Peculiarities of the iron reduction mechanism in Fe-Al-K system

Article abstract of DOI:10.1134/S0036024411010146

Temperature-programmed reduction was used in combination with measurements of magnetization to determine the peculiarities of iron reduction in the Fe-K-Al system. It was found that reduction by hydrogen proceeds with the formation of metallic iron through the stage of magnetite formation (Fe₃O ₄); the effective activation energies are 63 and 39 kJ/mol for the I and II stages, respectively. It was shown that substituting carbon oxide for hydrogen leads to iron reduction proceeding only to the stage of magnetite formation (E^Fe3O4 = 94 kJ/mol). The magnetite interacts with CO to produce carbide (presumably Hegge carbide Fe₂C). Iron reduction in the synthesis gas occurs with the preferential participation of hydrogen or carbon dioxide, depending on the rate of temperature rise. Pleiades Publishing, Ltd., 2011.

Full text of DOI:10.1134/S0036024411010146

ISSN 0036ꢀ0244, Russian Journal of Physical Chemistry A, 2011, Vol. 85, No. 1, pp. 55–61. © Pleiades Publishing, Ltd., 2011.
Original Russian Text © A.Yu. Krylova, P.A. Chernavskii, A.S. Lyadov, S.A. Sagitov, M.V. Krylova, S.N. Khadzhiev, 2011, published in Zhurnal Fizicheskoi Khimii, 2011, Vol. 85,
No. 1, pp. 61–67.
CHEMICAL KINETICS
AND CATALYSIS
Peculiarities of the Iron Reduction Mechanism
in Fe–Al–K System
A. Yu. Krylovaa, P. A. Chernavskiib, A. S. Lyadova, S. A. Sagitova, M. V. Krylovaa,
and S. N. Khadzhieva
a Topchiev Institute of Petrochemical Synthesis, Russian Academy of Sciences, Leninskii pr. 29, Moscow, 119991 Russia
b Faculty of Chemistry, Moscow State University, Moscow, 119991 Russia
eꢀmail: krylova@ips.ac.ru
Received March 9, 2010
Abstract—Temperatureꢀprogrammed reduction was used in combination with measurements of magnetizaꢀ
tion to determine the peculiarities of iron reduction in the Fe–K–Al system. It was found that reduction by
hydrogen proceeds with the formation of metallic iron through the stage of magnetite formation (Fe3O4); the
effective activation energies are 63 and 39 kJ/mol for the I and II stages, respectively. It was shown that subꢀ
stituting carbon oxide for hydrogen leads to iron reduction proceeding only to the stage of magnetite formaꢀ
tion (
= 94 kJ/mol). The magnetite interacts with CO to produce carbide (presumably Hegge carbide
E
Fe3O4
Fe2C). Iron reduction in the synthesis gas occurs with the preferential participation of hydrogen or carbon
dioxide, depending on the rate of temperature rise.
Keywords: iron, reduction, temperatureꢀprogrammed reduction.
DOI: 10.1134/S0036024411010146
INTRODUCTION
If continuous measurements of the sample’s magꢀ
netization are made during the performance of TPR,
we can obtain fundamentally new information about
the nature of the initial compound and those that form
as a result of reduction [6, 7]. In the present work,
TPR was used in combination with measurements of
magnetization to study the reduction peculiarities of
a nitrate blend of iron, potassium and aluminum
(Fe–K–Al), an analogue of the iron catalyst of Fisꢀ
cher–Tropcsh synthesis.
Apart from their active components, the ironꢀconꢀ
taining catalysts of Fischer–Tropsch synthesis also
contain various promoting additives (e.g., K, Si, and
Al oxides). Prior to synthesis, the catalysts are usually
subjected to activation (reduction by hydrogen or carꢀ
bon oxide). The character of the reduction of the iron
catalyst of Fischer–Tropcsh synthesis depends conꢀ
siderably on its composition [1]. Temperatureꢀproꢀ
grammed reduction (TPR) is usually used to obtain
data on the catalyst’s reducibility [2, 3]. This method
allows us not only to obtain the TPR spectrum of the
initial sample, but also to calculate the activation
energy of the reduction of individual compounds.
The basis of reduction activation energy determinaꢀ
tion is the dependence of the peak position in the
TPR spectrum on the heating rate [4]. To define the
activation energies from TPR data, we can use the
equation [5]
Three samples of the iron catalysts were studied by
TPR with the simultaneous measurement of magnetiꢀ
zation in situ: iron nitrate (the sample of comparison);
a blend of nitrates of iron and aluminum; and a blend
of nitrates of iron, aluminum, and potassium. Model
blends were prepared by mechanical mixing with subꢀ
sequent drying in air at 150°С
.
Figure 1 gives the kinetics of iron catalyst reduction
at 300 revealed in the appearance and rise of sample
°С
magnetization. Treating the iron nitrate hydrogen for
15 min with hydrogen caused the sample magnetizaꢀ
tion to rise to 2.5 rel. units before this factor stabilized
(Fig. 1). Calculations show that the iron fraction was
7.9% after reduction for 30 min. The magnetization of
the blend of nitrates of iron and aluminum increased
over the 30 min of catalyst treatment with hydrogen.
The obtained fraction of reduced iron was 6.9%, i.e.,
slightly less than in the reduction of pure iron nitrate.
The magnetization of the blend of the nitrates of iron,
ln(b/ E/RTmax + ln(AR/E) + С,
T2max) = –
where
ture of the peak maximum,
stant, is a preexponential factor, and
b
is the rise in temperature, Тmax is the temperaꢀ
is a universal gas conꢀ
is a constant.
R
А
C
The activation energy is determined from the slope
ratio of the line constructed in the coordinates
T2max) – 1/Tmax
ln(b/ .
55

56
KRYLOVA et al.
M, rel. units
6
II
4
2
0
I
III
0
30
, min
t
Fig. 1. Kinetics of the reduction of iron catalysts at 300°C; (I) Fe(NO ) · 9H O, (II) Fe(NO ) · 9H O + Al(NO ) · 9H O, (III)
3 3
2
3 3
2
3 3
2
Fe(NO ) · 9H O + Al(NO ) · 9H O + KNO . The area of heating the samples to 300°С is shown before the arrow; the behavior
3 3
2
3 3
2
3
of the samples under isothermal conditions for a period of 30 min is shown after the arrow.
M is magnetization.
aluminum, and potassium also increased over the energies calculated on the basis of these data have
30 min of catalyst treatment with hydrogen. The
reduction of this sample proceeded more intensively;
according to our calculations, ~20% of the initial iron
transitioned to the zeroꢀvalence state during the
experiment. The high reducibility of sample III is
probably associated with the influence of potassium,
which is an energetic promoter of iron.
Differential curves of magnetization are preꢀ
sented in Fig. 2. It can be seen that the curves conꢀ
tain two wellꢀformed peaks in the temperature
region of 550–750 K. The maxima of these peaks
shift toward higher temperatures with an increase in
the heating rate. The low temperature peak with a
the following values:
39.0 kJ/mol.
= 63.1 kJ/mol and ЕFe =
E
Fe3O4
In the literature, there are many data on the activaꢀ
tion energies of the processes of the reduction of
hematite into magnetite and of magnetite into metalꢀ
lic iron. These data, however, often correlate poorly
with one another. The activation energy of the aboveꢀ
mentioned processes depends substantially on the
applied method of analysis, the conditions under
which the experiment is performed, and the precursor
that is used and the size of its particles [9–12]. On
average, the activation energy of magnetite formaꢀ
tion (Fe3O4) is 100–150 kJ/mol, and the activation
energy of the process for reducing magnetite to
metallic iron is 50–90 kJ/mol. Our data on the values
of the activation energies of the twoꢀstage reduction
maximum at 570–620 K (~300–350°С) on the difꢀ
ferential curves of magnetization obtained for the
MB Fe–K–Al blend probably corresponds to the
reduction of Fe (III) into magnetite, while the high
temperature peak with a maximum at 650–730 K
probably corresponds to the reduction of magnetite
into metallic iron [8].
The differential curves of magnetization presented
in Fig. 2 thus allow us to calculate the activation enerꢀ
gies of the following processes:
process (
and EFe) thus indicate a decrease in
E
Fe3O4
the effective activation energies of both stages of
reduction when the MB Fe–K–Al model blend is
used (as opposed to the reduction of pure iron oxide,
Fe2O3).
It is well known that carbon oxide or synthesis gas,
rather than hydrogen, are often used for the reduction
of the iron catalysts in a slurry reactor [13].
2Fe2О3 + H2 = 2Fe3O4 + H2O (EFe O ),
3
4
Figure 3 presents the curves of magnetization for
the MB Fe–K–Al blend, obtained while performꢀ
ing TPR in carbon oxide at various rates of temperꢀ
ature increase (0.08, 0.14, and 0.28 K/s). They difꢀ
Fe3O4 + 4H2 = 3Fe + 4H2O (ЕFe).
Table 1 presents the data used to construct the
2
dependence ln(
b
/
) – 1/Tmax. The activation
Tmax
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A
Vol. 85
No. 1
2011

PECULIARITIES OF THE IRON REDUCTION MECHANISM
57
II
I
IV
III
400
500
600
700
800
, K
T
Fig. 2. Differential curves of magnetization for the reduction of MS Fe–K–Al in hydrogen, obtained at various rates of temperꢀ
ature increase: (I) 0.08, (II) 0.14, (III) 0.28, (IV) 0.42 K/s.
M, rel. units
4
3
2
1
0
II
I
III
400
450
500
550
600
650
, K
T
Fig. 3. Curves of magnetization for the reduction of MS Fe–K–Al in carbon oxide, obtained at various rates of temperature
increase, (I)–(III) (see Fig. 2).
fer considerably in shape from the curves obtained
earlier during catalyst reduction with hydrogen: a
rather sharp burst of magnetization (almost a vertiꢀ
cal line) associated with the reduction of threeꢀ
valence iron into magnetite is observed at ~500 K
The following reaction thus occurs during the
interaction of MB Fe–K–Al and carbon oxide:
3Fe2О3 + СО = 2Fe3O4 + СО2
(
).
Fe3O4
E
Table 1 gives the results from the performed measureꢀ
ments. The value of the activation energy calculated
on their basis is 94 kJ/mol (114–125 kJ/mol [14, 15]).
(
~230°С). On the differential curves, only one peak
is fixed.
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A
Vol. 85
No. 1
2011

58
KRYLOVA et al.
Table 1. Data for calculating the activation energy of reꢀ
duction with hydrogen, carbon oxide, and synthesis gas
(CO/H2 = 1/1) for MS Fe–K–Al by TPR
reducing agent. Just as when hydrogen is used, the
activation energy of the process for reducing the
model blend MB Fe–K–Al with carbon oxide proꢀ
ceeds faster than in the reduction of pure iron oxide
b
, K/s
T
max, К 103/
H2, peak I
1.835
T
max, K–1
Tm2ax
b/ )
–ln(
(
Fe2O3).
It should be noted that the intensive formation of
carbon dioxide is observed when the model blend MB
Fe–K–Al is reduced with carbon oxide (Fig. 4). The
first peak, with Тmax at ~530 K, obviously corresponds
to the release of СО2 that forms during the reduction
of hematite into magnetite. The slightly asymmetrical
shape of this peak can be probably be explained by the
partial interaction of the formed oxide Fe3О4 and carꢀ
bon oxide, with the subsequent formation of nonstoꢀ
ichiometric carbides. The Boudouard reaction
(2CO = C + СО2) evidently proceeds above 650 K
0.08
0.14
0.28
0.42
545
575
585
610
15.12730
14.67485
14.01619
13.69442
1.739
1.709
1.639
H2, peak II
0.08
0.14
0.28
0.42
588
642
657
707
1.701
1.558
1.522
1.414
15.27918
14.89529
14.24833
13.98956
(380°C).
A method for estimating magnetic particle size by
the shape of the hysteresis loop obtained in the coorꢀ
dinates of magnetization (the value of the magnetic
field) was proposed in [16, 17]. Figure 5 shows the hysꢀ
teresis curve obtained for the catalyst MB Fe–K–Al.
The narrowing of the loop in the region of the abscissa
axis (its “waist”) indicates the presence in the system
of superparamagnetic particles of magnetite with parꢀ
ticle sizes of less than 10 nm. Figure 6 presents the
curves of magnetization for the MB Fe–K–Al blend,
obtained while performing TPR in the synthesis gas (a
blend of carbon oxide and hydrogen) at various rates of
temperature increase (0.055, 0.08, 0.11, and 0.14 K/s).
The shape of these curves is considerably transformed
upon a change in the heating rate. At low rates of heatꢀ
CO
0.08
0.14
0.28
486
505
515
2.058
1.98
9.84669
10.483
1.942
11.2154
CO/H2 = 1/1
0.055
0.080
0.110
476
2.101
2.092
2.083
9.43041
9.81349
10.1403
478
480
The obtained data show the higher activation
energy of the process for reducing threeꢀvalence iron
into magnetite when carbon oxide is used as the ing (0.055 and 0.08 K/s), the shape of the magnetizaꢀ
M, rel. units
[CO2], rel. units
4
6
3
2
1
0
4
2
0
300
400
500
600
700
800
T
, K
Fig. 4. Dynamics of CO release during the reduction of MS Fe–K–Al by carbon oxide.
2
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A
Vol. 85
No. 1
2011

PECULIARITIES OF THE IRON REDUCTION MECHANISM
59
metallic iron interacts with carbon to form FeCх
iron carbide with a considerably lower Curie temꢀ
perature than metallic iron. If we calculate the actiꢀ
vation energy of the reduction of Fe(III) + H2 =
Fe3O4 on the basis of these data, we obtain a value of
336.4 kJ/mol, which considerably exceeds the value
H
, E
1.0
0.5
of
obtained during the reduction of the same
E
Fe3O4
catalyst by hydrogen (Table 1). It is entirely possible
that this high value of the activation energy of reducꢀ
tion by hydrogen can be explained by the influence of
the carbon oxide present in the blend, which is assoꢀ
ciated with the high adsorption of CO on transition
metals and their oxides.
−6
−
4
−2
2
4
M
6
, rel. units
The release of carbon dioxide was registered during
the reduction of the MB Fe–K–Al blend in synthesis
gas at various rates of temperature increase (Fig. 7).
Since the first wellꢀformed peak of СО2 was in all
cases detected in the region of hematite reduction
into magnetite (above 470 K or 200°С), we may
assume that it characterizes the process of Fe3O4 forꢀ
mation. The activation energy of the reduction of
hematite into magnetite, calculated on the basis of the
data on the carbon dioxide release rate, is 116.3 kJ/mol
(Table 2). This is close to the activation energy of the
formation of Fe3O4 from hematite during reduction
by pure CO.
−
1.0
Fig. 5. Hysteresis loop obtained for the MS Fe–K–Al
system.
tion curves before 600
the MS Fe–K–Al blend was reduced with hydrogen
(Fig. 2); above 600 , it recalls those obtained when
°С reminds those obtained when
°С
the MB Fe–K–Al blend was reduced with carbon
oxide (Fig. 3). At higher rates (0.11 and 0.14 K/s) the
shape of spectrum over the temperature interval is simꢀ
ilar to the curve obtained during reduction with carbon
oxide.
Peak
2 on each curve in Fig. 7 characterizes the
release of carbon dioxide, which is probably due to the
formation of iron carbides. A sharp reduction in the
magnetic properties of the sample is observed on the
curves of magnetization in this field (Fig. 6), since iron
We may assume that at low rates of temperature
increase, the twoꢀstage reduction of the initial iron
into metallic iron by hydrogen occurs (through the
stage of magnetite formation); above 600°С, the carbides are less magnetized than magnetite or iron at
M, rel. units
IV
3
III
2
1
0
II
I
300
400
500
600
700
, K
T
Fig. 6. Curves of magnetization for the reduction of MS Fe–K–Al in synthesis gas, obtained at various rates of temperature
increase: (I) 0.055, (II) 0.08, (III) 0.11, (IV) 0.14 K/s.
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A
Vol. 85
No. 1
2011

60
KRYLOVA et al.
[CO2], rel. units
25
20
15
10
5
III
I
II
0
350
400
450
500
550
600
650
, K
T
Fig. 7. Release of carbon dioxide during the reduction of MS Fe–K–Al in synthesis gas, obtained at various rates of temperature
increase: (I) 0.08, (II) 0.11, (III) 0.14 K/s.
the same temperature. The activation energy of iron material [18]. The Curie temperature calculated
from our thermomagnetic curve is close to the Curie
temperature of Hegg’s carbide Fe2C (540.3 K). We
may thus assume that this carbide is formed during
the interaction of MS Fe–K–Al blend and synthesis
gas.
carbide formation, calculated on the basis of these
data, is 66.4 kJ/mol.
It is well known that the Curie temperature, deterꢀ
mined from the dependence of magnetization on
temperature, is an individual property of a magnetic
REFERENCES
Table 2. Data for calculating the activation energy of reꢀ
duction with a blend of carbon oxide and hydrogen for MS
Fe–K–Al (peak 1) and the formation of iron carbide
(peak 2) by carbon dioxide release
1. E. T. Turkdogan and J. V. Vinters, Metall. Trans., No. 2,
3175 (1971).
2. N. W. Hurst, S. J. Gentry, A. Jones, and B. D. McNicol,
Catal. Rev., No. 24, 233 (1982).
b
, K/s
T
max, К 103/ max, K–1
T
–ln(
Tm2ax
b/ )
3. J. L. Lemaitre, in TemperatureꢀProgrammed Methods in
Characterization of Heterogeneous Catalysts, Ed. by
F. Delannay (Marcel Dekker, New York, 1984),
pp. 29–70.
Peak 1
475
0.055
0.080
0.110
0.140
2.105
2.088
2.053
1.972
9.4262
9.8177
4. H. E. Kissinger, Anal. Chem., No. 29, 1702 (1957).
479
5. O. J. Wimmers, P. Arnoldy, and J. A. Moulijn, J. Phys.
487
10.1693
10.4909
Chem. 90, 1331 (1986).
507
6. A. Y. Khodakov, Chu. Wei, and P. Fongarland, Chem.
Rev. 107, 1692 (2007).
Peak 2
556
7. P. A. Chernavskii, Kinet. Katal. 46, 634 (2005) [Kinet.
Catal. 46, 634 (2005)].
0.055
0.080
0.110
0.140
1.790
1.795
1.724
1.686
9.7411
10.1194
10.5188
10.8043
557
8. E. E. Unmuth, L. N. Schwartz, and J. B. Butt, J. Catal.
61, 242 (1980).
580
9. K. Piotrowsky, K. Mondal, T. Wiltowski, et al., Chem.
593
Eng. J. 131, 73 (2007).
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A
Vol. 85
No. 1
2011

PECULIARITIES OF THE IRON REDUCTION MECHANISM
61
10. G. Munteanu, L. Iieva, and D. Andreeva, Thermoꢀ 15. D. Dobovisek and B. Rjrousic, RudarskoꢀMetalur.
chem. Acta 291, 171 (1997).
Zbornik, No. 2, 127 (1968).
11. G. Munteanu, L. Iieva, and D. Andreeva, Thermoꢀ
16. N. S. Perov, N. Yu. Sudarikova, and A. A. Bagrets,
chem. Acta 329, 157 (1999).
J. Magn. (Korean Magn. Soc.)
8 (1), 7 (2003).
12. M. J. Tierman, P. A. Barnes, and G. M. Parkes, J. Phys.
17. P. A. Chernavskii, J.ꢀA. Dalmon, N. S. Perov, and
Chem. B 105, 220 (2001).
A. Y. Khodakov, in Oil and Gas Science and Technology
,
13. R. J. O’Brien, L. Xu, D. R. Milburn, et al., Topics
Catal., No. 2, 1 (1995).
Rev. IFP (2008), p. 24.
14. A. Pineau, N. Kanari, and I. Gaballah, Thermochem. 18. E. Z. Meilikhov, Pis’ma Zh. Eksp. Teor. Fiz. 69, 579
Acta 447, 89 (2006).
(1999) [JETP Lett. 69, 623 (1999)].
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY A
Vol. 85
No. 1
2011