In situ X-ray absorption of Co/Mn/TiO2 catalysts for fischer-tropsch synthesis

Article abstract of DOI:10.1021/jp0403846

The reduction behavior of Co/TiO₂ and Co/Mn/TiO₂ catalysts for Fischer-Tropsch synthesis has been investigated by soft X-ray absorption spectroscopy (XAS). In situ XAS measurements of the L_2,3 edges of Co and Mn have been carried out during reduction treatments of the samples in H₂ at a pressure of 2 mbar and at temperatures up to 425?°C. The changes of Co and Mn 3d valences and the symmetries throughout the reduction have been determined by comparison with theoretical calculations based on the charge transfer multiplet code. Furthermore, bulk Co ₃O₄ has been reduced under the same conditions to evaluate the effect of TiO₂ as a support on the reducibility of Co oxides. The average Co valence at the various temperatures has been determined from a linear combination of the reference spectra. It was found that the unsupported Co₃O₄ was easily reduced to Co⁰ at 425?°C, whereas the Co₃O₄ supported on TiO₂ catalysts was only reduced to a mixture of CoO and Co⁰, even after 12 h reduction at 425?°C. The presence of Mn further retards the reduction of the supported Co₃O₄ particles. The Mn^III ions were easily reduced to MnO at temperatures lower than 300 ?°C, and they remained in this oxidation state even after further temperature increase. In addition, catalytic tests in the Fischer-Tropsch synthesis reaction at a pressure of 1 bar indicate that the selectivity of these catalysts might be related to the extent of Co reduced after the activation treatment (i.e., the reduction with H₂).

Full text of DOI:10.1021/jp0403846

J. Phys. Chem. B 2004, 108, 16201-16207
16201
In Situ X-ray Absorption of Co/Mn/TiO2 Catalysts for Fischer-Tropsch Synthesis
†
†
†
‡
Fernando Morales, Frank M. F. de Groot, Pieter Glatzel, Evgueni Kleimenov,
Hendrik Bluhm,‡ Michael H a1 vecker, Axel Knop-Gericke, and Bert M. Weckhuysen*
Department of Inorganic Chemistry and Catalysis, Utrecht UniVersity, Debye Institute, Sorbonnelaan 16,
584 CA Utrecht, The Netherlands, and Fritz-Haber-Institut der MPG, Faradayweg 4-6,
.§
‡
‡
,†
3
D-14195 Berlin, Germany
ReceiVed: May 26, 2004; In Final Form: August 5, 2004
The reduction behavior of Co/TiO
investigated by soft X-ray absorption spectroscopy (XAS). In situ XAS measurements of the L2,3 edges of Co
2
and Co/Mn/TiO
2
catalysts for Fischer-Tropsch synthesis has been
2
and Mn have been carried out during reduction treatments of the samples in H at a pressure of 2 mbar and
at temperatures up to 425 °C. The changes of Co and Mn 3d valences and the symmetries throughout the
reduction have been determined by comparison with theoretical calculations based on the charge transfer
multiplet code. Furthermore, bulk Co
of TiO as a support on the reducibility of Co oxides. The average Co valence at the various temperatures has
been determined from a linear combination of the reference spectra. It was found that the unsupported Co O
4
3 4
O has been reduced under the same conditions to evaluate the effect
2
3
0
was easily reduced to Co at 425 °C, whereas the Co
O
3 4
supported on TiO
2
catalysts was only reduced to a
mixture of CoO and Co , even after 12 h reduction at 425 °C. The presence of Mn further retards the reduction
0
III
of the supported Co
3 4
O particles. The Mn ions were easily reduced to MnO at temperatures lower than 300
°
C, and they remained in this oxidation state even after further temperature increase. In addition, catalytic
tests in the Fischer-Tropsch synthesis reaction at a pressure of 1 bar indicate that the selectivity of these
catalysts might be related to the extent of Co reduced after the activation treatment (i.e., the reduction with
H ).
2
Introduction
investigated over recent decades with the aim of enhancing the
catalytic performances of Co-based catalysts. The use of
manganese compounds as promoters for the hydrogenation of
carbon monoxide has been previously investigated. It was shown
that manganese promoters can increase the olefin selectivity of
Co-based catalysts in the FTS reaction and under certain
circumstances can enhance their activity and shift the distribution
The transformation process from syngas to hydrocarbons,
Fischer-Tropsch synthesis (FTS), is gaining in importance due
to the need for high purity fuels, which can be obtained from
feedstocks other than crude oil (i.e., natural gas, charcoal, or
biomass). These materials are converted into syngas by partial
1
oxidation or steam reforming processes. Supported Co catalysts
are claimed to be the most effective in achieving higher chain
12,13
of products toward larger chain hydrocarbons.
However, the
exact role of such promoters and their influence on the state of
cobalt particles throughout the FTS reaction remains unclear.
The use of in situ techniques for the characterization of
surfaces in heterogeneous catalysts is becoming an important
tool to investigate these materials under reaction conditions.
Photons in the range 100-1000 eV are suitable probes for the
electronic structure of reacting surfaces. Their interaction with
solid matter leads to photoabsorption processes, which can be
detected by the electrons created typically from a surface depth
of several nanometers, making the low energy XAS technique
surface sensitive. In recent years, it has become possible to
perform XAS measurements at pressure ranges of 1-10 mbar,14
which allows the carrying out of in situ treatments of the samples
growth probability values (R) and turnover rates compared to
other metal based catalysts. This makes them the catalysts of
choice for FTS.2,3 Reducibility of supported-cobalt particles
4-6
strongly depends on the support used and on the preparation
7,8
method. For supports without a strong interaction with cobalt,
such as silica, highly reducible cobalt oxide clusters are normally
9
,10
formed.
TiO2-supported catalysts have been shown to have
6
a strong metal-support interaction, which makes the cobalt
species difficult to be reduced, most likely due to a strong Co-O
interaction with the support or the formation of other stable
compounds (e.g., cobalt titanate (CoTiO3)) that are reduced only
at very high temperatures. The formation of the cobalt titanate
species has been reported before,6,11 and it is thought to occur
(e.g., calcinations and reductions), providing useful information
during the reduction from Co3O4 crystallites to CoO. During
II
about the surface composition of the catalysts under conditions
that are closer to the actual operating conditions. The L2,3 XAS
the reduction process, Co can diffuse into the support lattice,
leading to the formation of these compounds. The use of other
metals in small amounts (i.e., promoter elements) has been
spectral shapes of the 3d transition metal oxides can be simulated
accurately using the charge transfer multiplet code.1
5-17
These
calculations take into account the local symmetry and hybridiza-
tion. In general, they reproduce the L2,3 XAS very well, thereby
yielding information on the local symmetry (octahedral vs
tetrahedral, high spin vs low spin) including the crystal field
*
To whom correspondence should be addressed. E-mail:
b.m.weckhuysen@chem.uu.nl.
†
Utrecht University.
Fritz-Haber-Institut der MPG.
‡
§
Present address: Chemical Sciences Division, Lawrence Berkeley
National Laboratory, One Cyclotron Road, Berkeley, California 94720.
1
8
values and symmetry dependent covalence.
1
0.1021/jp0403846 CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/18/2004

16202 J. Phys. Chem. B, Vol. 108, No. 41, 2004
Morales et al.
In our study, in situ XAS measurements during reduction
treatments were carried out with two catalysts (Co/TiO2 and
Co/Mn/TiO2) and with a bulk Co3O4. With this work, we aim
to investigate the valences and symmetry of states for Mn and
Co at the different temperatures during the reduction and,
moreover, to evaluate the influence of TiO2 support on the
reducibility of the Co oxides due to the so-called metal-support
interaction. Bazin et al.19 have performed similar experiments
before on Co/SiO2 catalysts. In their study, they carried out the
XAS measurements in a UHV chamber, which was coupled to
another chamber for the sample treatments. However, this fact
can significantly affect the cobalt sites. For instance, the changes
in the pressure/gas environment before and during the measure-
ments can influence the average valence state of the cobalt sites.
In situ XAS studies on other catalyst systems reveal that
significant changes in the average valence can occur within
2
0
3 4
Figure 1. Co L2,3 edges for bulk Co O during reduction from RT to
oxide materials.
4
25 °C. All spectra have been normalized from zero to one.
Experimental Section
. Materials and Catalyst Preparation. Two TiO2-supported
The experiments were carried out during reduction treatments
under a dynamic atmosphere (i.e., flowing in H2). The samples
were inserted into the chamber, which was subsequently pumped
down to vacuum. After approximately 5 min, the chamber was
brought under pure hydrogen (100 mL/min) using a mass flow
controller. A gas mixing system regulated the flow and the total
pressure at the sample position, which was 2 mbar during the
experiments. After optimization of the signal by tuning the slit
size and the spot of the incoming beam, the temperature was
increased from RT to 425 °C with a ramp of 5 °C/min. The
ramp was temporarily stopped at 300, 350, and 425 °C to
measure both the Co and the Mn L edges. The bulk Co3O4 was
measured at 385 °C, in addition to the same temperatures used
to measure the catalysts. In addition, the catalysts were held
under H2 atmosphere at 425 °C for 12 h to evaluate further
changes in the spectra with reduction time. The electron yield
was measured by applying a voltage of 27 V on the first
aperture, which is electrically isolated from the cell.
1
catalysts were synthesized using the homogeneous deposition
precipitation (HDP) technique to load the Co and incipient
wetness impregnation (IWI) to load the Mn. Commercial Degusa
2
P25 titania (surface area of 45 m /g and pore volume of 0.27
3
cm /g) was used as the support material. The powder titania
was suspended in an aqueous solution containing cobalt nitrate,
and the pH was slowly increased by urea decomposition at 90
°
C under continuous stirring for 18 h. After drying at 110 °C
overnight, the material was sieved and a portion was calcined
in air at 400 °C for 4 h with a heating rate of 5 °C/min, obtaining
the Co/TiO2 catalyst. For the preparation of the Mn-promoted
catalyst, the other portion was impregnated with an aqueous
solution of manganese nitrate followed by the same calcination
procedure described above, obtaining the Co/Mn/TiO2 catalyst.
The Co and Mn loadings of both calcined catalysts were
determined by X-ray fluorescence analysis (XRF). The loadings
obtained were 7.5 wt % Co and 2 wt % Mn, respectively. The
high purity Co3O4 material was obtained from Merck, and the
reference compounds MnO, Mn2O3, and MnO2 were obtained
from Aldrich.
4. Temperature Programmed Reduction Experiments. The
reduction behavior of the supported metal oxides was studied
at a pressure of 1 bar by temperature programmed reduction
TPR) using a “Thermo Electron TPDRO 1100” piece of
(
equipment. The experiments were performed in a tubular quartz
reactor. After loading the sample, the reactor was flushed with
Ar at 120 °C for 1 h and then cooled to RT in flowing Ar.
Subsequently, the gas flow was adjusted to 5% H2/Ar and the
temperature was raised at the rate of 10 °C/min from RT to
2
. X-ray Diffraction. The bulk phase composition of both
catalysts was investigated after calcination by X-ray diffraction
XRD). Powder XRD measurements were performed using an
(
ENRAF-NONIUS XRD system with a curved position sensitive
INEL detector and applying a Co KR1 radiation source (λ )
.788 97 Å). In addition to the reflections of the TiO2 support,
only the diffraction peaks corresponded to the Co3O4 phase were
detected. The mean Co3O4 particle sizes were determined from
the line broadening of the diffraction line situated at 43.3°,
applying the Scherrer equation. The particle sizes obtained were
600 °C. The content of H2 in the outflowing gas was monitored
1
with a thermoconductivity detector throughout the reduction
treatment.
5. Catalytic Testing. The FTS reaction was carried out using
a glass plug flow reactor in which the calcined catalysts were
diluted in SiC. Before starting the reaction, the samples were
in situ activated in 20 mL/min flow of H2 at 350 °C for 2 h.
The reaction was carried out at a pressure of 1 bar, a temperature
of 220 °C, and a H2/CO ratio of 2 in the syngas composition.
The gas hourly space velocity (GHSV) was adjusted to obtain
CO conversions of ∼3%. The reaction temperature was
measured with a thermocouple inserted in the catalyst bed. The
hydrocarbon products’ composition was analyzed on-line using
a Varian CP-3800 gas chromatograph equipped with a fused
silica column of 50 m length and a flame ionization detector.
∼
15 and ∼13 nm for the Co/TiO2 and Co/Mn/TiO2 catalyst,
respectively.
3
. In Situ Soft X-ray Absorption Spectroscopy. The soft
X-ray absorption spectra of the manganese L2,3 edge (635-
60 eV), the cobalt L2,3 edge (770-800 eV), and the titanium
6
L2,3 edge (450-470 eV) were measured at beamline U56/2
PGM-2 at BESSY (Berlin). The spectral resolution of the
monochromator was ∼0.2 eV. The instrumentation for in situ
2
1
XAS and XPS measurements is described elsewhere.
A
stainless steel in situ cell was used in which the powdered
samples were fixed in a sample holder. The X-ray absorption
spectral shape was measured with the ionized gas conversion
total electron yield, which has a probing depth of approximately
Results
1. Co 2p XAS of Co3O4. Figure 1 shows the Co L2,3 edge
4
nm.
spectra of the bulk Co3O4 measured during the reduction

In Situ X-ray Absorption of Co/Mn/TiO2 Catalysts
J. Phys. Chem. B, Vol. 108, No. 41, 2004 16203
III
Figure 3. CTM calculations of octahedral Co (thin solid) and
II
tetrahedral Co (dashed). These spectra have been added with a 2:1
ratio (thick solid) to simulate the experimental Co
O
3 4
spectrum (points).
II
calculations of the cobalt sites. CoO contains octahedral Co
III
sites, and LiCoO2 contains octahedral Co sites. The charge
transfer parameters of the Co (Oh) sites have been optimized
to CoO, following the literature data. These parameters are
collected in Table 1. The CTM spectrum of Co (Oh) includes
2
Figure 2. Co L2,3 edges of the reference compounds LiCoO and CoO,
together with charge transfer multiplet (CTM) calculations of the cobalt
II
2
5
sites. From bottom to top, the spectra are respectively, the CTM
II
calculation of Co (O
h
), the CoO spectrum, the CTM calculation of
II
II
III
Co (T
d
), the CTM calculation of Co (O
The CoO and LiCoO
de Groot, are reproduced by the calculations shown below them.
h 2
), and the LiCoO spectrum.
the calculated energy intensity multiplet spectrum. These stick
spectra have been broadened with the 2p lifetime broadening
and an experimental broadening that is assumed to be a
Gaussian. The Lorentzian broadening is 0.4 eV for the L3 edge
and 0.8 eV for the L2 edge. The Gaussian broadening is 0.4
24
2
spectra, reproduced from de Groot et al. and
2
5
treatment from RT up to 425 °C. It can be observed that the
spectrum shape changes by increasing the temperature. At the
same time, the maximum energy peak shifts to lower energies.
These spectral changes are due to a shift from higher to lower
cobalt valence and variations in the symmetry of the cobalt
atomic states upon reduction. Comparison of the experimental
spectra (Figure 2) with the theoretical simulations led us to a
better understanding of these spectrum shapes. The spectrum
II
eV. Similarly, the Co (Oh) site has been optimized to LiCoO2.
In the case of LiCoO2 the agreement is not exactly correct, and
the experimental spectrum has a clear shoulder at 781 eV and
a less structured shoulder/peak at 785 eV.
Figure 3 shows the comparison of the CTM calculations of
III
II
octahedral Co and tetrahedral Co with Co3O4. These simula-
tions confirm that Co3O4 indeed consists of octahedral low spin
II
III
measured at RT corresponds to a mixture of Co and Co
oxidation states (ratio of 1/2) characteristic of Co3O4. Co3O4
has a spinel type structure where the oxygen atoms are arranged
1
III
4
II
( A1) Co and tetrahedral high spin ( A2) Co sites. Comparison
to the LiCoO2 spectrum in Figure 2 showed that LiCoO2 also
contains low spin ( A1) Co sites. The addition of twice the
II
III
1
III
in a fcc structure, and Co (Td) and Co (Oh) cations are located
2
2
III
II
in tetrahedral and octahedral coordination, respectively. As
soon as the temperature increases, the shape of the spectra
spectrum of Co plus one of the spectrum of Co is used to
simulate the experimental spectrum of Co3O4. Some discrep-
ancies are visible in the simulation, but these are mostly due to
the imperfect simulation of the low spin ( A1) Co sites, as it
was also the case for the LiCoO2 spectrum in Figure 2. Exactly
like the LiCoO2 spectrum, Co3O4 also has extra intensity at 781
eV and a less pronounced structure at 785 eV. Taking this into
account, the simulations confirm that Co3O4 consists of octa-
II
becomes more purely Co since Co3O4 is reduced to CoO. In
II
1
III
the CoO, the Co cations are octahedral coordinated by the
oxygen atoms. When the temperature reaches 385 °C, the
measured spectrum corresponds to pure CoO, as shown in the
comparison with a reference sample (Figure 2). This confirms
II
the complete shift of cobalt to Co (Oh) at 385 °C. The
0
1
III
4
II
consecutive reduction of CoO to Co easily occurs at 425 °C,
hedral low spin ( A1) Co and tetrahedral high spin ( A2) Co
sites.
0
at which point the pure Co spectrum is obtained. This spectrum
consists of a single peak at the L3 edge, characteristic of Co
Figure 4 shows the comparison of the CoO reference spectrum
with the experimental spectrum of Co3O4 at 385 °C. Because
the L3 edge shows much more structure than the L2 edge, we
only show the L3 edge region. The good agreement confirms
metal.24
Figure 2 shows the Co 2p X-ray absorption spectra of CoO
and LiCoO2, together with the charge transfer multiplet (CTM)
TABLE 1: Charge Transfer Parameters Used for the Calculations of Theoretical Spectraa
II
parameter
site symmetry
electronic symmetry
CoO
LiCoO
2
Co in Co
3
O
4
II
CoIII (O
II
Co (O
h
)
h
)
Co (T
d
)
high spin: 4T
, split by spin orbit to E
low spin: 1A
high spin: 4A
,
1
1
2
b
ionic crystal field [10Dq]
0.6
3.0
2.0
1.0
1.9
4.5
2.0
1.0
-0.3
7.0
c
charge transfer energy [∆]
b
hopping e
g
-electrons [t
e
]
1.0
2.0
b
hopping t2g-electrons [t
t
]
a
b
All values are given in eV. Hopping adds a “covalent contribution” to the overall crystal field (or ligand field) strength of ∼0.3 eV for the
c
values used. Core hole potential Q is assumed to be 2.0 eV higher than the 3d3d repulsion energy U, which implies that the final state charge
transfer energy ∆′ ) ∆ - 2.0 eV.
2
5,26

16204 J. Phys. Chem. B, Vol. 108, No. 41, 2004
Morales et al.
Figure 4. Comparison of the room temperature CoO reference
spectrum (dotted) with the experimental spectrum of Co at 385 °C
solid, with solid squares).
Figure 5. Co L2,3 edges measured for the Co/Mn/TiO
reduction. At RT, the spectrum corresponds mainly to Co
2
catalyst during
, which is
3
O
4
3
O
4
(
reduced to CoO at 350 °C. After 12 h of reduction at 425 °C, the
0
spectrum corresponds to CoO + Co .
that at 385 °C, Co3O4 is indeed fully reduced to CoO. There
are some small differences visible. The shoulder at 777 eV is
smaller in the reduced Co3O4, and the peaks at ∼779-780 eV
are slightly lower in the reduced Co3O4. A reason for these small
differences can be found in the difference in measurement
temperature between the CoO reference (RT) and the Co3O4
sample (385 °C). It is known that the X-ray absorption spectra
of CoO have a temperature dependence due to the population
2
4,25
of excited states.
In addition, it is likely that in the reduced
Co3O4 the crystallinity of the CoO is not perfect and there will
be some variation in site symmetry, oxygen vacancies, etc. These
variations will make the resulting spectrum slightly different
and can explain the small differences observed.
In conclusion, it is found from the reference compounds and
from the CTM calculations that Co3O4 is a mixture of octahedral
1
III
4
II
(
A1) Co and tetrahedral ( A2) Co sites. During reduction,
III
4
II
the Co is reduced to octahedral ( T1) Co ; in addition, the
tetrahedral ( A2) Co sites are modified into octahedral ( T1)
Co sites in CoO. All spectra at intermediate temperatures can
4
II
4
II
Figure 6. Comparison of the average Co valence for the bulk Co O
3
4
be considered as a linear combination of the spectra of Co3O4
and CoO.
2
and the Co/Mn/TiO catalyst during the reduction treatment.
2
. Co 2p XAS of the (Mn)/Co/TiO2 Catalysts. The Co L2,3
edges of the catalysts were measured at 25, 300, 350, and 425
C, in addition to another measurement performed after 12 h at
25 °C. It was found that the reduction behavior was rather
the bulk cobalt during the reduction. Figure 6 shows a
comparison of the bulk Co3O4 with the Co/Mn/TiO2 catalyst.
It can be observed that unsupported Co3O4 is reduced to Co at
425 °C, in contrast with the TiO2-supported cobalt catalyst that
is only partly reduced under the same conditions. Furthermore,
some differences in the amount of cobalt metal were found
between both catalysts after the full reduction. The surface Co
compositions were calculated for both catalysts after 12 h of
0
°
4
similar in both catalysts, but significantly different from the bulk
cobalt oxide discussed above. As an example, we show the
spectra of the Co/Mn/TiO2 catalyst in Figure 5. At RT, the
spectra of both catalysts correspond to (mainly) Co3O4. The
reduction of Co3O4 to CoO is achieved at lower temperatures
than in the bulk Co3O4. This might be due to the lower cobalt
content in the catalyst, which results in less time needed to
achieve the complete reduction to CoO. In none of the catalysts
0
reduction at 425 °C. The compositions obtained were 39% Co
0
and 61% CoO for the Co/TiO2 catalyst and 25% Co and 75%
CoO for the Co/Mn/TiO2 catalyst. Hence, the presence of Mn
0
in the catalyst was found to decrease the amount of Co obtained
0
did the CoO compounds significantly reduce to Co at 425 °C.
after H2 reduction at a pressure of 2 mbar.
Instead, the spectra consist mainly of CoO with only a small
3. Mn 2p XAS of the Mn/Co/TiO2 Catalysts. The three
manganese reference compounds were measured at RT to obtain
0
amount of Co . These results indicate that CoO supported on
0
IV
III
II
TiO2 is difficult to reduce to Co at the conditions used in this
the spectra of pure Mn , Mn , and Mn . Figure 7 shows the
2p X-ray absorption spectra of the MnO, Mn2O3, and MnO2
references. The Mn spectra can again be simulated using CTM
work (i.e., 2 mbar of H2 at 425 °C).
To quantify the changes of the Co valence throughout the
reduction, linear combinations of the spectra of the pure Co
compounds were used to calculate the Co average valence at
each temperature. For these calculations, the experimental
spectra of bulk cobalt measured at RT, 385, and 425 °C were
2
5
calculations, as discussed previously. We will now use these
reference compounds in the analysis of the Mn spectra of the
catalyst.
The Mn L2,3 edges of the Co/Mn/TiO2 catalyst were measured
at the same conditions as Co, at RT and during the reduction at
300, 350, and 425 °C. Changes in the shape of the spectra were
found between RT and 300 °C, whereas the spectra did not
0
used as the spectra of the pure Co3O4, CoO, and Co compounds,
respectively. Hence, the change of cobalt valence with temper-
ature could be followed and compared in both the catalysts and

In Situ X-ray Absorption of Co/Mn/TiO2 Catalysts
J. Phys. Chem. B, Vol. 108, No. 41, 2004 16205
Figure 7. Mn L2,3 edges of the manganese reference compounds MnO
2 3 2
(solid), Mn O (dots), and MnO (dashed).
2 2
Figure 9. TPR profiles of Co/TiO (solid) and Co/Mn/TiO (dashed)
catalysts.
TABLE 2: FTS Catalytic Performances
sample
code
Co-time % CH
4
% C5+
R
butane/butene
ratio
yielda
selec.
selec. valuesb
Co/TiO
Co/Mn/TiO
2
2.35
2.50
31.6
24.9
28.0
34.8
0.54
0.58
0.34
0.22
2
a
Co-time yield: 10-5 mol CO (g Co)-1 s-1. b As calculated from
and C in the ASF distribution plot, according to
/n ) log(ln R) + n log R, where W
fraction of the products with n carbon number.
the slope between C
the equation log W
3
7
2
n
n
is the weight
of Co-support interaction present in the catalyst. The Co/Mn/
TiO2 catalyst shows a similar TPR profile, although the
maximum temperature peak of the second reduction step is
somewhat more pronounced. Therefore, this experiment indi-
cates a decrease of the cobalt reducibility in the Co/Mn/TiO2
catalyst with respect to the Co/TiO2 catalyst. In addition, the
small peak appearing at ∼210 °C for the Co/Mn/TiO2 catalyst
is attributed to the reduction of the Mn oxides.
2
Figure 8. Mn L2,3 edges of Co/Mn/TiO catalyst before (dashed) and
after (solid) the reduction treatment.
change more by raising the temperature above 300 °C. For this
reason, we only show the spectra before and after the treatment
5. Catalytic Testing. Catalytic performances of the catalysts
(Figure 8). By comparing the measured spectra of the catalyst
were compared at pseudo-steady-state conditions after a 40 h
FTS reaction. The activities were calculated as the % CO
converted into hydrocarbons without taking into account other
side reactions (i.e., CO2 formation). The paraffin/olefin ratios
were calculated using the butane and butene yields. Table 2
gathers the resulting catalytic performances for both catalysts
after 40 h of reaction. The Co/Mn/TiO2 catalyst gave a slightly
higher Co-time yield than the Co/TiO2 catalyst, although this
small difference is within the measurement error. On the other
hand, the presence of Mn in the Co/Mn/TiO2 catalyst resulted
in a suppression of the methane make and enhanced the
selectivity toward high weight hydrocarbons (lower % CH4 and
higher % C5+). In addition, a small increase of the chain growth
probability (R) was found. This shift in the product’s mass
distribution for the Mn-promoted catalyst was accompanied by
an increase of the olefinic product yield.
with the reference compounds, it can be concluded that the Mn
II
contained in the calcined catalyst consists of a mixture of Mn
III
IV
and Mn , with possibly some admixture of Mn . The spectral
III
shape indicates that Mn is the predominant Mn species, which
could be in the form of Mn2O3, although it is also plausible
that some mixed oxides such as Co2MnO4 and Mn2CoO4 are
present, as reported in the literature.26 However, the spectral
quality does not allow for an accurate quantification of the
amount of each compound. Upon reduction, the Mn present in
the catalysts is already fully reduced to MnO at 300 °C, and no
other variations are found in the spectra recorded at higher
temperatures. The absence of the peak at 639 eV in the
experimental spectra of the catalysts is due to a smaller value
of the crystal field, which is influenced by defects in the crystal
structure. These results indicate a high stability of MnO at these
reduction conditions and are in agreement with other work where
MnO was also found to be the main compound in supported
manganese oxide catalysts after reduction.27
Discussion
4
. Temperature Programmed Reduction. Figure 9 shows
It has been shown that the use of soft X-ray absorption
spectroscopy as a tool for studying the composition of catalyst
surfaces can be of great interest for determining the oxidation
state of 3d transition metals during the activation of the catalysts.
Moreover, the fact that most catalytic reactions take place at
the surface of the metal particles makes this technique very
attractive for studying catalytic systems, since the information
obtained corresponds to a depth of a few nanometers into the
metal surface, not to the bulk. Nevertheless, the low pressure
the TPR profiles obtained for both catalysts. The reduction of
the Co/TiO2 catalyst results in two main peaks with maxima at
80 and 455 °C. We attribute these two H2-consumption peaks
2
to the reduction steps from Co3O4 to CoO and from CoO to
0
Co . The areas of the two peaks are in agreement with the
stoichiometry of both reduction steps, which is in the ratio of
/3. Moreover, the second reduction step shows two different
maxima (350 and 455 °C), and this suggests a different extent
1

16206 J. Phys. Chem. B, Vol. 108, No. 41, 2004
Morales et al.
conditions (2 mbar) required to perform these measurements
are still too far from the actual operating conditions (1 bar or
higher). Therefore, the information obtained with the XAS
technique cannot be directly correlated with that from other
characterization techniques.
be due to the modified catalyst surface resulted upon the Mn
addition, which restrains the secondary hydrogenation of olefins
with respect to the Co/TiO2 catalyst. The chain growth prob-
abilities are thought to be clearly influenced by the rate of
R-olefin reattachment on the Co surface and subsequent
29
growth. Therefore, the restriction of the secondary hydrogena-
tion of olefins leads to a more effective chain growth in the
promoted catalyst. Other work has also reported changes in the
With the current experiments, we intended to investigate the
surface composition of the cobalt and manganese contained in
two FTS catalysts. The XAS measurements allowed us to
calculate the valences of both transition metals, which play a
role during FTS reaction (i.e., after the activation treatment). It
turned out that the support oxide has a strong influence on the
reduction behavior of the cobalt oxides, since CoO could not
3
0
R as a function of paraffin/olefin ratios.
Conclusions
TiO influences the reduction behavior of Co O particles
2
3
4
0
be reduced to Co in any of the supported catalysts, whereas a
by decreasing the extent of reduced cobalt present in the
bulk Co3O4 sample could be fully reduced to metallic Co. This
should be due to the existence of a strong cobalt-support
interaction, which makes it difficult for the CoO reduction to
occur in the presence of TiO2 support. This fact suggests that
the active sites of FTS catalysts may consist of a mixture of
catalysts after reduction. The final cobalt composition of the
TiO -supported catalysts was mainly CoO and some amount
2
0
0
of Co , whereas bulk Co O could be reduced to pure Co after
3
4
the same treatment. These results point toward the existence of
a strong Co-TiO (metal-support) interaction, which affects
2
0
II
Co and Co and that their composition depends on the
pretreatment performed before FTS reaction and on the extent
of metal-support interaction present in the catalyst. We note
that the present result does not allow us to judge exactly what
the composition of the reduced catalysts. The Mn contained in
the Co/Mn/TiO catalyst after calcination was found to be
2
II
III
IV
present as a mixture of Mn , Mn , and Mn , and as soon as
the catalyst was reduced at 300 °C, only MnO was measured.
Furthermore, these Mn compounds had an extra effect on the
Co reducibility of this catalyst, since the final amount of Co0
was significantly lower than in the Co/TiO2 catalyst. This
decrease of the Co reducibility resulted in an increase of the
C5+ yield and a higher olefin selectivity without any loss in
FTS activity.
0
II
kind of mixture between Co and Co exists. Possible options
are a physical mixture of Co and CoO, as well as the existence
of another phase (for example, a suboxide and/or a changed
microstructure). Here, we note that the CoO nanoparticles are
quite small (∼10-20 nm), and in case of a mixed metal-oxide
particle there will be a significant percentage of cobalt atoms
that will feel neither a potential as pure Co (nanosize) metal
nor a potential of pure CoO. The only conclusion we would
like to draw is the measured ratio of Co to CoO, which best
explains the observed spectral shape.
Acknowledgment. The authors thank the BESSY staff for
their continual support during the XAS measurements at the
synchrotron in Berlin. The authors gratefully acknowledge
financial support from Shell Global Solutions and the many
fruitful discussions with Heiko Oosterbeek, Carl Mesters, and
Herman Kuipers. We thank Mr. I. Boussadkat (Utrecht Uni-
versity) for assisting with the data analysis and calculations.
The research of P. G. is supported by grants from Netherlands
Scientific Organization-Chemical-Sciences (NWO-CW), and the
research of FMFdG is supported by The Netherlands Research
School Combination on Catalysis (NRSCC) and by a Science-
Renewal Fund of NWO-CW.
The XAS measurements allowed measuring of the amount
0
of Co , and the surface of the cobalt particles contained in the
three samples after the reduction treatments can be ordered, from
most to least metallic, as follows: Co3O4 > Co/TiO2 > Co/
Mn/TiO2. From this calculation, we can see that both TiO2 and
0
Mn have an effect on the extent of Co at the surface of the Co
particles. The presence of Mn in the catalyst prevents the CoO
reducibility to some extent at 2 mbar conditions. In addition, a
similar trend is found with the reduction experiments at 1 bar.
This decrease of the Co reducibility is reflected in the catalytic
performances mainly by an increase in olefin yield and a
decrease in CH4 formation. These results suggest that a more
oxidic composition of the cobalt surface favors the chance to
terminate the hydrocarbon growth by â-hydrogen abstraction,
leading to an increase of the R-olefin content in the product
composition. However, this suppression in the hydrogenation
rate found for the Co/Mn/TiO2 catalysts might be merely related
to the presence of MnO interacting with the Co particles, rather
than the more oxidic Co surface resulting upon the decrease of
the Co reducibility. We note that even though the TPR
experiments provide a trend similar to the XAS measurements
concerning the Co reducibility in both catalysts, it is uncertain
what their Co surface composition is under FTS conditions, since
the Co sites may suffer structural and compositional changes
during FTS.28 Hence, the changes in selectivity may be also
accounted to the presence of MnO. Similar effects on Mn-
References and Notes
(1) Rostrup-Nielsen, J. R. Catal. Today 2002, 71, 243.
(
(
(
2) Iglesia, E. Appl. Catal., A 1997, 161, 59.
3) Schulz, H. Appl. Catal., A 1999, 186, 3.
4) Bartholomew, C. H.; Reuel, R. C. Ind. Eng. Chem. 1985, 24, 56.
(5) Riva, R.; Miessner, H.; Vitali, R.; Del Piero, G. Appl. Catal., A
999, 196, 111.
1
(
6) Vob, M.; Borgmann, D.; Wedler. J. Catal. 2002, 212, 10.
(7) Jacobs, G.; Das, T. K.; Zhang, Y.; Li, J.; Racoillet, G.; Davis, B.
H. Appl. Catal., A 2002, 233, 263.
(
(
8) Kraum, M.; Baerns, M. Appl. Catal., A 1999, 186, 189.
9) Coulter, K. E.; Sault, A. G. J. Catal. 1995, 154, 56.
(
10) Ernst, B.; Libs, S.; Chaumette, P.; Kiennemann, A. Appl. Catal., A
1
2
999, 186, 145.
(11) Brik, Y.; Kacimi, M.; Ziyad, M.; Bonzon-Verduraz, F. J. Catal.
001, 202, 118.
(
12) Barrault, J. Stud. Surf. Sci. Catal. 1982, 11, 225.
(13) Van der Riet, M.; Copperthwaite, R. G.; Hutchings, G. J. J. Chem.
Soc., Faraday Trans. 1 1987, 83, 2963.
14) Knop-Gericke, A.; Havecker, M.; Schedel-Niedrig, T.; Schlogl, R.
Top. Catal. 2000, 10, 187.
15) de Groot, F. M. F.; Fuggle, J. C.; Thole, B. T.; Sawatzky, G. A.
(
12,13
promoted catalysts have been observed before by other groups.
(
Therein, they found a higher selectivity toward light olefins and
a lower CH4 production in the FTS upon Mn addition to Co-
based catalysts, although no characterization was conducted to
explain the reasons for these selectivity changes. Furthermore,
the enhancement of the C5+ selectivity found in this work may
Phys. ReV. B. 1990, 42, 5459.
(16) de Groot, F. M. F. J. Elect. Spectrosc. Relat. Phenom. 1994, 67,
5
29.
(
17) de Groot, F. M. F. Chem. ReV. 2001, 101, 1779.
(
18) Wasinger, E. C.; de Groot, F. M. F.; Hedman, K. O.; Hodgson, K.
O.; Solomon, E. I. J. Am. Chem. Soc. 2003, 125, 12894.

In Situ X-ray Absorption of Co/Mn/TiO2 Catalysts
J. Phys. Chem. B, Vol. 108, No. 41, 2004 16207
(
19) Bazin, D.; Kovacs, I.; Guczi, L.; Parent, P.; Laffon, C.; de Groot,
F. M. F.; Ducreux, O. J. Catal. 2000, 189, 456.
20) Heijboer, W. M.; Battiston, A. A.; Knop-Gericke, A.; Havecker,
(24) de Groot, F. M. F.; Abbate, M.; Van Elp, J.; Sawatzky, G. A.; Ma,
Y. J.; Chen, C. T.; Sette, F. J. Phys.: Condens. Matter 1993, 5, 2277.
(25) de Groot, F. M. F. Ph.D. Thesis, 1991.
(
M.; Mayer, R.; Bluhm, H.; Schlogl, R.; Weckhuysen, B. M.; Koningsberger,
D. C.; de Groot, F. M. F. J. Phys. Chem. B. 2003, 107, 13069.
(26) Cuse, K.; Rapp, H. J. Anal. Chem. 1993, 346, 84.
(27) Boot, L. A.; Kerkhoffs, M. H. J. V.; Van der linder, B. T.; Dillen,
A. J. V.; Geus, J. W.; Van Buren, F. R. Appl. Catal., A 1996, 137, 69.
(28) Schulz, H.; Nie, Z.; Ousmanov, F. Catal. Today 2002, 71, 351.
(29) Iglesia, E.; Soled, S. L.; Fiato, R. A.; Via, G. H. J. Catal. 1993,
143, 345.
(21) Ogletree, D. F.; Bluhm, H.; Lebedev, G.; Fadley, C. S.; Hussain,
Z.; Salmeron, M. ReV. Sci. Instrum. 2002, 73, 3872.
(
22) Kim, K. J.; Park, Y. R. Solid State Commun. 2003, 127, 25.
(23) Chen, C. T.; Idzerda, Y.; Lin, H. J.; Smith, N. V.; Meigs, G.;
Chaban, E.; Ho, G. H.; Pellegrin, E.; Sette, F. Phys. ReV. Lett. 1995, 75,
(30) Kuipers, E. W.; Vinkenburg, I. H.; Oosterbeek, H. J. Catal. 1994,
152, 137.
1
52.