Kinetics of reduction of iron oxides by H2. Part II. Low temperature reduction of magnetite

Article abstract of DOI:10.1016/j.tca.2007.01.014

This study deals with the reduction of Fe₃O₄ by H₂ in the temperature range of 210-950 °C. Two samples of Fe₃O₄ produced at 600 and 1200 °C, designated as Fe₃O₄₍₆₀₀₎ and Fe₃O_4(1200), have been used as starting material. Reduction of Fe₃O₄₍₆₀₀₎ by H₂ is characterized by an apparent activation energy 'E_a' of 200, 71 and 44 kJ/mol at T < 250 °C, 250 °C < T < 390 °C and T > 390 °C, respectively. The important change of E_a at 250 °C could be attributed to the removal of hydroxyl group and/or point defects of magnetite. This is confirmed during the reduction of Fe₃O_4(1200). While transition at T ≈ 390 °C is probably due to sintering of the reaction products as revealed by SEM. In situ X-rays diffraction reduction experiments confirm the formation of stoichiometric FeO between 390 and 570 °C. At higher temperatures, non-stoichiometric wu?stite is the intermediate product of the reduction of Fe₃O₄ to Fe. The physical and chemical modifications of the reduction products at about 400 °C, had been confirmed by the reduction of Fe₃O₄₍₆₀₀₎ by CO and that of Fe₃O_4(1200) by H₂. A minimum reaction rate had been observed during the reduction of Fe₃O_4(1200) at about 760 °C. Mathematical modeling of experimental data suggests that the reaction rate is controlled by diffusion and SEM observations confirm the sintering of the reaction products. Finally, one may underline that the rate of reduction of Fe₃O₄ with H₂ is systematically higher than that obtained by CO in the explored temperature range.

Full text of DOI:10.1016/j.tca.2007.01.014

Thermochimica Acta 456 (2007) 75–88
Review
Kinetics of reduction of iron oxides by H2
Part II. Low temperature reduction of magnetite
A. Pineaua,b,c,d,∗, N. Kanaria,b,c,d, I. Gaballaha,b,c,d
a Laboratoire Environnement et Mine´ralurgie, Rue du Doyen M. Roubault, BP 40, 54501 Vandœuvre Cedex, France
b Centre National de la Recherche Scientifique, 3 Rue Michel-Ange, 75794 Paris Cedex, France
Ecole Nationale Supe´rieure de Ge´ologie, Rue du Doyen M. Roubault, BP 40, 54501 Vandœuvre Cedex, France
d Institut National Polytechnique de Lorraine, 2 Rue de la Foreˆt de Haye, 54501 Vandœuvre Cedex, France
c
´
Received 12 July 2006; received in revised form 10 January 2007; accepted 12 January 2007
Available online 20 January 2007
Abstract
This study deals with the reduction of Fe3O4 by H2 in the temperature range of 210–950 ◦C. Two samples of Fe3O4 produced at 600 and 1200 ◦C,
designated as Fe3O4(600) and Fe3O4(1200), have been used as starting material.
Reduction of Fe3O4(600) by H2 is characterized by an apparent activation energy ‘Ea’ of 200, 71 and 44 kJ/mol at T < 250 ◦C, 250 ◦C < T < 390 ◦C
and T > 390 ◦C, respectively. The important change of Ea at 250 ◦C could be attributed to the removal of hydroxyl group and/or point defects of
magnetite. This is confirmed during the reduction of Fe3O4(1200). While transition at T ≈ 390 ◦C is probably due to sintering of the reaction products
as revealed by SEM.
In situ X-rays diffraction reduction experiments confirm the formation of stoichiometric FeO between 390 and 570 ◦C. At higher temperatures,
non-stoichiometric wu¨stite is the intermediate product of the reduction of Fe3O4 to Fe.
The physical and chemical modifications of the reduction products at about 400 ◦C, had been confirmed by the reduction of Fe3O4(600) by CO and
that of Fe3O4(1200) by H2. A minimum reaction rate had been observed during the reduction of Fe3O4(1200) at about 760 ◦C. Mathematical modeling
of experimental data suggests that the reaction rate is controlled by diffusion and SEM observations confirm the sintering of the reaction products.
Finally, one may underline that the rate of reduction of Fe3O4 with H2 is systematically higher than that obtained by CO in the explored
temperature range.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Magnetite; Reduction; Hydrogen; Carbon monoxide; Low temperature
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2. Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3. Material and experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.1. Reduction of Fe3O4(600) by hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.2. Reduction of Fe3O4(600) by CO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.3. Reduction of Fe3O4(1200) by H2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.4. Analysis of experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
DOI of original article:10.1016/j.tca.2005.10.004.
Corresponding author at: Laboratoire Environnement et Mine´ralurgie, Rue du Doyen M. Roubault, BP 40, 54501 Vandœuvre Cedex, France.
∗
Tel.: +33 383 59 63 56.
E-mail address: Annye.Pineau@ensg.inpl-nancy.fr (A. Pineau).
0040-6031/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2007.01.014

76
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
1. Introduction
4. check the eventual reduction kinetic’s modifications during
the magnetite reduction and compare them with those of
hematite reduction.
As mentioned in part I [1], more than 2 t of carbon dioxide are
generated for the production of 1 t of iron metal. Low temper-
ature study of the reduction of iron oxides with hydrogen may
contribute to energy saving and low CO2 emission. However,
despite serious efforts [2–4], the cost of hydrogen production is
currently high.
This paper is focused on the reduction of magnetite with
hydrogen in the temperature range of 210–950 ◦C using ther-
mogravimetric analysis (TGA) and a furnace allowing the ‘in
situ’ X-rays diffraction. The objectives of this study are to
2. Literature review
Studies devoted to iron oxide reduction are abundant. Due to
the industrial importance of this topic, extensive work had been
done from the academic and applied point of view. However,
results are contradictories as will be demonstrated below. This
is because most authors use:
1. samples, either natural or pure, with different levels of impu-
rities;
1. compare the reduction kinetic of Fe3O4, with hydrogen or
carbon monoxide, with those of the reduction of hematite
performed in part I;
2. diverse experimental apparatus;
3. various temperature ranges;
4. distinctive gas mixtures with different levels of purity.
¨
2. confirm the presence of stoichiometric wustite, at tempera-
tures lower than 570 ◦C, as intermediate of the reduction of
magnetite to iron using H2;
3. explore the probable relation between the reactivity of mag-
netite and its initial physical characteristics by using two
samples issued from pure hematite reduction by H2–H2O
at 600 and 1200 ◦C;
Consequently, the comparison of the reaction rate and the
apparent activation energy ‘Ea’ obtained by different authors is
impractical. Fig. 1 illustrates these contradictions. For instant,
the reaction rate of the magnetite reduction, at the same tempera-
ture, can be different by several orders of magnitude. Moreover,
Table 1a shows that Ea can vary from 13.4 to 167 kJ/mol.
Table 1a
Bibliographic survey of the effect of magnetite material on the apparent activation energy (kJ/mol) of its reduction by H2 and suggested rate controlling process
[5–16]
Material
Temperature
range (◦C)
Activation
energy
Material, experimental conditions and kinetics controlling mechanism
Ref.
Natural and pure magnetite
200–280
234–450
60.6
77.7
Pure magnetite (0.5–3 ␮m), important interparticle diffusion phenomena
Thin (89 ␮m) magnetite, chemical reaction, film mass-transfer effects (H2,
water vapor) above 450 ◦C, star-like pores in iron (smaller at higher
temperatures)
[9]
[10]
400–500
400–500
65.6
72.1
Magnetite crystals, abnormal low reduction rate above 500 ◦C: phenomenon of
decrease in microporosity of the reduced layer
[11]
[6]
Natural magnetite (2% SiO2), the area of the inner unreacted sphere is the rate
controlling factor with the resistance diffusion of H2, H2O in metal
Natural magnetite, reaction order n = 1
Pure magnetite (99.95%), parabolic law of the degree of reduction α = (kt)0.5
Disk and sheet samples, reduction rate controlled at Fe3O4-Fe interface
Disk samples, reduction rate controlled at FeO–Fe interface, dense and stable
FeO formed an protective layer
390–580
325–600
400–585
585–900
86.0
69.3
61.4
13.4
[12]
[5]
[13]
[13]
910–1000
550–650
800–1000
≈ or >62
62.7
62.7 & 146.3
Disk samples, abnormal increase in reduction rate (␣ to ␥ Fe transition)
Briquettes of magnetite, phase boundary reaction
Briquettes of magnetite, 25–55% of reduction: Ea = 62.7 (phase boundary
reaction), later stages of reduction Ea = 146.3 (diffusion of O2− ions in iron)
Iron ore, Fe3O4 → Fe
Dense Fe3O4, H2 pressure = 1–40 atm: as the H2 pressure increases, the
reduction rate approaches a maximum rate: saturated surface with adsorbed
H2, phase boundary-controlled reaction (contracting sphere)
Magnetite pellets, controlling process at the hydrogen wu¨stite interface
(interfacial gas-solid reaction)
[13]
[8]
[8]
1200–1300
350–500
68.6
56.4
[14]
[7a]
Pellets
430–600
46
[15]
900–1000
700–900
46
Magnetite pellets, controlling process at the hydrogen wu¨stite interface
Magnetite pellets, controlling process at the hydrogen wu¨stite interface and
diffusion of iron in wu¨stite (solid sate diffusion)
[15]
[15]
≈167 or <167
Above 570
Above 570
1200–1300
1200–1300
57.1
54.1
68.6
61.5
Step Fe3O4 → Fe, pellets containing fine magnetite and coal
Step Fe3O4 → FeO, pellets containing fine magnetite and coal
Step Fe3O4 → Fe, iron ore
[16]
[16]
[14]
[14]
Step Fe3O4 → FeO, iron ore

A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
77
temperature [6,8,11,13,15,19]. They attributed this increase or
decrease of the reaction rate to different physical modifications
as summarized in Table 2.
Lien et al. [8] related the variations of the reduction rate to
imperfections of the crystal lattices either as default or impuri-
ties. Colombo et al. [20] demonstrated that micro-deformation
and defaults introduced by mechanical treatment of magnetite
are eliminated by a thermal treatment at 500 ◦C [21,22]. They
also mentioned that H2O associated to magnetite is removed at
about 600 ◦C. On the other hand, it is well known that the reac-
tion rate of the gas-solid interaction could increase or decrease
as solids are subjected to first or second-order transitions. This
phenomenon is known as Hedvall effect [23–25].
Thus, the reactivity of iron oxides could be affected by the
level of impurities, the extent of crystal defaults, the compo-
sition of the gas mixture and eventual phase transitions. In
order to avoid the effect of impurities relatively pure sample
of hematite had been used for the synthesis of magnetite used
in this study. The reducing gas mixture is relatively pure and
had a constant composition. On the other hand, two samples of
Fe3O4 produced at 600 and 1200 ◦C have been used as starting
material to verify the eventual effect of crystal defaults on their
reactivity.
Fig. 1. Arrhenius diagrams for the reduction of magnetite by hydrogen.
Table 1b compares the Ea of the magnetite reduction using
different gas mixtures as well as the controlling rate step as sug-
gested by different authors. Again, using the same gas mixture
for the magnetite reduction gives different apparent activation
energies that are more or less related to the explored temper-
ature range and the samples initial characteristics. As could
be expected, the use of different gas mixtures leads to differ-
ent values of Ea. However, the Ea values, for the reduction
of Fe3O4 by different reducing gas mixtures, are more or less
equivalents.
3. Material and experimental procedure
The magnetite used in this study is obtained by the reduction
of hematite supplied by Merck using a H2–H2O gas mixture
of adequate composition. Two samples are synthesized at 600
and 1200 ◦C and designed as Fe3O4(600) and Fe3O4(1200). Their
˚
crystal parameter is 8.395 A. The grain size of Fe3O4(600) is
about 1–2 ␮m and its apparent specific surface area is 0.7 m2/gr.
While the grain size of Fe3O4(1200) is 10–20 ␮m and its specific
surface area is about 0.1 m2/g.
Several authors have found that the rate of magnetite
reduction decreases or increases as function of the reaction
Table 1b
Bibliographic survey of the effect of gas mixture on the apparent activation energy (kJ/mol) of the reduction of Fe3O4 and suggested rate controlling process
[5,7b,12,14–18]
Gas mixture
H2–H2O–N2
Temperature range (◦C)
400–500
Activation energy
56.8
Material and experimental conditions and kinetics controlling mechanism
Ref.
[7b]
Dense Fe3O4, reduction rate approaches zero at magnetite/wu¨stite equilibrium,
phase boundary-controlled reaction (contracting sphere)
Synthetic magnetite + catalysts (SiO2, Al2O3, CaO, MgO, K2O), the reduction is
shifted towards higher temperatures relatively to pure magnetite
Commercial synthetic Fe3O4 + a few catalysts (SiO2, A12O3, CaO, MgO, K2O)
Slightly promoted magnetite: 2.2% of SiO2 + MgO + CaO + Al2O3 + · · ·
Heavily promoted magnetite: 7.8% of CaO + Al2O3 + SiO2 + MgO + · · ·
Magnetite pellets and B2O3, solid sate diffusion of ferrous ions is increased by
trivalent ions (bore) soluble in wu¨stite
H2
490–700
85.8
[12]
450–650
325–600
325–600
430–1000
46.5
88.3
117.7
46.0
[12]
[5]
[5]
[15]
CO
600–900
600–900
Above 570
78.2
64.4
69.6
Fe3O4 → FeO in (porous) Fe2O3 reduction, shrinking occur during reduction
Fe3O4 → FeO in Fe3O4 reduction, shrinking occur during reduction
Fe3O4 → FeO, pellets containing fine magnetite and coal, chemical reaction and
mixed control with Boudouard reactions
[17]
[17]
[16]
Above 570
1200–1300
1200–1300
1200–1300
76.9
65.7
69.7
73.6
FeO → Fe, pellets containing fine magnetite and coal
[16]
[14]
[14]
[14]
Fe3O4 → FeO, iron ore
FeO → Fe, iron ore
Fe3O4 → Fe, iron ore
CO + CO2
800–1050
121.2
50% CO2, step Fe3O4 → FeO, step-wise reduction of hematite pellets
[18]

78
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
Table 2
Variations of reaction rate in the range 600–700 ◦C during the reduction of Fe3O4 by H2
Raw material
Temperature range (◦C) of decreasing reaction rate and
explanation
Temperature range (◦C) of increasing
reaction rate and explanation
Ref.
Synthetic magnetite
Magnetite crystals
585–592: dense and stable wu¨stite forms a protective layer 600–900: porous wu¨stite formed is
over Fe3O4 surface non-protective
Above 500: decrease in microporosity of the reduced layer 600–900, for 20–40% of reduction; decrease
of reduction rate after 40% of reduction due
[13]
[11]
to the drop of ultra- and microporosity of the
reduction products
Briquettes of Fe3O4 sintered at 1200 ◦C
650–800: presence of a secondary reaction zone:
individual wu¨stite crystals into the iron band (solid-state
diffusion through the recrystallized iron surrounding the
individual wu¨stite crystals - crystal imperfections of iron
are removed). Certain impurities in the lattice of iron
oxides cause an increase in the number of imperfections
and eliminate the rate minimum
800–1000: during the earlier stages of
reduction (25–55%), phase boundary
reaction (rapid diffusion of ions through the
large number of crystals defects and
imperfections in the newly formed wu¨stite
and iron phases). In the later stages (85% of
reduction): solid-state diffusion
[8]
Magnetite concentrate
600–700: sintering and consolidation of a shell of
metallization advances from the outside of grain toward
the interior with increasing time, the metal jacket resists to
penetration by H2 and especially by water vapor. Sintering
is greatly enhanced by the presence of sulfides and silica
600–700 and 700–800 for later stages of reduction:
lagging phenomenon due to entrapped H2O, the factor
responsible for lagging has the nature of a protective
700–1000: restoration of diffusion
[6]
Magnetite, 4% SiO2, 4% Al2O3
700–800 for previous stages of reduction:
the trapped gaseous phase at high internal
pressures is able to burst the enclosing film,
[19]
envelope, either the metallic Fe itself or a nonmetallic film disruption of the metallic shell destroys the
metastable equilibrium conditions and the
reduction proceed again
Pure magnetite pellets
About 700–900: control of reduction is increasingly
interfered by diffusion of iron in wu¨stite and solid-state
reaction, a substantial retention (50%) of wu¨stite in the
iron occurs at 700 ◦C, the retained wu¨stite becomes
progressively coarser and its amount less from 700 to
900 ◦C
Increasing rate of reduction at all temperatures: the rate of solid-state diffusion of ferrous ions in wu¨stite is
increased by the presence of the soluble trivalent bore ions that is soluble in wu¨stite, the reduction proceeds
with the formation of discrete layers of reaction products. Contrarily, the rate of reduction is decreased when
trivalent (Al, Cr) ions insoluble in wu¨stite are present in wu¨stite
At 900 ◦C, islands of wu¨stite diminish in size
towards the outside of the specimen
[15]
[15]
Pure magnetite pellets with controlled
amount of B2O3
Thermogravimetric analysis tests were performed using
100 mg of the sample and a Cahn microbalance. The experi-
mental protocol and the details of this apparatus are described
in part I of this paper.
‘In situ’ reduction of magnetite was carried out using an X-
rays furnace. This furnace allows automatic data collection of
formed solids for θ = 0–60◦. The goniometer speed was adapted
to the reaction kinetics at different temperatures and varied from
θ = 0.25◦ to 1◦/min.
is about 500, 140, 11 and 3 min at 239, 258, 352 and 426 ◦C,
respectively.
The Arrhenius diagram traces the evolution of ln V versus 1/T
(◦K) as shown by Fig. 3. At temperatures lower than 250 ◦C, Ea
is about 200 kJ/mol. It decreases to about 71 kJ/mol for the tem-
perature range 250–390 ◦C. At temperatures higher than 390 ◦C,
Ea is equal to 44 kJ/mol.
The important change of the Ea value around 250 ◦C could
be attributed to
Raw samples and reaction products are systematically exam-
ined by X-rays diffraction (XRD), scanning electron microscope
(SEM) and eventually by chemical analysis.
1. a modification of the physical characteristic of the generated
iron in this temperature range;
2. the suppression or the modification of configuration of point
defects of Fe3O4(600) [26];
3. change of status of H2O associated to the magnetite [27–30];
4. low flow rate of the reducing gas mixture leading to starvation
and decrease of the reaction rate.
4. Results and discussion
4.1. Reduction of Fe3O4(600) by hydrogen
Fig. 2 groups the isotherms of this magnetite reduction
versus the reaction time at different temperatures. These
isotherms have the sigmoid form. For a reduction extent
(%R = weight loss × 100/27.64) of 50%, the reaction duration
The first hypothesis can be considered as several authors
[31–33] have observed a peak of internal friction of iron at about
250 ◦C. The second hypothesis is confirmed by several authors

A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
79
Fig. 2. Evolution of reduction extent vs. reaction time for the reduction of Fe3O4(600) with hydrogen.
[34–36] during the oxidation of magnetite. These authors used
differential thermal analysis ‘DTA’ and found an exothermic
peak at 250 20 ◦C during the oxidation of magnetite. They
contributed this peak to the formation of ␥-Fe2O3. However,
Behar and Collongues [37] found that disordered ␥-Fe2O3, pre-
pared at low temperatures, is well ordered at 250 ◦C.
The third hypothesis based on the consideration that the oxy-
gen of magnetite, prepared at T ≤ 600 ◦C, could be replaced by
OH− ions [27–29]. This had been confirmed by Colombo et al.
[20] as they found that magnetite losses its associated water at
temperatures higher than 600 ◦C. This could be related to our
experimentation for the unsuccessful preparation of magnetite at
400 ◦C. This is because the prepared magnetite using H2–H2O
gas mixture, at this temperature, is pyrophoric in contact with
air. Thus one should expect that reduction of magnetite, prepared
at 1200 ◦C, should not exhibit any anomaly at 250 ◦C due to the
elimination of protons and/or hydroxyl groups at T > 600 ◦C.
This will be examined in the next paragraph.
The fourth hypothesis cannot be considered as the increase of
the flow rate from 100 to 300 l/h does not augment the reaction
rate.
One may conclude that the first hypothesis can be excluded
as the anomaly at 250 ◦C is experienced for the reduction and
oxidation of magnetite. Thus this anomaly is probably due to
intrinsic property of magnetite. The second and third hypothesis
can be validated by considering that the association of protons
and/or the hydroxyl group to point defects that are modified at
about 250 ◦C.
The mathematical modeling of the experimental data of Fig. 2
is presented in Fig. 4. The models used are summarized in
Table 3. Eqs. (1)–(12) covers most of the probable mechanisms
that controls the gas–solid reactions. At temperatures lower than
400 ◦C, Fig. 4 suggests that the reaction rate is controlled by
phase boundary according to Eq. (3) or Eq. (4). At temperatures
higher than 390 ◦C, Eq. (11) seems to fit the experimental data.
Fig. 3. Arrhenius diagram for the reduction of Fe3O4(600) by hydrogen.

80
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
Fig. 4. Mathematical modeling of experimental data of Fig. 2.
This equation suggests that the reaction rate is controlled by
two-dimensional growth of nuclei.
tion of the reduction products occurs around 390 ◦C designed as
transition temperature ‘Tt’.
The morphological aspects of Fe3O4(600) and the evolution
of the reduced magnetite versus the temperature are grouped in
Fig. 5. It seems that there are no morphological modifications of
the reaction products reduced at temperatures lower than 390 ◦C
(pictures A–C). However, at temperatures higher than 390 ◦C,
the grain size of the reaction product increased demonstrating
their sintering. This suggests that there is a physical modifica-
To check this hypothesis, in situ X-rays diffraction exper-
imentation of the reduction of Fe3O4(600) by H2 in the
temperature range of 250–700 ◦C had been performed. Eq. (13)
summarizes the obtained results.
At temperature lower than 390 ◦C, magnetite is reduced
directly to iron. Between 390 and 570 ◦C, reduction of mag-
¨
netite passes by an intermediate that is the wustite before the
Table 3
Suggested mathematical modeling of reaction kinetics
Equation
Shape factor
Mechanism
kt = 1 − (1 − X)1/F
(1)
(2)
(3)
(4)
(5)
General equation [38]
p
kt = X
1
2
3
1
Phase-boundary-controlled reaction (infinite slabs)
Phase-boundary-controlled reaction (contracting cylinder)
Phase-boundary-controlled reaction (contracting sphere)
One-dimensional diffusion
kt = 1 − (1 − X)1/2
kt = 1 − (1 − X)1/3
kt = X2
kt = X + (1 − X)ln(1 − X)
(6)
(7)
2
3
Two-dimensional diffusion
Three-dimensional diffusion
kt = 1 − 3(1 − X)2/3 + 2(1 − X)
ꢀ
ꢁ
ꢂꢃ
2
3
kt = 1 −
X
− (1 − X)2/3
(8)
3
Three-dimensional diffusion (Ginstling–Brounshtein equation) [39]
2
kt = [1 − (1 − X)1/3
kt = [−ln(1 − X)]
]
(9)
(10)
3
1
Three-dimensional diffusion (Jander equation) [39]
Random nucleation; unimolecular decay law (first-order)
kt = [−ln(1 − X)]1/2
kt = [−ln(1 − X)]1/3
(11)
(12)
2
3
Two-dimensional growth of nuclei (Avrami–Erofeyev equation)
Three-dimensional growth of nuclei (Avrami–Erofeyev equation)
where, k = constant, t = reduction time, X = extent of reduction (X = 0 at the beginning of the reduction and X = 1 at the end of reduction), Fp = particle shape factor (1
for infinite slabs, 2 for long cylinders, and 3 for spheres).

A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
81
Fig. 5. Evolution of morphological aspects of the reduction products of Fe3O4(600) vs. the temperature (X = 10 000 1000).
¨
formation of iron. The crystal parameter of this wustite is
It seems that the decrease of Ea around 390 ◦C is related
˚
¨
about 4.33 A confirming the formation of stoichiometric FeO.
to the formation of stoichiometric wustite during the reduc-
At temperatures higher than 570 ◦C, the reduction of mag-
netite proceeds through the formation of non-stoichiometric
tion of Fe3O4(600) with H2. This behavior is similar to that of
the reduction of hematite with hydrogen around 420 ◦C [1].
To check wither this behavior is related to the nature of the
¨
wustite.

82
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
Fig. 6. Evolution of reduction extent vs. reaction time for the reduction of
Fe3O4(600) with CO.
reducing gas or to the intrinsic physical properties of magnetite,
reduction of Fe3O4(600) with CO was studied between 300 and
500 ◦C.
Fig. 7. Arrhenius diagram for the reduction of Fe3O4(600) by CO.
4.2. Reduction of Fe3O4(600) by CO
Fig. 6 shows the isotherms of the magnetite reduction with
carbon monoxide. The decrease and the increase of the sam-
ple weight is due to the reduction of magnetite in iron followed
(13)
Fig. 8. Mathematical modeling of experimental data of Fig. 6.

A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
83
Fig. 9. Morphological aspects of reduced Fe3O4(600) by CO at different temperatures (X = 7000 700).
Fig. 10. Evolution of reaction extent vs. reaction time for the reduction of Fe3O4(1200) with H2.

84
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
by the formation of iron carbide resulted from the decomposi-
tion of carbon monoxide. Only the beginning of the isotherm
(%R = 10–35) is employed for the construction of the Arrhenius
diagram as shown by Fig. 7. This figure indicates a decrease of
Ea from 150 to 64 kJ/mol at temperatures lower and higher than
428 ◦C, respectively.
Mathematical modeling of experimental data of Fig. 6 is
grouped in Fig. 8. At temperatures lower than 430 ◦C, the reac-
tion rate is controlled by two and/or three-dimensional nuclei
growth. At higher temperature, diffusion controls the reaction
rate. Fig. 9 exhibits some of the morphological aspects of the
reduction products. This figure suggests that sintering occurs at
temperatures higher than 430 ◦C.
4.3. Reduction of Fe3O4(1200) by H2
Fig. 10 groups the isotherms of reduction of magnetite at
different temperatures. This figure shows that at temperatures
lower than 600 ◦C, the isotherms have the traditional sigmoid
form. While at temperatures higher than 600 ◦C, the sigmoid
form of isotherms is replaced gradually by the parabolic one.
Fig. 11 groups the Arrhenius diagrams for different reduc-
tion extent of Fe3O4(1200) by H2. This figure suggests that for a
reduction extent of
1. 5–20%, Ea decreases from 77 to 27 kJ/mol for temperatures
lower and higher than 450 10 ◦C (Fig. 11A);
2. 20–60%, Ea decreasesattemperatureshigherthan390 ◦Cand
the reduction rate passes by a maximum at about 577 5 ◦C.
Starting at about 600 ◦C, the reaction rate decreases to attaint
a minimum at 766 5 ◦C. Between 770 and 950 ◦C, the
reduction rate increases with a discontinuity at around 900 ◦C
(Fig. 11B);
One may underline that the transition temperatures for the
reduction of magnetite by H2 and CO are comparables. It seems
that the kinetic modification is related to the intrinsic physical
properties of magnetite.
To confirm this hypothesis, reduction of Fe3O4(1200) by
hydrogen is performed between 210 and 950 ◦C.
Fig. 11. (A–D) Arrhenius diagrams for the reduction of Fe3O4(1200) by H2.

A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
85
3. 60–85%, the evolution of the reaction extent is identical to
that described in 2 (Fig. 11C).
One may underline that the comparison between Figs. 5
and 13 indicates that the grain size of Fe3O4(1200) is bigger than
that of Fe3O4(600). For both samples, the reaction products of the
magnetite reduction are agglomerated and begin at temperatures
higher than 400 ◦C.
Finally Fig. 14 compares the Arrhenius diagrams of the
reduction of hematite and Fe3O4(1200). At temperatures lower
than 400 ◦C, the rates of reaction of the two solids are almost
identical. However, at temperatures higher than 600 ◦C, the
behavior of the two solids is different.
One may underline that the kinetic modifications observed at
about 250 ◦C during the reduction of Fe3O4(600) does not exist
for both solids. This is probably due to absence of point defects
that are eliminated during the synthesis or the purification of
these solids at high temperature.
Fig. 11D compare the Arrhenius diagrams for the reduction
of Fe3O4(600) and Fe3O4(1200) for a reaction extent of 20–60%.
One may underline:
1. the modification of the reaction rate at 250 ◦C observed dur-
ing the reduction of Fe3O4(600) does not exist during the
reduction of Fe3O4(1200)
;
2. between 250 and 500 ◦C, the reaction rates of the two mag-
netite are comparables;
3. at temperatures higher than 600 ◦C, the reaction rate of
Fe3O4(1200) decreases while that of Fe3O4(600) increases.
Fig. 12 groups the mathematical modeling of experimental
data shown in Fig. 10 at 357, 478 and 878 ◦C. This figure sug-
gests that the reduction rate for 357 and 478 ◦C is controlled
by phase boundary according to Eqs. (3) and (4). At 878 ◦C, it
seems that Eq. (6) is the best fit for the experimental data. It
indicates that the reaction rate is controlled by diffusion.
Fig. 13 exhibits the evolution of the morphological aspects
of the reaction product of the Fe3O4(1200) reduction versus the
reaction temperature. The photos A–C of this figure show a
start like bursts. The size of these stars decreases as temperature
increaseswhiletheirnumberincreases. PhotosDandE, obtained
at temperatures higher than 700 ◦C, confirm the sintering of the
reaction products as compared to the initial sample.
4.4. Analysis of experimental results
The results of different series of experiments and the above
discussion confirm that the kinetic modification of the reduction
rate could be affected by point defects and/or crystal defaults
depending on the temperature range of the reduction of iron
oxides. This is because:
1. The kinetic modification of the reduction rate, around 400 ◦C,
occurs for both Fe3O4(600) and Fe3O4(1200)
.
2. Use of H2 or CO as reducing gas, does not affect the kinetic
modification at ≈400 ◦C.
Fig. 12. Mathematical modeling of experimental data of Fig. 10.

86
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
Fig. 13. Evolution of morphological aspects of the reduction products of Fe3O4(1200) by hydrogen at different temperatures (X = 1500 150).
3. Increasing the flow rate of the reducing gas does not change
this modification.
5. The agglomeration of the reduction products starts at tem-
peratures > 400 ◦C.
4. The kinetic modification at 250 ◦C occurs only during the
reduction of Fe3O4(600). This is probably due to the presence
of point defects or water in the magnetite structure prepared
at relatively low temperature.
6. The formation of stoichiometric wustite, during the reduc-
¨
tion of Fe2O3 and Fe3O4 is proved only in the temperatures
range 450–570 ◦C. This could be attributed to better crystal
structure due to the annealing of defaults.

A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
87
Fig. 14. Comparison of the Arrhenius diagrams for the reduction of Fe2O3 and Fe3O4 by H2.
7. The phenomenon of the minimum of the reaction rate, at
boundary reaction. At temperature higher than 430 ◦C for
Fe3O4(600) reduced with CO and higher than 650 ◦C for
Fe3O4(1200) reduced by H2, the controlling mechanism is
diffusion.
T ≈ 750 ◦C, occurs only during the reduction of Fe3O4(1200)
.
This tends to confirm that the severe sintering of the reduction
product is responsible of this behavior.
8. The reduction rate of Fe3O4 with H2 is systematically higher
than that obtained with CO in the explored temperature range.
9. SEM’s observations of the reduction products show their
gradual agglomeration and sintering as the temperature
increases. This leads to a decrease of Ea thus lowering the
thermal efficiency of the reduction process.
5. Conclusions
Results of the reduction of two samples of Fe3O4 prepared
at 600 and 1200 ◦C [Fe3O4(600) and Fe3O4(1200)] with hydrogen
in the temperature range of 210–950 ◦C leads to the following
conclusions:
From the practical point of view, one should emphasis that
low temperature reduction of iron oxides with H2 is handicapped
by the cost of hydrogen production. On the other hand, the
important decrease of Ea at temperatures higher than 400 ◦C sug-
gests that the best energy efficiency is around this temperature.
However, it is also in this temperature range that the pyrophoric
character of the reduction products is manifested.
While results of this work outline the important role of point
and linear defects on the reduction rate of iron oxides, one should
underline the lack of data about the behavior of point and linear
defects of iron oxides as function of temperature. The role exact
of the protons and hydroxyl groups in the magnetite structure
and their eventual association with the crystal defects is almost
unknown. Research in this area could boost the large-scale low
temperature reduction of iron oxides with hydrogen.
1. The Arrhenius diagram presents two modifications of Ea
at 250 and at 390 ◦C. The apparent activation energy Ea
decreases from 200 kJ/mol to about 71 kJ/mol between 250
and 390 ◦C. At higher temperatures Ea falls to 44 kJ/mol.
2. These kinetic modifications could be related to the intrinsic
physical properties of Fe3O4 and/or reduction products.
3. The kinetic modification at 250 ◦C is not observed during the
reduction of Fe3O4(1200). This is probably due to the anneal-
ing of point defects during the synthesis of this magnetite.
4. It seems probable that the kinetic modifications of the
reduction rate at about 380–420 ◦C, during the reduction
of Fe3O4(600), Fe3O4(1200) and Fe2O3, to the formation of
stoichiometric FeO between 390 and 570 ◦C and to the
agglomeration of the reduction products.
5. The observed minimum reduction rate of Fe3O4(1200) at about
760 ◦C seems to be related to severe sintering of the reduction
product.
Acknowledgments
6. At temperatures higher than 570 ◦C, the reduction of mag-
netite proceeds through the formation of non-stoichiometric
Part of this work was performed in ‘Laboratoire de Chimie du
Solide’oftheUniversityofNancy, FranceandtheMassachusetts
Institute of Technology ‘MIT’, Dept. of Material sciences. The
authors thank Prof. J. Szekely and Dr. Ch. Gleitzer for their help
and discussions. They also indebted to Mrs. Ch. Richard for her
kind help in technical and administration work.
¨
wustite.
7. Mathematical modeling of experimental data, at temper-
atures lower than 430 ◦C, suggests that the controlling
mechanism of the reaction rate is nucleation and/or phase

88
A. Pineau et al. / Thermochimica Acta 456 (2007) 75–88
[19] G. Specht Jr., C.A. Zaffe, TMS AIME 167 (1946) 237.
References
[20] U. Colombo, F. Gazzarrini, G. Lanzaveccha, Mater. Sci. Ing. 2 (2) (1967)
445.
[21] F. Gazzarini, G. Lanzavecchia, in: J.W. Mitchell, et al. (Eds.), Proceeding
of 6th ISRS, Wiley Interscience, 1969, p. 57.
[22] J. Fagherazzi, G. Lanzavecchia, Mater. Sci. Eng. 5 (2) (1970) 63.
[23] J.A. Hedvall, Z. Anorg, Allg. Chem. 61 (1924) 135.
[24] (a) H. Forestier, R. Lille, Comp. Rend. 204 (1937), p. 265, 1254;
(b) H. Forestier, R. Lille, Comp. Rend. 208 (1939) 891.
[25] H. Forestier, M. Daire, Akad. Der Wiss. Lit. 7 (1966) 705.
[26] C.G. Nestler, A. Faust, B. Kampfe, D. Lange, Kristall Technik 9 (1974)
k57.
[1] A. Pineau, N. Kanari, I. Gaballah, Thermochim. Acta 447 (2006) 89.
[2] Anonymous, Hydrogen production, in: US Climate Change Technology
Program—Technology Options for the Near and Long Term, November,
2003, p. 75. Edited in Technology options, 2005. http://www.climatscience.
gov/library/.
[3] L. Barreto, A. Makihira, K. Riahi, Int. J. Hydrogen Energy 28 (2003) 267.
[4] A. Carlson, Energy Policy 31 (2003) 951.
[5] P.G. Caceres, M.H. Behbehani, Appl. Catal. A: Gen. 109 (2–3) (1994) 211.
[6] M.C. Udy, C.H. Lorig, TMS AIME 154 (1943) 162.
[7] (a) W.M. McKewan, AIME Trans. 224 (1962) 387;
(b) W.M. McKewan, AIME Trans. 221 (1961) 140.
[27] S. Starke, Z. Phys. Chim. 42B (1939) 159.
[28] V.J. Arkharov, V.N. Bogosluvsky, F.S. Met, Metallov 3 (1965) 254.
[29] G. Fagherazzi, Chem. Ind. (Milan) 47 (1965) 75.
[30] M.L. Garcia-Gonzalez, P. Grange, B. Delmon, Proceeding of the 8th ISRS,
Go¨thenburg, June, 1976.
[31] W. Koster, L. bangert, R. Hahn, Arch. Eisenhuttenw. 25 (1954) 569.
[32] A.R. Rosenfield, Nature 196 (1962) 1083.
[33] P. Barrand, G.M. Leak, Nature 198 (1963) 279.
[34] R. Schrader, W. Vogelsberger, Z. Chim. 9 (1969) 354.
[35] M. Schneider, C.E. Beaulieu, Mem. Sci. Rev. Metall. 65 (1968) 869.
[36] B. Gillot, J. Tyranowicz, A. Rousset, Mater. Res. Bull. 10 (1975) 775.
[37] Behar, Collongues, C.R. Acad. Sci. 244 (1957) 617.
[38] J. Szekely, J.W. Evans, H.Y. Sohn, Gas–Solid Reactions, Academic press,
New York, 1976, p. 115, 232.
[8] H.O. Lien, A.E. El-Mehairy, H.U. Ross, J. Iron Steel Inst. 209 (1971) 541.
[9] G. Liberti, G. Servi, A. Gaulio, C. Ruffino, N. Pernicone, J. Therm. Anal.
6 (1974) 183.
[10] M.M. Al-Kahtany, Y.K. Rao, Ironmak. Steelmak. 7 (1980) 49.
[11] G.I. Tschufarov, et al., Acta Physicochim. (URSS) 3 (1935) 957.
[12] S.S. Bogdanov, B.D. Aleksi, I.G. Mitov, D.G. Klisurski, N.A. Petranovi,
Thermochim. Acta 173 (27) (1990) 71.
[13] M. Quets, M.E. Wadsworth, J.R. Lewis, AIME Trans. 218 (1960) 545.
[14] S. Sun, W.-K. Lu, ISIJ Int. 39 (1999) 123.
[15] V.J. Moran, A.E. Jenkins, JISI 199 (1961) 26.
[16] P.C. Coetsee, E.E. Pistorius, de Villiers, Miner. Eng. 15 (11 suppl. 1) (2002)
919.
[17] K.L. Trushenski, W.O. Philbrook, Metall. Trans. 5 (1974) 1149.
[18] T. Murayama, Y. Ono, Y. Kawai, Trans. Iron Steel Inst. Jpn. 18 (1978) 579.
[39] A. Ortega, Thermochim. Acta 284 (1996) 379.