reported that the ature had been reached, the syngas was introduced to the reactor.
amount of heavy unsaturated hydrocarbons or coke deposited on At the same time the stirring rate was changed to 1000 rpm. The
 D. Peña et al. / Catalysis Today 228 (2014) 65–76 67

Table 1
Catalytic performance of the 15 wt.% Co/Al2 O3 catalyst under different operating conditions (after 350 h on stream).

Operating conditions Test Selectivity (%) on carbon basis O/P (C10 –27 ) ␣ (C15 –60 ) FT rate
(mol CO g−1 h−1 ) × 10−2

CO2 CH4 C2 –C4 C5+ C-OH

3 −1 −1
GHSV = 7000 cm g h Standard – 9.8 8.4 79.9 1.9 0.032 0.887 3.48
H2 /CO = 2, T = 230 ◦ C
GHSV = 7000 cm3 g−1 h−1 High H2 Pressure – 14.1 10.2 73.7 2 0.029 0.869 2.43
H2 /CO = 3, T = 220 ◦ C
GHSV = 1900–800 cm3 g−1 h−1 High CO Conversion 0.2 11.0 10.3 78.0 0.4 0.042 0.875 0.69
H2 /CO = 2, T = 220 ◦ C
GHSV = 1900–1300 cm3 g−1 h−1 High CO Pressure – 7.9 11.2 79.4 1.5 0.090 0.905 0.53
H2 /CO = 1, T = 220 ◦ C

syngas was diluted with Ar at the beginning of reaction. After reach- of the Soxhlet extractor. The Soxhlet extractor was placed onto a
ing the reaction temperature, Ar and syngas were introduced to the flask containing around 100 mL of toluene, which was heated at
reaction system in concentration of 75% and 25% respectively for 150–160 ◦ C to reflux, and the boiling vapor travels up a distilla-
1 h. Then, the Ar fraction was decreased to 25%, while syngas frac- tion arm, and floods into the chamber housing the thimble. The
tion was increased to 75% for 5 h. Finally, the reactor was fed with chamber was slowly filled with toluene and the remaining wax
pure syngas and its chemical composition kept constant until the was then dissolved. When the chamber was almost full, it was
end of reaction. This moment was considered as the beginning of automatically emptied by a siphon side arm, with toluene running
reaction. The catalytic tests lasted for a minimum of 350 h. A sim- back down to the distillation flask. After several cycles of Soxhlet
ilar procedure was previously used by our group for the fixed bed extraction, the catalyst grains did not show any visible traces of
reactor [26]. wax and were ready for ex situ characterization.
The catalytic performance and catalyst deactivation were stud-
ied under two types of the operating conditions: at constant 2.4. Ex situ catalyst characterization
gas space velocity (at iso-GHSV) and at constant conversion (at
iso-conversion). During iso-GHSV experiments, carbon monoxide The Brunauer–Emmett–Teller (BET) area, pore diameter and
conversion gradually decreased with the reaction time. During iso- pore volume were measured from nitrogen adsorption isotherms
conversion experiments, the conversion was adjusted to a desired after sample outgassing at 150 ◦ C during 2 h using a Micromeritics
value on daily basis by decreasing the gas space velocity. The exper- ASAP 2010 apparatus.
imental conditions of the catalytic tests are specified in Table 1. Differential thermal analysis (DTA) and thermogravimetric
The FT reaction products were analyzed both on-line and off- analysis (TGA) were conducted using a TA Instrument SDT 2960.
line by gas chromatography (GC). The carbon monoxide contained DTA–TGA analyses were performed in pure nitrogen at temper-
5% of nitrogen, which was used as an internal standard for cal- atures ranging from room temperature to 1000 ◦ C in the aim to
culating CO conversion. The periodical analysis of unreacted N2 determine the weight loss related to desorption of carbon species.
and CO, and produced CH4 and CO2 were performed using a The nitrogen flow rate (100 mL min−1 ), sample loading (about
packed CTR-1 column and a thermal conductivity detector (TCD) 20 mg) and a heating rate (10 ◦ C min−1 ) were kept constant in all
in order to determine CO conversion and CO2 and CH4 selec- experiments.
tivities. Light hydrocarbons (C1 –C7 ) were analyzed on-line using In order to differentiate the reactivity of each carbon species
a capillary CP-PoraPLOT Q column by a flame-ionization detec- toward hydrogen, temperature programmed hydrogenation (TPH)
tor (FID). High-molecular-weight products (C7 + hydrocarbons) and coupled with mass spectroscopy (MS) was carried out using an
oxygenate products were collected in two traps thermostated at AutoChem II 2920 V3 0.2 Micromeritics. About 80 mg of catalyst
150 ◦ C and 10 ◦ C respectively. The liquid hydrocarbons products after Soxhlet extraction was used for each TPH analysis using a tem-
were weighted, dissolved in carbon disulfide (CS2 ) and ana- perature ramping of 4 ◦ C min−1 from room temperature to 1000 ◦ C.
lyzed by a Varian CP-3800 GC using a CP-SimDist column. The The methane production using m/z = 15 signal and production of
Anderson–Schulz–Flory chain growth probability (˛) was calcu- Cx Hy hydrocarbons using m/z = 28 were monitored using a Balzers
lated in C15 to C60 carbon range. The O/P ratio was calculated from Omnistar mass spectrometer. The gas flow of pure hydrogen was
C10 to C30 hydrocarbons. Alcohols products were analyzed using a 50 mL min−1 .
Varian 3900 GC equipped with a ZB-WAX-ZB-1 capillary column.
The reaction rates (expressed in mol h−1 g−1 ) were defined as 2.4.1. Characterization of strongly adsorbed carbon species
the number of moles of CO converted per hour per gram of cata- After the removal of paraffinic wax by Soxhlet procedure,
lyst. The product selectivity (S) was reported as the percentage of more strongly adsorbed carbon species in the used catalysts
CO converted into a given product calculated on carbon basis. The were extracted using a solution of dichloromethane–methanol
carbon mass balances in the range of 95–100% were calculated at (CH2 Cl2 :CH3 OH). CH2 Cl2 dissolves a wide range of organic com-
every step of trap collection, including products analyzed on-line pounds and is widely employed as organic solvent. Similarly,
and off-line. methanol is another laboratory solvent with affinity to organic
molecules especially to alcohols and carboxyl acids. The extracted
2.3. Wax extraction procedure organic compounds were then analyzed by GC–MS.
Initially, the sample was dissolved in a solution of 2:1
After reaction, the catalyst covered with wax was discharged CH2 Cl2 :CH3 OH employing ultrasound during 2 h. After that, the
from the reactor. The wax from the catalyst was removed using solid was decanted and the solution with organic compounds was
Soxhlet extraction. The procedure is described in detail in Refs. filtered using a column, filled in the bottom by glass wool and in the
[8,9]. The catalyst with the layer of wax was placed inside a top by chromatography sand. The filtered solution of around 2 mL
thimble of cellulose, which was loaded into the main chamber was recovered in a vial. Then, the filtered solution was evaporated
68 D. Peña et al. / Catalysis Today 228 (2014) 65–76

R GHSV
CO conversion R GHSV
100 0,038
70
95
R (mol CO. g.cata . h )
-1 -1 0,054 7600 7600
65 90 CO conversion 0,036
60 0,052 85 -1 -1
3 -1 -1
GHSV cm g .h 7400 80 R (mol CO. g.cata . h ) 0,034
55 7400
0,050 75

CO conversion (%)
3 -1 -1
50 70 GHSV cm .g .h
0,048
0,032
65
CO conversion (%)

45 7200
7200
60
0,046 0,030
40 55
35 0,044 7000 50
0,028 7000
45
30 0,042 40
6800 0,026
25 35
0,040 30 6800
20 0,024
25
15 0,038 6600
20
15 0,022 6600
10 0,036
0.036
6400 10
5 0,020
0,034 5
0 0 6400
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Time on stream (h) Time on stream (h)

Fig. 1. Evolution of CO conversion and reaction rate with time on stream (TOS) at iso-GHSV at (a) Standard conditions and (b) High H2 Pressure conditions (Table 1).

under N2 at atmospheric pressure. Afterwards, the organic com- 10% and 80% respectively. Very low selectivity to alcohols (around
pounds extracted were recovered with around 100 ␮L of CH2 Cl2 . 2%) was observed, in agreement with previous results [29].
One microliter of this solution was injected in the GC–MS using a C2 –C4 selectivity was also stable, not exceeding 10%. However,
DB-5 MS non-polar column. higher syngas ratio i.e. H2 /CO = 3 (High H2 Pressure in Table 1)
The organic compounds present in the extracted solution were and constant GHSV of 7000 cm3 g−1 h−1 conditions led to rather
also analyzed using FTIR. The liquid sample was placed in a CaF2 different catalytic performance (Fig. 1b). A continuous decrease in
cell. The FTIR spectra were recorded at room temperature using CO conversion was observed, achieving a final CO conversion of
a Thermo Nicolet 460 spectrometer equipped whit a DTGS (CsI) around 35% after 350 h of the reaction. As expected, higher H2 /CO
and processed using the Nicolet OMINCTM software at 64 averaged ratio led to a higher CH4 selectivity (near to 14%) and a drop of C5+
scans at a 2 cm−1 of resolution. selectivity to around 75% (Table 1). C2 –C4 and alcohol selectivities
Scanning transmission electron microscopy coupled with elec- showed similar values compared to those obtained under Standard
tron energy loss spectroscopy (STEM-EELS) experiments were conditions. Interestingly, hydrocarbon selectivity kept almost
performed using a 100 keV STEM microscope (VG HB 501) spec- unchanged with time on stream. Zhou et al. reported similar result
trometer equipped with a cold field emission source (CFEG) and a [30] suggesting also catalyst deactivation at higher H2 /CO ratio and
Gatan 666 parallel-EELS spectrometer. EEL spectra were recorded relative stability of hydrocarbons selectivity with time on stream.
with a charge-coupled device (CCD) camera optically coupled to At High H2 Pressure conditions, the O/P ratio was also low. At this
a scintillator in the image plane of a Gatan magnetic sector. The condition, ␣-olefins are preferentially hydrogenated due to higher
high brightness CFEG source allows the formation of small probes H2 partial pressure.
(∼0.5 nm) and an energy resolution close to 0.5 eV. Both carbon and Higher catalytic conversion of about 65–70% at H2 /CO = 2 (High
cobalt maps in the catalyst were obtained by acquiring series of CO Conversion conditions, Table 1) was maintained by progressively
spectra when scanning the subnanometer probe size at the surface decreasing gas space velocity (Fig. 2a). Despite the decrease in reac-
of a region of interest. At each pixel of the scan (pixel size is about tion rate, the hydrocarbon distribution did not change greatly with
0.5 nm × 0.5 nm) a full EELS spectrum is acquired [27]. The whole time on stream. Some increase in O/P ratio was observed (Table 1). A
set of spectra is then processed to extract the chemical maps. slight decrease in alcohol selectivity (∼0.4%) and higher CO2 selec-
tivity were observed (∼0.2%). These results are in agreement with
3. Results and discussion previous reports. Indeed, low alcohols selectivity has been reported
at CO conversion around 50% for cobalt-based catalysts [31]. It is
3.1. Catalyst performance widely known that an increase in CO conversion resulted in a higher
H2 O partial pressure that could be favorable for the water-gas-shift
The evolution of catalytic performance with time on stream reaction resulting in CO2 production. Note that CO2 formation was
under Standard conditions (H2 /CO = 2, temperature = 230 ◦ C, not detected in the other tests (Table 1).
GHSV = 7000 cm3 g−1 h−1 ) is shown in Fig. 1a. A slow decrease in At the conditions of High CO Pressure
CO conversion and reaction rate was observed during the first (GHSV = 1900–1300 cm3 g−1 h−1 , H2 /CO = 1, temperature = 220 ◦ C,
200 h on stream. After that, CO conversion stabilized at around 40%. Table 1) CO conversion showed a moderate decrease with time
The turnover frequency (TOF) was calculated using the character- on stream (Fig. 2b). CH4 selectivity was slightly increasing after
ization data for the freshly reduced catalyst (cobalt particle size the first 150 h of reaction; nevertheless it did not exceed 10%
of 7.5 nm and extent of reduction 42% [8]). It was compared with (Fig. 3). Additionally, we noted a decrease in C5+ selectivity from
the literature data. The TOF was 0.085 s−1 with our catalyst which 85% to 80% approximately from 200 h on stream; the alcohols
is similar with the TOF of 0.09–0.05 s−1 calculated by Gnanamani and C2 –C4 selectivities were less affected (Fig. 3). As expected
[28] for silica supported cobalt catalyst with larger metal particles the chain growth probability (˛) and the O/P ratio increased with
tested in slurry reactor (T = 220 ◦ C, GHSV = 3000 cm3 h−1 g−1 cat ,
the decrease in H2 /CO ratio. Similar results have been reported
P = 19.9 bar, H2 /CO = 2). The CH4 and C5+ selectivities were about previously in the literature [30].
 D. Peña et al. / Catalysis Today 228 (2014) 65–76 69

R GHSV R GHSV
CO conversion 0,017 0,013
65
95 -1 -1
2000
90 R (mol CO. g.cata . h ) 0,016 60 2000
1900 0,012
85
GHSV cm g .h
3 -1 -1 0,015 55
80 1800 CO conversion 1900
0,011
75 0,014 50
1700
70
CO conversion (%)

-1 -1
1600 45 R (mol CO. g.cata . h ) 0,010 1800
65 0,013

CO conversion (%)
60 1500 40 3 -1 -1
55 0,012 GHSV cm g .h 0,009 1700
1400 35
50
0,011
45 1300 30 0,008
1600
40 0,010
1200 25
35 0,007
30 0,009 1100 20 1500
25 0,006
0,008 1000 15
20 1400
15 900 10
0,007 0,005
10
800 5 1300
5 0,006
0,004
0 700 0
0 50 100 150 200 250 300 350 400 450 500 550 600 650 0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750
Time on stream (h) Time on stream (h)

Fig. 2. Evolution of CO conversion and reaction rate at iso-conversion at High CO Conversion conditions (a) and at High CO Pressure conditions (b) (Table 1).

3.2. Catalyst texture, DTA–TGA and TPH-MS analyses of Sasolwax® , which was not removed after wax extraction. The
weight loss in this case is about 5%. This result indicates that after
The textural properties such as BET area, pore volume and pore Soxhlet extraction a minimum of 5% of hydrocarbon species remain
diameter before and after reaction are summarized in Table 2. in the catalyst. Above 200 ◦ C the weight loss of the catalyst after
Under Standard conditions, the textural properties were not very catalytic test under Standard conditions was near to 9.5% while for
different from those of the freshly reduced catalyst after wax the other conditions the weight losses were even more significant
extraction, suggesting only very slight plugging of catalyst pore (about 13%, Table 2). The observed weight losses are consistent with
probably by residual wax. On the contrary, BET area and total pore the decrease in pore volume. This suggests that the carbon species
volume under the other catalytic conditions (High H2 Pressure, High are principally localized in catalyst pores.
CO Conversion, High CO Pressure) showed an important decrease. The TPH-MS experiments were carried out in order to study the
The decrease in BET surface area and pore volume after FT tests reactivity toward hydrogen of the carbon species remaining onto
suggests carbon deposition in the catalyst pores [24,32–35]. The catalyst surface after wax extraction. The production of methane
decrease in accessible catalyst surface area is in line with the drop was observed using m/z = 15 signal and production of hydrocarbons
of catalytic activity observed under these conditions. (Cx Hy ) was detected using m/z = 28. The methane and Cx Hy profiles
The catalysts weight loss data after the catalytic tests are also for the catalyst before and after reaction are shown in Fig. 4a and b
shown in Table 2. For the fresh passivated and freshly reduced respectively. They exhibit the presence of several peaks which cor-
catalysts after extraction of Sasolwax, the observed initial weight respond to different carbon species. The methane and Cx Hy profiles
loss corresponds to the removal of chemisorbed water from the suggest that the peak around 250 ◦ C corresponds to desorption of
catalyst [23] while, the weight loss above 200 ◦ C in the freshly hydrocarbons, while high temperature peaks detected in CH4 pro-
reduced catalyst might be attributed to decomposition/desorption duction profiles correspond to hydrogenation of carbon species. A
decrease in CH4 and Cx Hy production rates was observed when
the catalyst showed less significant decline in catalytic activity
90 (Standard conditions). The Gaussian deconvolution of methane pro-
85
duction profile using the approach described in our earlier work
80
[23] confirms the presence of three types of chemisorbed hydrocar-
75
70
bon/carbon species onto catalyst surface (Table 3). The Cx Hy profiles
65 of the spent catalysts (Fig. 4b) exhibit a weak shoulder at higher
Hydrocarbons Selectivity (%)

60 temperatures (∼300 ◦ C). The profile is probably composed of two

55 C5+ peaks, the first one being centered near to 250 ◦ C and another less
50 C2-C4 intense one situated at around 300 ◦ C. According to our previous
45 R-OH report [8] the most intense methane production signal registered
40 CH4 near to 250 ◦ C corresponds to hydrogenolysis/desorption of hydro-
35
carbons (probably wax) adsorbed on catalyst surface, while the
30
25 shoulder at 300 ◦ C corresponds to the desorption of hydrocarbons
20 which are adsorbed more strongly or remain inside the narrow-
15 est catalyst pores. This attribution is consistent with the literature.
10 Gruver et al. [36] had reported that FT wax is hydrogenated at
5 around 250 ◦ C. Similarly, Moodley et al. [17] suggested that the
0 hydrogenolysis of remaining wax occurs at 250 ◦ C. Also the metha-
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700
nation of surface carbide species was reported to occur also at a
Time on stream (h) similar temperature (around 250 ◦ C) [17]. However, cobalt carbide
species are rarely detected by any ex situ characterization tech-
Fig. 3. Evolution of hydrocarbons selectivity with TOS at iso-conversion at High CO
Pressure conditions. nique [17,23].
70 D. Peña et al. / Catalysis Today 228 (2014) 65–76

Table 2
Textural properties and weight loss of the freshly reduced catalyst in passivated and wax-extracted form and spent catalyst under different operating conditions.

Spent catalyst BET area (m2 /g) Average pore Total pore volume Weight loss %
diameter (Å) (cm3 /g support) 200–900 ◦ C

Fresh passivated catalyst 121 86 0.35 –

Freshly reduced catalyst (after extraction of Sasol wax) 103 80 0.27 4.5
Standard 102 73 0.28 9.5
High H2 Pressure 85 74 0.25 13.0
High CO Conversion 80 75 0.24 12.5
High CO Pressure 81 76 0.23 13.5

Fig. 4. TPH-MS profiles of the spent catalyst (a) CH4 production rate (m/z = 15) and (b) Cx Hy production rate over spent catalysts exposed to FT reaction under different
conditions.

In agreement with previous reports [23], the most resilient (Table 1) was initially treated in hydrogen at 280 ◦ C to remove the
carbon species (near 400 ◦ C) are attributed to polymeric (not more reactive carbon species, leaving strongly adsorbed hydrocar-
graphene/graphite) carbon adsorbed on metallic surface or catalyst bons and polymeric species. Another sample of the same catalyst
surface. This confirms again that graphene/graphitic species do not was submitted to hydrogen at 380 ◦ C, taking-off strongly adsorbed
form over the catalyst even under H2 deficiency the conditions [8]. hydrocarbons, leaving only polymeric carbon species.
The hydrogenolysis of polymeric carbon on alumina-supported cat- Fig. 5 displays cobalt and carbon maps for both samples. The
alyst has been reported to occur above 400 ◦ C [14,17,20]. Polymeric sample treated at 280 ◦ C showed higher concentration of carbon
carbon seems to be related to the decline of catalytic activity with (carbon EELS map Fig. 5c). Carbon however was dispersed hetero-
time on stream. When polymeric carbon is deposited on the cata- geneously over the catalyst. Interestingly, carbon deposition was
lyst, it may cause physical blocking active sites. Strongly adsorbed observed on both cobalt particles and support. This suggests that
hydrocarbons on catalyst pores could be also related to catalyst treatment in hydrogen at 280 ◦ C is not sufficient to remove carbon
deactivation under specific reaction conditions. deposition from the metal particles and some of cobalt sites can-
not be available for the reaction. As expected, the sample treated in
3.3. Localization of carbon species in the catalyst by STEM-EELS hydrogen at 380 ◦ C showed smaller amount of carbon (carbon EELS
map Fig. 5f). In this case carbon was mostly located on the support.
STEM-EELS analysis was performed in order to establish local- In agreement with Moodley et al. [17], carbon deposition seems
ization of carbon species on the surface of wax-extracted spent to occur on cobalt nanoparticles with further migration of carbon
catalysts. To determine the location of carbon species based on their species to the support. The results of the present work suggest that
reactivity toward hydrogen, two samples were prepared. The spent carbon located on the support is more reluctant to hydrogenation
catalyst exposed to FT synthesis test at High CO Pressure conditions than that on cobalt particles.

Table 3
Relative fraction of different carbon species obtained from deconvolution of the TPH profile of the wax-extracted 15 wt.%. Co/Al2 O3 spent catalysts under the different
conditions.

Carbon species TPH peak Fraction of carbon species

temperature (◦ C)

Standard Chi2 / High H2 p. Chi2 / High CO Conv. Chi2 / High CO p. Chi2 /

DOF = 1.24753 × 10−20 DOF = 1.12578 × 10−19 DOF = 7.2961 × 10−20 DOF = 1.00339 × 10−19

Hydrocarbons (surface) 250–275 0.21 0.24 0.28 0.31

Strongly adsorbed 300–330 0.59 0.40 0.49 0.36
hydrocarbons (pores)
Polymeric carbon (not 400–440 0.20 0.36 0.23 0.33
graphene/graphite)

Relative concentrations obtained by normalization of each peak area.

Chi2 = Chi-square value, DOF = degree of freedom.
 D. Peña et al. / Catalysis Today 228 (2014) 65–76 71

Fig. 5. STEM-EELS images: (a) HAADF of spent catalyst at High CO Pressure conditions after TPH at 280 ◦ C, (b) cobalt map, (c) carbon map, (d) HAADF of spent catalyst at High CO
Pressure after TPH at 380 ◦ C, (e) cobalt map, (f) carbon map. The areas with the highest concentration of cobalt and carbon are shown with brightest intensity (yellow-orange)
in each map. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

3.4. Identification of strongly adsorbed carbon species in spent extracted solution. Cobalt FT catalysts produce mostly paraffins
catalysts by GC–MS especially in higher hydrocarbon range. Higher concentration of
olefins than paraffins on the catalyst surface could be attributed to
TPH-MS analysis indicated the presence of three carbon a stronger adsorption of olefins.
species in the spent FT catalyst which differ in their reactiv- In order to calculate ␣-olefins distribution we measured the
ity toward hydrogen: (i) hydrocarbons (probably on the surface), m/z = 55 ion pattern for all the samples (Fig. 7). Linear ␣-olefins
(ii) strongly adsorbed hydrocarbons (probably in small pores) showed a distribution from C6 to C33 , however some differences
and (iii) polymeric (not graphene/graphite) carbon. However, TPH were detected depending on the operating conditions. A decrease in
and STEM-EELS could not provide detailed information about the the relative abundance of ␣-olefins between C6 –C17 was observed
molecular structures and possible pathways of carbon deposition. when FT reaction was carried out at High H2 Pressure condition. This
These strongly adsorbed carbon species were extracted in the suggests that light olefins can be readily hydrogenated to paraffins
spent catalyst using an organic solvent mixture (CH2 Cl2 :CH3 OH in the presence of higher hydrogen concentrations.
2:1). The structure of the extracted compounds was then analyzed Additionally, 2-methyl fatty aldehydes and methyl ketones
by GC–MS. Fig. 6 displays a typical total ion current (TIC) mass were detected in the organic solutions extracted from the spent
fragmentogram of the organic extract showing a wide range of catalyst (Fig. 8) exposed to syngas under different operating
organic compounds such as linear ␣-olefins in high relative abun- conditions. The m/z = 58 ion mass fragment is characteristic of
dance (m/z = 55 ion typical of n-alkene [37]) followed by linear the McLafferty rearrangement of methyl ketones [38] and also for
n-paraffin (m/z = 57 ion fragment [37,38]) and branched alkanes. 2-methyl fatty aldehydes [39]. The 2-methyl fatty aldehydes follow
␣-Olefins were dominant chemical compounds present in the a distribution from C10 to C31 with a maximum between C16 –C20 .

Fig. 6. CG–MS total ion current (TIC) mass fragmentogram of the organic extract obtained for the spent catalyst at High CO Pressure conditions (GHSV = 1900–1300 cm3 g−1 h−1 ,
H2 /CO = 1, T = 220 ◦ C), *C18:1 = 1-octadecene.
72 D. Peña et al. / Catalysis Today 228 (2014) 65–76

Higher relative abundance of aldehydes was detected at High CO

Conversion conditions. These findings suggest hydroformylation
and isomerization of linear ␣-olefins especially at high CO con-
version, which is in agreement with the results reported by Bukur
et al. [40]. The m/z = 58 fragmentograms obtained under High CO
Pressure conditions exhibited low intense of peaks attributed to
methyl ketones and fatty aldehydes respect to alkanes/alkenes
which also present a small m/z = 58 ion mass fragment [37], which
is indicative of low concentration of these compounds.
The hydroformylation of ␣-olefins was discovered in the late
30s during an investigation of oxygenated products occurring in
cobalt catalyzed Fischer–Tropsch reaction [41]. Most aldehydes
are hydrogenated to alcohols under FT conditions and alcohols
were dominant oxygenated products of FT synthesis [41]. However,
heavy 2-methyl-branched fatty aldehydes seem to remain strongly
adsorbed on spent catalyst surface without any further hydrogena-
tion to alcohols especially, at High CO Conversion conditions.
Pinard et al. [20] have detected carboxylic acids on spent cobalt-
based FT catalyst surface. In the present work branched fatty acids
from C7 to C28 were detected either at High H2 Pressure or at High
CO Conversion conditions. Carboxylic acids were not detected under
Standard Conditions, while they were present in very low concen-
trations at High CO Pressure conditions. The distribution of these
acids identified using m/z = 74 signal shows a maximum at C14 and
C11 correspondingly (Fig. 9). Previously, it has been reported that
aldehydes can be oxidized or hydrated to the corresponding car-
boxylic acids in the presence of water [42]. Our results suggest
that formation of 2-methyl-branched fatty acids from the corre-
sponding 2-methyl fatty aldehydes can be promoted at higher H2 O
Fig. 7. GC–MS m/z = 55 ion fragmentogram from the spent catalysts exposed to FT
partial pressure i.e. at High H2 Pressure conditions and at High CO
reaction under (a) Standard conditions, (b) High H2 Pressure conditions, (c) High CO
Conversion, (d) High CO Pressure. Conversion [43,44].
Note that alcohols cannot be clearly identified using GC–MS
analysis. It is known that alcohols form hydrogen-bonded
R O H· · ·O H R complexes which make their analysis difficult
by GS–MS [45].
Surprisingly, a number of aromatic compounds identified as
alpha-alkyl cinnamic aldehydes according to open data base [46]
were mainly uncovered in the organic extraction of the spent
catalyst under High CO pressure conditions. Fig. 10 displays the
m/z = 91 + 129 ion fragmentogram, which was measured for the
extracted solution. Alpha-alkyl cinnamic aldehydes follow a dis-
tribution from C12 to C32 with a maximum centered on C22 . To
confirm attribution of MS patterns, we performed analysis of alpha-
hexyl cinnamic aldehyde purchased from Sigma–Aldrich (≥95% of
purity). The fragmentation patterns of a commercial alpha-hexyl
cinnamic aldehyde and cinnamic aldehyde extracted from the cat-
alyst were identical (not shown). This confirms our suggestion
about formation of cinnamic aldehyde during FT synthesis with
low H2 /CO ratio.
Interestingly, 2-methyl branched fatty acids, and alpha-alkyl
cinnamic aldehydes did not appear under Standard conditions (low-
est decline in catalytic activity). Alpha-alkyl cinnamic aldehydes
(very low relative abundance) and heavier 2-methyl branched fatty
acids were present on spent catalyst surface at High H2 Pressure con-
ditions, while 2-methyl branched fatty acids were detected at High
CO Conversion conditions. Finally, alpha-alkyl cinnamic aldehydes
and 2-methyl branched fatty acids (very low relative abundance)
were detected at High CO Pressure conditions.

3.5. Detection of carbon species in the extracted solutions by

Fourier transform infrared (FTIR) spectroscopy

The organic solution extracted from the spent catalysts using

CH2 Cl2 :CH3 OH (2:1) solutions were further analyzed using FT-IR
Fig. 8. GC–MS m/z = 58 ion fragmentogram from organic extract from the spent cat-
alyst (a) Standard conditions, (b) High H2 Pressure conditions, (c) High CO Conversion.
spectroscopy to corroborate the results obtained by GC–MS. The
assessment of these peaks is given in Table 4. The FTIR spectra
 D. Peña et al. / Catalysis Today 228 (2014) 65–76 73

Fig. 9. GC–MS m/z = 74 ion fragmentogram from organic extract: (a) High H2 Pressure, (b) High CO Conversion. A MS pattern of 2-methyl C14-fatty acid is shown in the bottom.

of the organic extraction at Standard, High H2 Pressure and High associated to n-alkenes. The peak at 2980 cm−1 is due to asymmet-
CO Conversion conditions are shown in Fig. 11a. The observed ric CH3 stretching vibration, while the peaks at 2916 cm−1 are due
most intense broad band at 3000–3600 cm−1 was assigned to the to symmetric and asymmetric stretching vibration of primarily
hydrogen-bonded hydroxyl groups. These groups could be linked CH2 and possibly terminal CH3 groups. The band at 2836 cm−1 was
to alcohols or water molecules present in the reaction products. attributed to symmetric vibration of CH2 group [37]. The weak
The band at 3057 cm−1 was attributed to C H stretching vibration band appearing at 1707 cm−1 indicates the presence of C O groups
[47] probably from aldehydes and ketones. The band at 1648 cm−1
may originate from COO− stretching vibration in the carboxylic
Table 4 acids [48] confirming GC–MS result. Note that this band is absent
Assignment of the observed IR frequencies.
for the solution extracted from the catalyst exposed to syngas
Frequencies (cm−1 ) Assignments under Standard conditions, which is in agreement with GC–MS
3626, 3350 O H free and polymeric (water, alcohols) analysis of this sample. Indeed, no aldehydes, ketones or carboxylic
3054, 3057 C H (alkenes, aromatic) acids were detected in this spent catalyst. The band at 1453 cm−1
2988, 2980 CH3 corresponds to CH2 /CH3 asymmetric bending vibrations [37],
2974, 2957 CH2
having probably a contribution of C H scissoring from n-alkenes.
2928, 2916 CH2 primary (alkenes)
2874, 2854, 2836 CH2 (alkenes, alkanes)
A peak at 1423 cm−1 corresponds to C H bending vibration of
1707, 1704 C O (aldehyde, ketones) aliphatic chains [52] but it could be present a contribution of
1679 C O (cinnamic aldehyde) C O H bending [48]. Finally at 1271 cm−1 could be assigned to
1648 COO− (carboxylic acid) the vibrations of the C OH bond [49].
1601, 1552, 1494 C C (aromatic)
Unexpectedly, the FT-IR spectrum of the organic extract
1455 CH2 /CH3
1437 C C (aromatic) obtained for the spent catalyst under High CO Pressure is different
1423 C H or C O H from that of other samples (Fig. 11b). Hydroxyl groups were not
1377 C H detected. FTIR spectra suggest the presence of aromatic compounds
74 D. Peña et al. / Catalysis Today 228 (2014) 65–76

Fig. 10. GC–MS m/z = 74 ion fragmentogram from organic extract: (a) High H2 Pressure, (b) High CO Conversion. A MS pattern of 2-methyl C14-fatty acid is shown in the
bottom.

Fig. 11. (A) FT-IR of the organic extract for the spent catalysts exposed to FT reaction under (a) Standard conditions, (b) High H2 Pressure conditions, (c) High CO Conversion,
(B) FT-IR of the organic extract for the test at High CO Pressure conditions.
 D. Peña et al. / Catalysis Today 228 (2014) 65–76 75

Fig. 12. Possible pathways for formation of carbon species in FT catalysts in slurry reactor.

which is also consistent with GC–MS analysis. A triplet of weak presence of hydrogen, olefins can be hydrogenated to paraffins
bands at 1494, 1552 and 1601 cm−1 is associated to C C stretch- (principal products of low temperature FT synthesis). They can
ing vibration of aromatic rings [50]. The bands between 2800 and undergo isomerization [53] to produce 2-methyl-alkenes/alkanes
3000 cm−1 attributed to C H stretching in alkyl groups were rather and 4-methyl alkanes (alkenes/alkanes also detected) which were
intense for this catalyst. The peak at 3054 cm−1 may be related mostly detected at High CO Pressure conditions. The formation of
to C H stretching vibration from aromatic rings. Similarly, a peak 2-methyl-branched fatty aldehydes and ketones proceeds by the
at 1704 cm−1 was detected which corresponds to CO probably in hydroformylation of ␣-olefins. In the presence of hydrogen the
aldehydes and ketones. A very slight peak detected at 1679 cm−1 aldehydes and ketones can be hydrogenated to alcohols.
could be related to C O stretching vibration from cinnamic alde- The fatty aldehydes are considered as precursors of 2-methyl
hyde in agreement with GC–MS results. The sample also showed a fatty acids which form from these compounds via oxidation under
poorly resolved peak at 1455 cm−1 , which corresponds to CH2 /CH3 high H2 O pressure (High H2 Pressure and High CO Conversion condi-
asymmetric bending. Another contribution at 1437 cm−1 could be tions). Similarly, the formation of alpha-alkyl cinnamic aldehydes
attributed to C C aromatic ring vibration or also to alkyl-H vibra- may occur through the aldol condensation [52] involving eno-
tion [50,51]. Similarly, the sample showed a well-defined band at late formation from n-aldehydes, which could be stabilized by
1423 cm−1 of bending vibration of aliphatic groups with probably the support [53] and aromatic aldehydes (benzaldehyde). Aro-
some contributions from C O H bending. Finally, we found a peak matic aldehydes could probably form via dehydrocyclization of
at 1377 cm−1 (C H methyl rock), which corresponds to terminal ␣-olefins and then formylation of benzene [54] especially under
CH3 groups of alkyl chains [37]. H2 deficiency conditions. Another possible route of benzaldehyde
Fig. 12 displays a tentative mechanism of the formation of dif- formation (not shown) could include the initial formation of ben-
ferent species in alumina supported cobalt FT catalyst during the zene from ethylene and consequently formylation. Nevertheless,
reaction. The spent catalyst after Soxhlet extraction contains a further mechanistic studies are probably required to elucidate the
significant concentration of ␣-olefins, which are much strongly exact formation mechanism of benzaldehyde. Note that benzalde-
adsorbed on the catalyst than n-paraffins. It seems that these hyde did not been detected in Sasolwax and Durasyn 170® which
␣-olefins could be precursors of different carbon species. In the were used for the slurry reactor startup.
76 D. Peña et al. / Catalysis Today 228 (2014) 65–76

4. Conclusion [16] S-K. Ihm, D.-K. Lee, Stud. Surf. Sci. Catal. 68 (1991) 219–226.
[17] D.J. Moodley, J. van de Loosdrecht, A.M. Saib, M.J. Overett, A.K. Datye, J.W.
Niemantsverdriet, Appl. Catal. A 354 (2009) 102–110.
The catalytic performance, deactivation rate, type and concen- [18] K.F. Tan, J. Xu, J. Chang, A. Borgna, M. Saeys, J. Catal. 274 (2010) 121–129.
tration of deposited carbon on a conventional alumina-supported [19] K.F. Tan, J. Chang, A. Borgna, M. Saeys, J. Catal. 280 (2011) 50–59.
cobalt catalyst depend on the operating conditions of FT synthesis [20] L. Pinard, P. Bichon, A. Popov, J.L. Lemberton, C. Canaff, F. Mauge, P. Bazin, E.
Falabella S.-Aguiar, P. Magnoux, Appl. Catal. A 406 (2011) 73–80.
in a slurry type reactor. The presence of higher concentration of [21] J.A. Lapszewicz, H.J. Loeh, J.R. Chipperfield, J. Chem. Soc. Chem. Commun. 11
carbon species in the catalyst leads to more significant deactiva- (1993) 913–914.
tion. All the spent catalysts showed the presence of linear ␣-olefins [22] S. Tehrani, M. Irani, A. Tavasoli, Y. Mortazavi, A.A. Khodadadi, A.N. Pour, J. Nat.
Gas Chem. 20 (2011) 65–71.
with higher relative abundance, followed by n-paraffins, branched
[23] D. Peña, A. Griboval-Constant, V. Lecocq, F. Diehl, A.Y. Khodakov, Catal. Today
alkenes/alkanes, 2-methyl fatty aldehydes, 2-ketones and alpha- 215 (2013) 43–51.
alkyl cinnamic aldehydes. Higher CO conversion leads to higher [24] S.-J. Park, J.-W. Bae, Y.-J. Lee, K.-S. Ha, K.-W. Jun, P. Karandikar, Catal. Commun.
12 (2011) 539–543.
isomerization of linear ␣-olefins favoring formation of branched
[25] T.M. Duncan, P. Winslow, A.T. Bell, J. Catal. 93 (1985) 1–22.
alkenes/alkanes. The cinnamic aldehydes were detected under H2 [26] A. Jean-Marie, University of Science and Technologies of Lille, France, 2010
deficient conditions. Carbon species are probably nucleated on the (Ph.D. Thesis).
cobalt particles and then migrate to alumina support. The carbon [27] C. Jeanguillaume, C. Colliex, Ultramicroscopy 28 (1989) 252–257.
[28] M.K. Gnanamani, G. Jacobs, W.D. Shafer, B.H. Davis, Catal. Today 215 (2013)
species localized on the support were found to be more reluctant 13–17.
to hydrogenation. [29] V.A. de la Peña O’Shea, M.C. Alvarez-Galván, J.M. Campos-Martín, J.L.G. Fierro,
Appl. Catal. A: Gen. 326 (2007) 65–73.
[30] W. Zhou, J.-C. Chen, K.-G. Fang, Y.-H. Sun, Fuel Process. Technol. 87 (2006)
Acknowledgments 609–616.
[31] W. Ma, G. Jacobs, R.A. Keogh, D.B. Bukur, B.H. Davis, Appl. Catal. A: Gen. 437–438
D.P. thanks the Fundayacucho foundation for the Ph.D. stipend. (2012) 1–9.
[32] M.K. Niemelä, A.O.I. Krause, Catal. Lett. 42 (1996) 161–166.
The authors are indebted to O. Gardoll for his assistance with char- [33] J. Zhang, J. Chen, Y. Li, Y. Sun, J. Nat. Gas Chem. 11 (2002) 99–108.
acterization techniques and to IFP Energies Nouvelles for financial [34] S.A. Hosseini, A. Taeb, F. Feyzi, F. Yaripour, Catal. Commun. 6 (2005) 233–240.
support of this work. Dr. F. Agbossou-Niedercorn is acknowl- [35] G.W. Huber, C.G. Guymon, T.L. Conrad, B.C. Stephenson, C.H. Bartholomew,
Stud. Surf. Sci. Cat. 139 (2001) 423–430.
edged for her expertise in the mechanisms of organic reactions. [36] V. Gruver, R. Young, J. Engman, H.J. Robota, Prep. Pap. Am. Chem. Soc. Div. Petrol.
The authors acknowledge financial support from the CNRS-CEA Chem. 50 (2005) 164.
“METSA” microscopy French network for the STEM-EELS experi- [37] K. Arouri, P.F. Greenwood, M.R. Walter, Org. Geochem. 30 (1999) 1323–
1337.
ments.
[38] H.E. Bakouri, A. Aassiri, J. Morillo, J. Usero, M. Khaddor, A. Ouassini, Appl.
Geochem. 23 (2008) 3487–3497.
References [39] http://webbook.nist.gov/cgi/cbook.cgi?ID=C110418&Units=SI&Mask=200
#Mass-Spec
[40] D.B. Bukur, Z. Pan, W. Ma, G. Jacobs, B.H. Davis, Catal. Lett. 142 (2012)
[1] S. Fedou, E. Caprani, D. Douziech, S. Boucher, Petrol. Technol. Quart. 13 (2008)
1382–1387.
87–91.
[41] http://chem-faculty.lsu.edu/stanley/webpub/4571-chapt16-
[2] A.Y. Khodakov, W. Chu, P. Fongarland, Chem. Rev. 107 (2007) 1692–1744.
hydroformylation.pdf
[3] A.P. Steynberg, M.E. Dry, Fischer–Tropsch Technology, Stud. Surf. Sci. Catal. 152
[42] M. Lim, C.M. Yoon, G. An, H. Rhee, Tetrahedron Lett. 48 (2007) 3835–3839.
(2004) 406–481.
[43] G. Huffman, N. Shah, J. Zhao, F.E. Huggins, T.E. Hoost, S. Halvorsen, J.G. Goodwin
[4] N.E. Tsakoumis, M. Rønning, Ø. Borg, E. Rytter, A. Holmen, Catal. Today 154
Jr., J. Catal. 151 (1995) 17–25.
(2010) 162–182.
[44] M.K. Denk, Inorganic Chemistry II (Organometallic), CHM 331 University of
[5] A.Y. Khodakov, Braz. J. Phys. 39 (1A) (2009) 171–175.
Guelph, Canada, 1999.
[6] A.M. Saib, D.J. Moodley, I.M. Ciobîcă, M.M. Hauman, B.H. Sigwebela, C.J. West-
[45] http://cnx.org/content/m31896/latest/?collection=col11124/1.8
strate, J.W. Niemantsverdriet, J. van de Loosdrecht, Catal. Today 154 (2010)
[46] http://webbook.nist.gov/cgi/cbook.cgi?ID=C101860&Units=SI&Mask=200
271–282.
#Mass-Spec
[7] P.J. van Berge, R.C. Everson, Stud. Surf. Sci. Catal. 107 (1997) 207–212.
[47] A.P. Deshmukh, C. Pacheco, M.B. Hay, S.C.B. Mynemi, Geochim. Cosmochim.
[8] D. Peña, A. Griboval-Constant, F. Diehl, V. Lecocq, A.Y. Khodakov, ChemCatChem
Acta 71 (2007) 3533–3544.
5 (2013) 728–731.
[48] F.Z. Cui, S.Q. Sun, D.M. Zhang, Z.L. Ma, G.Q. Chen, Nucl. Inst. Met. Phys. Res. B
[9] H. Karaca, University of Lille 1, 2010 (PhD Thesis).
169 (2000) 48–53.
[10] M. Sadeqzadeh, J. Hong, P. Fongarland, D. Curulla-Ferré, F. Luck, J. Bousquet, D.
[49] M.B. Hay, S.C.B. Mynemi, Geochim. Cosmochim. Acta 71 (2007) 3518–3532.
Schweich, A.Y. Khodakov, Ind. Eng. Chem. Res. 51 (2012) 11955–11964.
[50] R.M. Silverstein, G.C. Bassler, T.C. Morill, Spectrometric Identification of Organic
[11] M. Sadeqzadeh, S. Chambrey, S. Piché, P. Fongarland, F. Luck, D. Curulla-Ferré,
Compounds, John Widely & Sons, Inc., USA, 1991, pp. 101–132.
D. Schweich, J. Bousquet, A.Y. Khodakov, Catal. Today 215 (2013) 52–59.
[51] I. Warad, H. Al-Ussain, R. Al-Far, R. Mahfouz, B. Hammouti, T.B. Hadda, Spec.
[12] C.H. Bartholomew, Appl. Catal. A 212 (2001) 17–60.
Act. A: Mol. Biochem. Spec. 95 (2012) 374–381.
[13] I.M. Ciobîcă, R.A. van Santen, P.J. van Berge, J. van de Loosdrecht, Surf. Sci. 602
[52] M. Vashishtha, M. Mishraa, D.O. Shah, Appl. Catal. A: Gen. 466 (2013) 38–44.
(2008) 17–27.
[53] T. Yokoyama, N. Yamagata, Appl. Catal. A: Gen. 221 (2001) 227–239.
[14] P.G. Menon, J. Mol. Catal. 59 (1990) 207–220.
[54] A.S. Singh, B.M. Bhanage, J.M. Nagarkar, Tetrahedron Lett. 52 (2011) 2383–
[15] D.-K. Lee, J.-H. Lee, S.-K. Ihm, Appl. Catal. 36 (1988) 199–207.
2386.