JOURNAL OF CATALYSIS 173, 64–73 (1998)

ARTICLE NO. CA971885

Synthesis of New Fe–Co Based Metal/Oxide Composite Materials:

Application to the Fischer–Tropsch Synthesis
C. Cabet, A. C. Roger, A. Kiennemann, S. Läkamp,∗ and G. Pourroy∗
LERCSI URA CNRS 1498, 1, rue Blaise Pascal, 67008 Strasbourg, France; and ∗ IPCMS UMR CNRS 046, 23, rue du Loess, 67037 Strasbourg, France

Received February 21, 1997; revised July 17, 1997; accepted August 12, 1997

proposed as well as the existence of metal–Mn solid solu-

An original synthesis of cobalt–iron alloy/cobalt doped magnetite tions at different steps of the preparation and more par-
catalysts (spinel structure) has been developed, in which cobalt and ticularly in the case of iron on a freshly reduced catalyst
iron are obtained in the metallic state without any reducing treat- (8–11). The high selectivity of C2–C4 has been correlated
ment. The chemical phenomena occurring during the preparation, to the presence of a solid solution (12), in Fe–Mn cata-
based upon the coprecipitation of the hydroxides in a boiling basic lysts which, for iron rich examples, was a mixed spinel
medium, are detailed. By heating under argon, the composition of
phase (Fe1−yMny)3O4 accompanied by two carbide phases
the metallic phase can be controlled within the range (Co00.8 Fe00.2 ) to
(13). For manganese rich systems, Butt (13) proposed two
(Co00.95 Fe00.05 ). X-ray diffraction, thermogravimetric analyses, and
X-ray photoelectron spectroscopy experiments lead to the charac- spinel phases, one abundant in iron Fe3−yMnyO4, the other
terization of the crystalline phases and their location in the bulk rich in manganese β-(Fe1−xMnx)2O3, two carbide phases
and at the surface of the catalysts. The catalytic behavior of these (ε 0-Fe2.2C and λ-Fe5C2), and the wustite phase. Unfortu-
composites in the hydrogenation of carbon monoxide is studied. No nately, in such catalysts, the carbide phase formed during
reducing treatment is necessary for them to be active for the Fischer– the reaction has a short lifetime, which therefore hinders
Tropsch synthesis. Moreover, an efficient treatment for protecting their industrial development. Cobalt does not carburize as
the spinel phase against reduction and subsequent carburization is readily as iron in presence of CO/H2 and numerous stud-
established. The ability of such catalysts to produce C2–C4 olefins ies have confirmed its efficiency for the obtention of light
is also discussed. °c 1998 Academic Press olefins (4, 14–16). As for iron based catalysts, the cobalt–
manganese solid solution is essential. A catalyst where the
Co : Mn atomic ratio is 1 : 1 is mainly composed of a M3O4
INTRODUCTION spinel in which the M cations (Co and Mn) were statisti-
cally distributed between the tetrahedral and octahedral
The Fischer–Tropsch catalysts which produce hydrocar- sites. After reduction, two phases were detected: MnO and
bons from syngas are highly diversified in their formulation metallic cobalt. At 220◦ C, under CO/H2 flow, a body cen-
as are the conditions under which they are used. However, tered cubic (bcc) cobalt phase was reported to be present
the metals readily available for obtaining industrially higher (17, 18).
hydrocarbons on an industrial scale are mainly restricted to From the studies discussing the synthesis of light olefins,
iron and cobalt, and, to a lesser extent, to ruthenium and the following point arises: the spinel structure plays a
nickel. The catalytic formulations have been developed in prominent part in the preparation step. In the case of iron
two ways: for the obtention of chemical families of products based catalysts, under the reaction conditions, the spinel is
(paraffins, olefins, alcohols) or narrow distributions in one only partially destroyed but free Fe readily carburizes and
given family (C2–C4 olefins, gasoline, diesel, waxes. . .). the catalytic activity is then reduced. In the case of cobalt
Due to the thermodynamic and kinetic limitations of the based catalysts the carbide formation is avoided. However,
reaction, few catalysts are able to amplify the C2–C4 hydro- the catalysts are not active in the spinel form but as metal
carbons fraction. However, some examples are reported in dispersed on partially reduced oxide matrix.
the literature and these are Fe and/or Co based catalysts The aim of this work was to combine the respective ad-
on partially reductible oxide supports such as MnO2, V2O5, vantages of iron and cobalt based catalysts, i.e., no de-
and TiO2 instead of the conventional inert supports like struction of the spinel structure and presence of a metal
SiO2 or Al2O3 (1–4). In view of producing C2–C4 olefins, dispersed on this spinel without formation of metallic car-
the Fe–Mn and Co–Mn catalysts, with or without alka- bides. The synthesis, the characterization, and the activity
line promoters, are the most studied (5–7). For these cata- in syngas reaction of Co metal/Fe–Co spinel catalysts are
lysts, a strong metal–manganese oxide interaction has been described. The preparative method is original and leads to
64
0021-9517/98 $25.00
Copyright ° c 1998 by Academic Press
All rights of reproduction in any form reserved.
 SYNTHESIS OF Fe–Co BASED METAL OXIDE COMPOSITES 65

a metallic phase (composed mainly of cobalt) without any

preliminary reducing treatment, as is usually the case for
the catalytic activation in the CO hydrogenation.

EXPERIMENTAL

1. Preparation of the Catalysts

The synthesis of the metal–ferrite compounds, developed
by A. Malats et al. (19–23), has been adapted to the spe-
cific requirements of catalysis, particularly to the necessity
of obtaining adequate specific surface areas. This synthesis
relies upon the coprecipitation of hydroxides from cobal-
tous and ferrous chlorides in boiling concentrated KOH
solution. Potassium hydroxide is preferred to sodium hy-
droxide because the solubility of KCl, formed during the
synthesis, in hot water is greater than that of NaCl, which
renders the washing of the catalyst much easier. Further-
more, the ionic radius of K+ (1.33 Å) is larger than that of
Na+ (0.97 Å) and avoids the possibility of cationic inser-
tions. The respective variation of KOH concentration and
of the Co/Fe atomic ratio can lead to a wide range of compo-
sitions and properties for the metal–ferrite compounds. In
the example discussed below, the Co/Fe atomic ratio equals
1/3 while the potassium hydroxide concentration is 2 mol
liter−1.
CoCl2 · 6H2O (4.02 g) (ACROS) and 10.08 g of FeCl2 ·
FIG. 1. Synthetic scheme for metal–ferrite compounds.
4H2O (ACROS) were dissolved in 28 ml of water at ambi-
ent temperature, in order to obtain a solution where [Co +
2. Characterization of the Catalysts
Fe] = 2 mol liter−1 (crystallization of water molecules be-
ing taken into account). This solution was then added with The samples were studied by X-ray diffraction (XRD).
stirring to 150 ml of boiling potassium hydroxide solution Data were collected at room temperature using a D5000
([KOH] = 2 mol liter−1) obtained by dissolution of 19.7 g of Siemens diffractometer equiped with a primary beam
KOH (BDH Anala R) in 148 ml of water. A black powder quartz monochromator (Co Kα 1 = 1.78897 Å).
is precipitated. In one case (drying method), the precip- The specific surface areas of the catalysts were measured
itate was dried by total evaporation of water during 1 h by the BET method based on the N2 physisorption capacity
with a final temperature attained of 152◦ C. In the second at 77 K. The values are summarized in Table 1.
case (reflux method), the precipitate was matured by re- The Co/Fe ratios were determined by atomic absorp-
flux at 124◦ C for 1 h in the boiling solution. The resulting tion at the Service Central d’Analyse du CNRS (Vernaison,
powders from drying and reflux methods were then filtered France) and by energy dispersive X-ray analysis. The mor-
and washed with hot water until the filtrate was neutral and phology of these materials were observed by scanning elec-
was presumed to contain no chloride ions (negative AgNO3 tron microscopy (SEM) on a JEOL 840 apparatus.
test). The precipitates were then dried at 40◦ C for 2 h. Fine Thermal gravimetric and thermal differential analyses
black powders were obtained. Accordingly, the material ob- (TGA–TDA) were carried out in alumina crucibles under
tained from the drying method is labeled “dried” catalyst atmospheres of nitrogen and air, using a SETARAM 92-12
and that from reflux method “refluxed” catalyst. The prepa- apparatus.
ration routes are summarized in Fig. 1. Both precipitates The surfaces of the samples were examined by X-ray pho-
were heated for 10 h under argon at 415◦ C and labeled, re- toelectron spectroscopy (XPS) on a VG-ESCA 3 apparatus.
spectively, treated “dried” and treated “refluxed” catalyst. Binding energies were calibrated with respect to the signal
Where necessary (see Results and Discussion Section 3.1), for adventitious carbon (binding energy = 284.8 eV).
the treated dried catalyst was submitted to various reduc-
ing treatments. It was heated under 10% H2/He from 20 3. Reactivity in the CO Hydrogenation
to 220◦ C (slope 0.2◦ C min−1) and maintained under these 3.1. CO disproportionation. Catalyst (0.1 g) was heated
conditions for either 0, 4, or 8 h. under helium (20 liters h−1 g−1 ◦ ◦ −1
cat ) at 220 C (1 C min ). Pulses
66 CABET ET AL.

TABLE 1 follows:
Specific Areas of the Fe–Co Catalysts moles of transformed CO
CO conversion (%) = 100 ×
moles of initial CO
Specific
Catalyst and surface area selectivity to products (%)
treatment (m2 g−1
cat )
CO moles transformed into the given product
Dried catalyst Fresh 28
= 100 ×
total transformed CO moles
Treated dried Fresh 20
catalyst mass of Ci
selectivity to Ci (wt%)=100 × .
Heated at 200◦ C under 10% H2/He 12 total mass of hydrocarbons
Heated and maintained 4 h at 200◦ C 13
under 10% H2/He
RESULTS AND DISCUSSION
Heated and maintained 8 h at 200◦ C 13
under 10% H2/He
After CO/H2 reaction <1 1. Characterization of Dried and Refluxed Catalysts
Refluxed catalyst Fresh 13
1.1. X-ray diffraction. The diffraction diagrams of
Treated refluxed Fresh 8
catalyst dried and refluxed catalysts are shown in Fig. 2. The com-
After CO/H2 reaction 2 pounds are highly crystallized (narrow diffraction peaks
and low background). At room temperature, they are made
Note. Dried catalyst, treated dried catalyst, refluxed catalyst, and of two phases, a spinel phase (named s) and a metallic phase
treated refluxed catalyst are prepared according to Fig. 1.
isomorphous to α-Fe having a bcc structure (named b). The
lattice parameter of the spinel is a = 8.402(5) Å. The corre-
sponding value, a = 2.838(5) Å, for the metallic phase indi-
of carbon monoxide (0.5 ml) were admitted onto the cata- cated that the bcc phase is an iron–cobalt alloy containing
lyst at steady intervals (8 min), using a six-port switching about 20% of Fe (24). Thus, the dried and refluxed cata-
valve. This process was repeated until a stabilization of the lysts can be formulated as (Co00.8 Fe00.2 )β [CoxFeyO4]. In both
amount of CO2 formed was observed. The exit gases (CO cases, a high background in the low angles region revealed
and CO2) were analyzed by gas chromatography. the presence of an amorphous phase, possibly hydroxides;
When necessary, the reducing treatments were per- it is more pronounced for the dried catalyst.
formed in situ under 10% H2/He (total gas flow: 30 liters
h−1 g−1 1.2. Elemental analysis—energy dispersive X-ray anal-
cat ).
ysis. The Co/Fe atomic ratios determined by elemental
3.2. Catalytic tests. Catalytic tests were carried out in a analysis are consistent with those determined by energy
fixed bed reactor. The feed gas flow rate was adjusted by dispersive X-ray analysis. The values obtained for the dried
mass flow controller. The pressure was regulated by cou- and refluxed catalysts are 0.42 and 0.40, respectively. For
pling a pressure comparator and an electronic control valve. the treated dried and treated refluxed catalyst, they are 0.43
For no reducing treatment, the catalyst (300 mg) was and 0.40, respectively. The Co/Fe ratios remain unchanged
heated to 200◦ C (0.2◦ C min−1) under nitrogen (2.3 liters h−1 during the thermal treatment. However, all of the ratios are
at 1 MPa). Then, the nitrogen was replaced by the CO/H2
mixture (CO/H2 = 1, total flow gas = 0.5 liter h−1, GHSV =
2000 h−1, P = 1 MPa) and the temperature increased to
220◦ C.
The catalysts were also tested under their reduced form.
The reducing treatment was performed in situ at 1 MPa
under a 10% H2/N2 (total flow gas 2.6 liters h−1) from 20 to
200◦ C (0.2◦ C min−1). At 200◦ C, the gases were substituted
by the CO/H2 mixture and the temperature was further
increased. For each temperature, on one side, the catalytic
results are given at the steady state after a stabilization of
almost 50 h, on the other side, the catalyst is maintained
120 h while the analyses are performed at regular intervals.
The exit gases were analyzed by on-line gas chromatogra-
phy. Liquid products were regularly collected from a cooled
trap and analyzed by chromatography. FIG. 2. X-ray diffraction patterns of the dried catalyst and the re-
The CO conversion (%), the molar selectivities (%) and fluxed catalyst prepared according to Fig. 1. s and b refer to spinel and
the hydrocarbon distributions (wt%) were calculated as Fe–Co alloy, respectively.
 SYNTHESIS OF Fe–Co BASED METAL OXIDE COMPOSITES 67

FIG. 3. Scanning electron micrographs (SEM) of the dried catalyst (left) and the refluxed catalyst (right) prepared according to Fig. 1.

found to be higher than the expected value of 0.33, that of

the starting solution. The values of Co/Fe ratio indicate that
some iron is lost during the preparation. Neither chlorine
nor potassium were detected in the bulk.
1.3. Scanning electron microscopy. The examination of
dried and refluxed catalysts (see Fig. 3) revealed structural
differences which are probably due to the maturation in
boiling KOH. The dried catalyst comprised large grains
of octahedral shape, having sides of almost 8 µm. These
large grains are embedded in a matrix consisting of small
grains. The refluxed catalyst is composed of small grains
(100–300 nm) which exhibited no defined geometry. Thus,
the former appeared more heterogeneous in terms of its
morphology. During the preparation, it was matured by
evaporation to dryness while the latter was matured in boil-
ing medium, which might explain why in the case of dried
catalyst, grain growth occurred in preferential sites.
1.4. TGA–TDA analyses. To quantify the amorphous
phase, thermal gravimetric analyses were performed un-
der inert gas (nitrogen) between 20 and 900◦ C (slope
5◦ C min−1). They are presented in Fig. 4a. For the dried cata-
lyst, the weight loss between 150 and 600◦ C, attributable
to the decomposition of the amorphous hydroxide phase
detected by XRD, is 2.71%. In the case of the refluxed cata-
lyst, the corresponding weight loss is only 0.77% and con-
firms that it contains less amorphous phase than the dried
catalyst.
The diffraction pattern of this latter following the above
thermal gravimetric analysis, indicated the presence of
cobalt doped magnetite spinel, a cobalt fcc phase (f) and
traces of cobalt oxide CoO. The presence of this oxide
would suggest that the hydroxide amorphous phase, initially
present in the dried catalyst, is mainly cobalt hydroxide.
A gravimetric analysis in air allows to precise their for- FIG. 4. Thermal gravimetric analysis (TGA) of the dried catalyst (—)
mula. Indeed, from the mass increase due to oxidation, β, and the refluxed catalyst (- -) prepared according to Fig. 1(a) under nitro-
x, and y of the formula (Co00.8 Fe00.2 )β [CoxFeyO4] can be gen and (b) under air.
68 CABET ET AL.

determined. The presence of an amorphous phase prevents It must be noted that the β value is close to 1.0, the high-
any precise formulation. However, approximations could est possible value. Indeed, the metallic phase results from
be made. For the dried catalyst, the weight increase due to the disproportionation of Fe(II) according to the reactions
oxidation is 7.24% (Fig. 4b), while that due to the hydrox- given above, thus the total amount of (Co + Fe) metallic
ide decomposition is 2.71%, giving a value of 7.44% for can not exceed a third of (Co + Fe) in the spinel.
the mass increase corrected with respect to the presence To summarize, the dried and refluxed catalysts are close
of hydroxides. The elemental analysis gave Co/Fe = 0.42, to identical in their composition. The differences between
this corresponds to a global composition of CoFe2.38Op. the compounds lie in their specific surface areas (respec-
The mass increase leads to the value of p = 3.42, so the tively 28 and 13 m2 g−1), in their morphological aspect and
global compound can be formulated as Co1.17Fe2.78O4. XRD in the amount of amorphous phase (probably cobalt rich hy-
suggested that the compound corresponds to (Co00.8 Fe00.2 )β droxides) they contain. These differences can be attributed
[FexCoyO4]. The identification of the parameters β, x, and to the methods of maturation during their preparation.
y leads now to the formula (Co00.8 Fe00.2 )0.95 [Fe2.66Co0.34O4]
for the dried catalyst. In the case of the refluxed catalyst, 2. Characterization of Treated “Dried”
the corrected weight increase due to oxidation is 7.46% and “Refluxed” Catalysts
(Fig. 4b) and the weight loss due to the hydroxide decom-
The structural evolution during heating under argon is
position is 0.77%. The calculated formula is (Co00.8 Fe00.2 )0.93
shown in the XRD patterns displayed in Fig. 5. From 300◦ C,
[Fe2.62Co0.38O4]. Due to the presence of an amorphous
the intensity of diffraction peaks of the cobalt–iron bcc al-
phase, it must be noted that these formulations are approxi-
loy (b) decreased in favor of the cobalt fcc phase (f). At
mative.
415◦ C only the fcc phase is present. According to the Fe–Co
The synthesis of these catalysts is believed to be based
phase diagram (24), this means that under heating the alloy
upon several steps as follows:
—the coprecipitation in boiling concentrated basic
medium of a cobalt iron mixed hydroxide (from the cor-
responding chlorides),

xCoCl2 · 6H2 O + yFeCl2 · 4H2 O

conc. KOH
−−−−→ Cox Fe y (OH)2(x+y) + 2(x + y)KCl;

—the dehydration of the hydroxide to an oxide in alka-

line medium,
OH−
Cox Fe y (OH)2(x+y) −−→ Cox Fe y Ox+y + (x + y)H2 O.

The formation of the metallic phase of the catalysts can

now be explained. Indeed, according to the literature, FeO
is known to disproportionate below 570◦ C (25):

4FeO → Fe3 O4 + Fe0 (1G ◦25 = −9.8 kcal).

However, the same phenomenon was reported for precipi-

tates or gels of Fe(OH)2 (26–28) according to

4Fe(OH)2 → Fe3 O4 + Fe0 + 4H2 O (1G ◦25 = −7.7 kcal).

In our case, disproportionation from the mixed hydroxide

CoxFey(OH)2(x+y) or from the mixed oxide CoxFeyOx+y is
possible. In both cases there is significant formation of Fe0.
Then the following redox reaction occurs to form Co0:

Co2+ + Fe0 → Co0 + Fe2+ E◦ (Fe2+/Fe) = −0.409 V

E◦ (Co2+/Co) = −0.28 V. FIG. 5. Evolution of the X-ray diffraction pattern of the dried catalyst
prepared according to Fig. 1, during heating under argon at (a) 300◦ C,
The reduction of Co2+ into Co0 is favored, which would (b) 350◦ C, (c) 370◦ C, (d) 400◦ C, (e) 415◦ C. s, b, and f refer to the spinel
explain why the alloy obtained is rich in Co ( ∼ 80% at Co). phase, the alloy of bcc structure, and the alloy of fcc structure, respectively.
 SYNTHESIS OF Fe–Co BASED METAL OXIDE COMPOSITES 69

FIG. 6. Scanning electron micrographs (SEM) of treated dried catalyst (left) and treated refluxed (right) catalyst prepared according to Fig. 1.

was reduced in iron content while simultaneously being The treated dried and treated refluxed catalyst can be
enriched in cobalt. The lattice parameter of 3.550(5) Å considered as metallic cobalt over cobalt doped magnetite:
observed for the fcc phase corresponds to a composition almost 17 wt% Co0/spinel in both cases. This amount of
of the alloy of the treated dried catalyst of Co00.95 Fe00.05 . metallic cobalt proved consistent with the concentration
The spinel phase is conserved and the lattice parameter range usually used in the Fischer–Tropsch synthesis (29).
of a = 8.394(5) Å remains almost unchanged. Once more, as for the nontreated samples, the formulations
The evolution of the refluxed catalyst is identical and of the treated dried and treated refluxed catalyst proved al-
the X-ray diffraction diagrams of treated dried and treated most identical. The differences of the specific surface areas
refluxed catalysts were equally similar. encountered for the nontreated samples were conserved
Thermal gravimetric analyses under argon were per- since for the treated dried and treated refluxed catalysts
formed on the treated dried and treated refluxed catalysts. these values were, respectively, 20 and 8 m2 g−1 cat .
The respective weight losses were only 0.25 and 0.21% They were also studied by scanning electron microscopy
(compared to 2.71 and 0.77% for the nontreated samples). (Fig. 6). The treated dried catalyst is very similar to the
This indicates that the hydroxides initially present decom- respective nontreated sample. The treated refluxed cata-
pose by heating at 415◦ C. In air, the weight increases were, lyst, unlike the nontreated sample, exhibited some grains
respectively, 6.68 and 6.22%. The calculated formula of the of octahedral shape. The crystallization of the amorphous
treated dried catalyst is (Co00.95 Fe00.05 )0.82 [Fe2.65Co0.35O4]; phase during heating at 415◦ C revealed by XRD and sub-
that of the treated refluxed catalyst is (Co00.95 Fe00.05 )0.74 sequent thermal gravimetric analysis was accompanied by
[Fe2.63Co0.37O4]. These were consistent with the fact that microstructural alteration.
the spinels remained almost unchanged by heating under Surface studies were carried out by XPS. The surface
argon. atomic ratios were calculated. For the dried catalyst, the
In contrast, heating under argon at 415◦ C led to par- surface atomic Co/Fe ratio is higher (0.5) than the Co/Fe
tial breaking down of the metallic phase (β equals 0.82 bulk ratio (0.42), indicating that at the surface the amount
for the treated dried catalyst and 0.74 for the treated re- of Co is higher than in the bulk. If it is considered that the
fluxed catalyst, respectively, compared to 0.95 for the dried phases present at the surface are the same as those in the
catalyst and 0.93 for the refluxed catalyst). The decom- bulk (alloy Co00.8 Fe00.2 and spinel Fe2.59Co0.41O4), this would
position of the hydroxides of the nontreated samples is mean that the surface of the dried catalyst is enriched in
thought to be responsible for the partial oxidation of the alloy with respect to the bulk. However, since no metallic
compounds. If no hydroxides remained in the nontreated element is detected by XPS, then Co0 and Fe0 are not in-
samples, the amount of metallic part would be conserved cluded in the surface Co/Fe ratio of 0.51. That cobalt rich
under heating. hydroxides are present at the surface of the dried catalyst
Under heating, the composition of the two phases was must be considered and equally that they shield metallic
changed. The metallic alloy was enriched in cobalt while cobalt, consequently protecting this layer from oxidation.
the spinel was enriched in iron according to This is consistent with the fact that hydroxides were de-
tected by XRD and thermal gravimetric analyses.
Ar
Fe0 + CoII The case of the refluxed catalyst is very similar. The Co/Fe
spinel −
−−→ FeII
spinel + Co .
0
415◦ C surface ratio is 0.84. No metallic element was detected.
70 CABET ET AL.

However, despite the high value of Co/Fe ratio at the sur-

face, the thermally treated sample did not contain as many
hydroxides as the dried catalyst. Therefore, in addition to a
cobalt rich hydroxide phase, a cobalt oxide phase has to be
considered.
For the treated dried and treated refluxed catalysts, the
calculation of the phase distribution at the surface is al-
lowed since these compounds have been shown to be free
of hydroxides. This time the two phases are Co00.95 Fe00.05
and Fe2.65Co0.35O4 (or Fe2.63Co0.37O4 for the treated re-
fluxed catalyst). The Co/Fe surface ratio of the treated dried
catalyst is 0.31 (lower than the bulk ratio). Metallic Co is
FIG. 7. Disproportionation curves of treated dried catalyst and
detected: the alloy (poor in Fe) is present at the surface
treated refluxed catalyst prepared according to Fig. 1 at various temper-
(and Fe0 is not detected). The calculation leads to the for- atures. (j) Treated dried catalyst at 220◦ C, (s) treated dried catalyst at
mula (Co00.95 Fe00.05 )0.50 [Fe2.65Co0.35O4] for the surface of the 240◦ C, (d) treated dried catalyst at 260◦ C, (h) treated refluxed catalyst at
treated dried catalyst. When compared with the bulk for- 220◦ C.
mula (where β = 0.82), the surface of the treated dried cata-
lyst was enriched in oxide phase (spinel) while the part of
sesses a metallic phase able to dissociate CO; thus reduction
the metallic phase is reduced. The same reasoning for the
is not necessary to produce metallic sites. The lower activ-
treated refluxed catalyst for which Co/Fe surface ratio is
ity of the reduced catalysts compared to that of the treated
0.44, gives (Co00.95 Fe00.05 )0.85 [Fe2.63Co0.37O4] for the surface
dried catalyst can be explained by the decrease of the spe-
composition. In this case the metallic content is both higher
cific surface area following reduction (13 m2 g−1 instead of
in the bulk of the treated refluxed catalyst and at the surface
20 m2 g−1). A more profound reduction of 8 h induced an
of the treated dried catalyst.
increase of the disproportionation activity, while the spe-
3. Reactivity of the Treated Dried and Treated Refluxed cific surface area remained stable. In this case, it can be
Catalyst in CO Hydrogenation suggested that the reduction generated metallic sites, ac-
tive for the disproportionation of carbon monoxide. This
3.1. CO disproportionation. As indicated in previous is not necessarily a favorable phenomenon since the active
works on Fe (30), Co (31, 32), or bimetallic Co–Cu (33) sites can be induced only by the spinel reduction. The sites
based catalysts, the carbon monoxide disproportionation for iron metal thus generated can easily carburize to the
reaction, detriment of the catalytic activity. Further, the diffraction
2CO → Csurface + CO2 , diagrams of the treated dried catalyst after an 8-h reduc-
tion treatment or not are similar but for d = 2.01–2.02 Å a
is a means of simulating the first step of the mechanism peak is present for the reduced catalyst. This peak signi-
of the Fischer–Tropsch synthesis, which can also be use- fies the presence of an iron rich metallic phase. The SEM
ful for comparing catalytic activities and here provides the examination reveals a breaking of the octahedral grains (see
means by which CO disproportionation on the treated dried Fig. 9).
and treated refluxed catalyst under various conditions was
studied.
Figure 7 presents the CO disproportionation curves at
220◦ C for the treated dried and treated refluxed cata-
lysts, and at 240 and 260◦ C for the treated dried catalyst.
Figure 8 relates the effect of the reduction (see experimen-
tal section) on the disproportionation activity of this latter
at 220◦ C.
From Fig. 7, at 220◦ C, the treated dried catalyst
(20 m2 g−1
cat ) would seem to be slightly more active than the
treated refluxed catalyst (8 m2 g−1). An increase in dispro-
portionation temperature induces higher activity. Figure 8
reveals that the reduction of the treated dried catalyst has
no effect upon the disproportionation activity (except for an
FIG. 8. Disproportionation curves of treated dried catalyst prepared
8-h reduction treatment). This phenomenon is surprising according to Fig. 1, at 220◦ C after various reducing treatments: (j) fresh,
for conventional Fischer–Tropsch catalysts but understand- (m) heated at 220◦ C under 10% H2/He, (4) maintained 4 h at 220◦ C under
able in our case since the treated dried catalyst already pos- 10% H2/He, (×) maintained 8 h at 220◦ C under 10% H2/He.
 SYNTHESIS OF Fe–Co BASED METAL OXIDE COMPOSITES 71

the reduced catalyst is less active than the nonreduced one

at 220◦ C (1.7% compared to 2.0%). At 220◦ C, the treated
refluxed catalyst remains inactive while the treated dried
catalyst converts 2.0% of CO. However, at 240 and 250◦ C,
the former becomes more active than the latter (5.0 and
9.0% of CO conversion for the former, and 4.1 and 6.9%
for the latter). As indicated in Fig. 7, with increasing tem-
perature, the CO conversion over the treated dried catalyst
increases (4.1% at 240◦ C and 6.9% at 250◦ C). This phe-
nomenon also occurs for reduced treated dried and treated
refluxed catalysts.
The molar selectivity for CO2 was high for the treated
dried catalyst and its reduced form, and reduced treated
refluxed catalyst, but decreased with increasing temper-
ature in favor of the hydrocarbon fraction: for reduced
FIG. 9. Scanning electron micrograph (SEM) of treated dried catalyst treated dried catalyst, the CO2 selectivity decreased from
prepared according to Fig. 1 after an 8-h reducing treatment at 220◦ C under
88.9 to 66.3% when the temperature was increased from 220
10% H2/He.
to 280◦ C, while the selectivity for hydrocarbons increased
3.2. Catalytic tests. The treated dried and treated re- from 7.7 to 33.0%.
fluxed catalyst were studied in the carbon monoxide hy- The treated refluxed catalyst behaved differently since
drogenation reaction for their capacity to produce C2–C4 selectivity for CO2 was only 40.4% at 240◦ C, while that for
olefins. the hydrocarbon was 56.4%. These results showed no im-
First, the catalysts were studied in their reduced form provement at higher temperatures. In all cases, the alcohol
(see Experimental). Then, they were tested without prior formation was low and alcohol selectivity never exceeded
reducing treatment and the CO/H2 mixture was introduced 3.4%. Notably, the higher alcohol detected was hexanol.
at 200◦ C under 1 MPa (see Experimental). Concerning the hydrocarbon distribution, once again, the
The catalytic results obtained are summarized in Table 2. treated dried catalyst and its reduced form, and reduced
The total CO conversions are consistent with the results treated refluxed catalyst, presented the same catalytic be-
obtained by the carbon monoxide disproportionation ex- havior. For reduced treated dried catalyst, the methane
periments. Indeed, in the case of the treated dried catalyst, fraction was 51.1% at 220◦ C, which decreased to 36.0%

TABLE 2
Catalytic Results of Treated Dried Catalyst, Treated Refluxed Catalyst, and the Corresponding Reduced Catalysts at Various
Temperatures (P = 1 MPa; CO/H2 = 1; GHSV = 2000 h−1)

Reduced treated Reduced treated Treated dried Treated refluxed

Catalyst dried catalyst refluxed catalyst catalyst catalyst

T (◦ C) 220 250 280 240 220 240 250 240 250

Total CO conversion (%) 1.7 9.3 24.0 8.8 2.0 4.1 6.9 5.0 9.0
Molar selectivities (%) into
CO2 88.9 83.2 66.3 80.6 84.2 78.5 68.8 40.4 55.8
Hydrocarbons 7.7 16.0 33.0 18.5 15.1 20.1 30.1 56.4 41.4
R–OH(<C6) 3.4 0.8 0.7 0.9 0.7 1.4 1.1 3.2 2.8
Hydrocarbons distribution (wt%)
C1 55.1 52.7 36.0 44.2 52.9 54.1 36.1 63.3 54.9
C= 2 8.8 6.9 3.4 6.4 8.7 7.8 4.9 4.9 3.9
C2 3.8 4.9 3.6 4.8 3.3 7.9 3.2 9.8 6.5
C= 3 9.6 9.1 5.0 9.6 8.7 7.9 5.9 6.8 7.6
C3 2.4 1.8 1.0 2.2 2.2 3.6 1.1 3.3 1.6
C= 4 5.2 4.9 2.5 6.9 5.3 4.5 3.3 3.7 4.2
C4 3.6 3.5 1.0 3.2 3.5 4.1 1.4 2.6 1.5
C+ 5 11.5 16.1 47.4 22.7 15.4 10.1 44.2 5.6 19.8
Total C2–C4 33.4 31.1 16.5 33.1 31.7 35.8 19.8 31.1 25.3
O/P 2.4 2.0 1.9 2.2 2.5 1.3 2.5 1.0 1.6

Note. Treated dried catalyst and treated refluxed catalyst have been prepared according to Fig. 1. The reduced form has been obtained by heating
the catalysts at 200◦ C under 10% H2/He. O/P, olefin/paraffin ratio in C2–C4 fraction.
72 CABET ET AL.

at 280◦ C. At this temperature, the C2–C4 fraction was also

lowered (from 33.4 to 16.5%) in favor of the C+ 5 fraction
(from 11.5 to 47.4%). Varying the temperature between 220
and 280◦ C induced a slight decrease of the olefin/paraffin
ratio (O/P) from 2.4 to 1.9. Reduced treated refluxed cata-
lyst, studied at 240◦ C, gave similar results to reduced treated
dried catalyst (CH4, 44.2%; C2–C4, 33.1%; C+ 5 , 22.7%; and
O/P = 2.2). For the treated dried catalyst at 220, 240, and
250◦ C, the variation in the catalytic results was identical
except for the value of the O/P ratio (2.5 at 250◦ C). The
C2–C4 fraction was of the order of 19.8% at 250◦ C with the
C+5 fraction representing 44.2% of the hydrocarbons.
In the case of treated refluxed catalyst, the catalytic re-
sults were different. At isoconversion of CO, the selectivity
toward methane was 54.9% at 250◦ C (compared to 52.7%
for reduced treated dried catalyst at 250◦ C and 44.2% at
240◦ C for reduced treated refluxed catalyst) and the C+ 5
FIG. 10. X-ray diffraction patterns of (a) treated dried catalyst after
catalytic test, (b) treated refluxed catalyst after catalytic test, (c) treated re-
fraction was smaller; the C2–C4 fraction remained satisfac- fluxed catalyst before catalytic test, (d) reduced treated dried catalyst after
tory. The chain growth probability (α), calculated in the catalytic test, and (e) reduced treated refluxed catalyst after catalytic test.
C2–C6 fraction, was the value of 0.44 for the treated re- The samples have been prepared according to Fig. 1 and the reduced form
fluxed catalyst at 240◦ C, while, at the same temperature, has been obtained by heating the catalysts at 200◦ C under 10% H2/He. s,
α equals 0.53 for the treated dried catalyst. This provided b, and f refer to the spinel phase, the alloy of bcc structure, and the alloy
of fcc structure, respectively.
confirmation that, at 240◦ C, the treated refluxed catalyst is
more favorable to the light hydrocarbons than the treated
dried catalyst. However, the most significant difference lies refluxed catalyst, while treated refluxed catalyst does not
in CO2 selectivity which is less than half for treated refluxed present the same catalytic results.
catalyst at 240◦ C than for the other three catalysts. The reduction of the spinel during the catalytic test of
The specific surface areas of the treated dried catalyst the reduced samples can be explained by the fact that the
and the treated refluxed catalyst after catalytic test were reduction before test partially reduced the spinel phase.
low (1 and 2 m2 g−1 cat , respectively). Some metallic Fe sites were created. These sites were pre-
The diffraction diagrams of the four catalyst after test cursors to very stable iron carbides and, under the reac-
are presented in Fig. 10, as well as that of fresh treated tion conditions, in the presence of Csurface coming from CO
refluxed catalyst. They were similar for treated dried cata- disproportionation displace the spinel reduction reaction
lyst, reduced one, and reduced treated refluxed catalyst af- according to
ter test. The peaks of the spinel phase are significantly de-
creased and a carbide phase was detected. The amount of
Cox Fe3−x O4 *
) (3−x)Fe +xCo (x ¿ 3−x, iron rich alloy)
0 0
this carbide phase increased with the duration of the cata- | {z }
lytic test, the temperature of reaction conditions, and iron ↓Csurface
carburization. (3 − x)/2Fe2 C + xCo0 .
For treated refluxed catalyst after test, the diffraction
peaks of the spinel (a cobalt doped magnetite with a = When treated refluxed catalyst is put in contact with
8.394 (5) Å) are almost unchanged with respect to those CO/H2 at 200◦ C, the catalyst activity, though weak, gen-
before test. Concerning the metallic phase, a fcc cobalt erated the formation of CO2 and H2O. The reducing ca-
phase was detected, as was, an iron rich Fe–Co alloy, pacity of syngas is then lowered and the reduction of the
as in the case of the treated dried catalyst after an 8-h spinel does not occur. For the treated dried catalyst, al-
reduction treatment (see Section 3.1) and a weak peak though treated under the same conditions, the spinel is re-
characteristic of Fe2C. Therefore, it seems that only the duced under test. In the case of the treated dried catalyst, it
metallic phase appeared to be affected by the reactivity must be remembered that the specific surface area is higher
under CO/H2. The treated refluxed catalyst is then the than for treated refluxed catalyst and the surface composi-
only catalyst whose spinel remains stable during CO hy- tions determined from XPS studies demonstrated that the
drogenation reaction. The other three catalysts studied re- surface of the treated dried catalyst was enriched in spinel
duce and carburize. These phenomena imply an explana- (β = 0.50) with respect to that of treated refluxed catalyst
tion for the similarities of the catalytic behaviors of the (β = 0.85). In brief, that more spinel is accessible to reducing
treated dried catalyst, reduced one and reduced treated gases can explain why the treated dried catalyst is reduced
 SYNTHESIS OF Fe–Co BASED METAL OXIDE COMPOSITES 73

under the reaction conditions, while the treated refluxed 3. Kugler, E. L., and Tauster, S. J., U.S. Patent 4,206,134 (1980) assigned
catalyst remains unchanged. to Exxon Research and Engineering Co.
4. Copperthwaite, R. G., Hutchings, G. J., Van der Riet, M., and
Woodhouse, J. R., Int. Eng. Chem. Res. 26, 969 (1987).
CONCLUSION 5. Kölbel, H., and Tillmetz, K. D., German Patent 2,507,647 (1976) U.S.
Patent 4,177,203 (1976) assigned to Ruhrchemie AG.
This work presents the preparation of cobalt–iron alloy/ 6. Bussenmeier, B., Frohning, C. D., Horn, G., and Klug, W., German
cobalt doped magnetite catalysts in which cobalt and iron Patent 2,518,964 (1976) assigned to Ruhrchemie AG.
are obtained in the metallic state. By heating at 415◦ C un- 7. Barrault, J., Renard, C., Yu, L. T., and Gal, J., “Proceedings for the
der inert gas, crystallographic restructurations occur in the 8th International Congress on Catalysis,” Vol. II, p. 101. Verlag Ed.,
Berlin, 1984.
compounds and the metallic phase becomes mainly com- 8. Jaggi, N. K., Schwartz, L. H., Butt, J. B., Papp, H., and Baerns, M.,
posed of elemental cobalt. Appl. Catal. 10, 347 (1985).
The methods by which the compounds were matured 9. Maiti, G. C., Malessa, R., and Baerns, M., Appl. Catal. 5, 151
during preparation induced morphological differences and (1983).
variations of both the specific surface areas and the metallic- 10. Barrault, J., and Renard, C., Appl. Catal. 14, 133 (1985).
11. Van Dijk, W. L., Niemantsverdriet, J. W., Van der Kraan, A. M., and
oxide phases distribution at the surface which influences the Van den Baan, H. S., Appl. Catal. 2, 273 (1982).
spinel stability toward reduction under the reaction condi- 12. Hutchings, G. J., and Boeyens, J. C. A., J. Catal. 100, 57 (1986).
tions. In fact, the greater the amount of the spinel phase at 13. Butt, J. B., Catal. Lett. 7, 83 (1990).
the surface of the catalyst, the more the oxide is exposed to 14. Hutchings, G. J., Van der Riet, M., and Hunter, R., J. Chem. Soc. Farad.
reducing gases. Then, as soon as metallic sites are created, Trans. 1 85, 2875 (1989).
15. Van der Riet, M., Copperthwaite, R. G., and Hutchings, G. J., J. Chem.
very stable iron carbides are formed which favor the spinel Soc. Farad. Trans. 1 83, 2963 (1987).
reduction. 16. Chen, H., and Adesina, A. A., Appl. Catal. A 112, 87 (1994).
The catalysts studied do not require any reduction step to 17. Colley, S., Copperthwaite, R. G., Hutchings, G. J., Loggenberg, P., and
be efficient in the Fischer–Tropsch synthesis, Further, when Van der Riet, M., Ind. Eng. Chem. Res. 27, 1339 (1988).
18. Colley, S., Copperthwaite, R. G., Hutchings, G. J., Terblanche, S. P.,
reducing treatments were performed before catalytic tests,
and Thackeray, M. M., Nature 339, 129 (1989).
the spinel phase of the compounds partially reduces and the 19. Malats i Riera, A., Pourroy, G., and Poix, P., J. Mat. Sci. Eng. A168,
catalysts easily carburize under the reaction conditions; this 245 (1993).
is to the detriment of the catalytic activity (high production 20. Malats i Riera, A., Pourroy, G., and Poix, P., J. Magn. Mater. 125, 125
of CO2). (1993).
An efficient treatment for protecting the catalyst from re- 21. Malats i Riera, A., Pourroy, G., and Poix, P., J. Magn. Mater. 134, 195
(1994).
duction and subsequent carburization was established: the 22. Malats i Riera, A., Pourroy, G., and Poix, P., J. Solid State Chem. 108,
compounds were heated at 200◦ C under nitrogen and then 362 (1994).
the inert atmosphere is substituted with the CO/H2 mixture. 23. Läkamp, S., Malats i Riera, A., Pourroy, G., and Poix, P., Eur. J. Solid
At 200◦ C, the catalyst activity, though weak, generated the State Inorg. Chem. 92, 159 (1995).
formation of CO2 and H2O and lowered the reducing ca- 24. Pearson, W. B., “Handbook of Lattice Spacing and Structures of
Metals.” Pergamon Press, 1964.
pacity of syngas. 25. Collongues, R., and Chaudron, G., C.R. Acad. Sci. 234, 728 (1952).
When the spinel is preserved, a selectivity to hydrocar- 26. Schikorr, G., Z. Elektrochem. 36, 65 (1929).
bons of 56.4% (31.1% C2–C4) can be obtained for a CO 27. Shipko, F. J., and Douglas, D. L., J. Phys. Chem. 60, 1519 (1956).
conversion of 5%. 28. Schrauzer, G. N., and Guth, T. D., J. Am. Chem. Soc. 98, 3508
(1976).
29. Hindermann, J. P., Hutchings, G. J., and Kiennemann, A., Catal. Rev.–
ACKNOWLEDGMENT Sci. Eng. 35, 1 (1993).
30. Kiennemann, A., Barama, A., Boujana, S., and Bettahar, M. M., Appl.
The authors thank Mr. P. Poix for his advice concerning the synthesis Catal. 99, 175 (1993).
of the composite materials. 31. Kiennemann, A., Ernst, B., and Chaumette, P., “Proceedings for the
27th International Symposium Automotive Technololgy and Automa-
tion,” Aachen, 333 (1994).
REFERENCES 32. Chaumette, P., Courty, P., Kiennemann, A., and Ernst, B., Top. Catal.
2, 117 (1995).
1. Tauster, S. J., Fung, S. C., and Garten, R. L., J. Am. Chem. Soc. 100, 33. Kiennemann, A., Boujana, S., Diagne, C., and Chaumette, P., “Pro-
170 (1978). ceedings, 10th International Congress on Catalysis, Budapest 1992”
2. Kugler, E. L., and Tauster, S. J., U.S. Patent 4,206,135 (1980) assigned (L. Guczi, F. Solymosi, and P. Tetenyi, Eds.), Akadémiai Kidaó,
to Exxon Research and Engineering Co. Budapest, 1993.