Kinetics of reduction of iron oxides by H2. Part I: Low temperature reduction of hematite

Article abstract of DOI:10.1016/j.tca.2005.10.004

This study deals with the reduction of Fe₂O₃ by H₂ in the temperature range of 220-680 °C. It aims to examine the rate controlling processes of Fe₂O₃ reduction by H₂ in the widest and lowes

Full text of DOI:10.1016/j.tca.2005.10.004

Thermochimica Acta 447 (2006) 89–100
Kinetics of reduction of iron oxides by H2
Part I: Low temperature reduction of hematite
A. Pineau, N. Kanari, I. Gaballah∗
Mineral Processing and Environmental Engineering Team, LEM,1 CNRS2 UMR 7569,
ENSG,3 INPL,4 BP 40, 54501 Vandœuvre, France
Received 13 July 2005; received in revised form 6 October 2005; accepted 6 October 2005
Available online 23 November 2005
Abstract
This study deals with the reduction of Fe2O3 by H2 in the temperature range of 220–680 ◦C. It aims to examine the rate controlling processes of
Fe2O3 reduction by H2 in the widest and lowest possible temperature range. This is to be related with efforts to decrease the emission of CO2 in
the atmosphere thus decreasing its green house effect.
Reduction of hematite to magnetite with H2 is characterized by an apparent activation energy ‘Ea’ of 76 kJ/mol. Ea of the reduction of magnetite
to iron is 88 and 39 kJ/mol for temperatures lower and higher than 420 ◦C, respectively. Mathematical modeling of experimental data suggests that
the reaction rate is controlled by two- and three-dimensional growth of nuclei and by phase boundary reaction at temperatures lower and higher
than 420 ◦C, respectively.
Morphological study confirms the formation of compact iron layer generated during the reduction of Fe2O3 by H2 at temperatures higher than
420 ◦C. It also shows the absence of such layer in case of using CO. It seems that the annealing of magnetite’s defects around 420 ◦C is responsible
for the decrease of Ea.
The rate of reduction of iron oxide with hydrogen is systematically higher than that obtained by CO.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Reduction; Iron oxides; Hydrogen; Carbon monoxide; Temperature
1. Introduction
vantages of using pure hydrogen for the reduction of iron oxides
at low temperature.
Reduction of iron oxide is probably one of the most stud-
ied topics. This is due to the importance of iron and steel in
the current and future technologies. More than 2 tonnes of car-
bon dioxide are generated for the production of 1 tonnes of iron
metal. It is well known that CO2 increases the green house effect.
For these reasons, this study explores the advantages and disad-
From the industrial point of view, direct reduction of iron ores
withH2 ornaturalgasescouldhaveseveraltechnical advantages,
such as:
1. replacement of the expensive metallurgical coke as reducing
agent;
2. low carbon content of the produced iron;
3. generated gases are essentially composed of H2O and H2,
thus avoiding the release of CO and CO2 and by the same
token avoid the projected Ecotax.
∗
1
Corresponding author.
Laboratoire Environnement et Mine´ralurgie, rue du Doyen M. Roubault, BP
However, reduction of iron ores with hydrogen leads to com-
pact iron layers that could slow their reduction rate. Moreover,
in some cases the produced iron is pyrophoric. Finally, and in
spite of serious effort [1–3], the cost of hydrogen production is
currently high.
40, 54501 Vandœuvre Cedex, France.
2
Centre National de la Recherche Scientifique, 3 rue Michel-Ange, 75794
Paris Cedex, France.
3
´
Ecole Nationale Supe´rieure de Ge´ologie, rue du Doyen M. Roubault, BP
40, 54501 Vandœuvre Cedex, France.
4
This paper is focused on the reduction of hematite with hydro-
Institut National Polytechnique de Lorraine, 2 rue de la Foreˆt de Haye, 54501
Vandœuvre Cedex, France.
gen in the temperature range of 220–680 ◦C.
0040-6031/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2005.10.004

90
A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
2. Literature review
Reduction of hematite by hydrogen proceeds in two or
three steps, under and above 570 ◦C, respectively, via magnetite
¨
(Fe3O4) and wustite (Fe(1−x)O) according to the Bell’s Diagram
(Fig. 1) and the following equations:
3Fe2O3 + H2 → 2Fe3O4 + H2O
Fe3O4 + 4H2 → 3Fe + 4H2O
(i)
(ii)
(1−x)Fe3O4 + (1−4x)H2 → 3Fe(1−x)O + (1−4x)H2O (iii)
Fe(1−x)O + H2 → (1−x)Fe + H2O
(iv)
◦
¨
Generally, it is admitted that wustite is unstable below 570 C
under thermo-dynamic equilibrium. However, as shown by
Fig. 1. Bell’s Diagram for the Fe–C–O and Fe–H–O equilibrium at 1 atm.
¨
Fig. 1, wustitecouldbeanintermediateproductduringthereduc-
tion of hematite by hydrogen at temperature lower than 570 ◦C
under irreversible thermodynamic conditions.
Table 1
Apparent activation energy of the hematite reduction by hydrogen (kJ/mol)
Parameter
Temperature range (◦C)
Activation energy
Material and experimental conditions
Ref.
Raw material
460–500
460–500
460–500
450–700
450–700
450–700
450–700
56.8
72.3
89.4
246.0
93.0
Pure Fe2O3
[4]
[4]
[4]
[8]
[8]
[8]
[8]
Pure Fe2O3 heated to 850 ◦C
Hematite ore
Fe2O3 → Fe3O4, precursor of Fe2O3was FeOOHa
Fe3O4 → Fe, precursor of Fe2O3 was FeOOHa
Fe2O3 → Fe3O4, precursor of Fe2O3 was ferrihydritea
Fe3O4 → Fe, precursor of Fe2O3 was ferrihydritea
162.0
104.0
Pellets
700–925
1000–1150
<500
109.9
18.0
30.1 and 56.4
Iron ore: 80% Fe2O3, 18% FeO, globular pellets
Iron ore: 80% Fe2O3, 18% FeO, globular pellets
Ferric oxide pellets, degree of reduction (Fe2O3 → Fe3O4) < 0.01,
electrical conductivity
[12]
[12]
[13]
Step Fe2O3 → Fe3O4
290–480
250–450
450–700
450–700
250–610
300–900
124.0
139.2
246.0
162.0
106.0
89.1
Fe2O3 → Fe3O4, 67% H2 (H2–Ar), ≈3% H2Oa
Fe2O3 → Fe3O4, hydrogen–argon mixture (10% H2)a
Fe2O3 → Fe3O4, the precursor of Fe2O3was FeOOHa
Fe2O3 → Fe3O4, the precursor of Fe2O3 was ferrihydritea
Fe2O3 → Fe3O4, 5% H2 in Hea
[6]
[7]
[8]
[8]
[9]
[10]
a
Fe2O3 → Fe3O4 5% H2 in N2
Step Fe3O4 → Fe
290–480
450–700
450–700
250–610
300–900
172.0
93.0
104.0
54.0
Fe3O4 → Fe0, 67% H2 (H2–Ar), ≈3% H2Oa
Fe3O4 → Fe, the precursor of Fe2O3 was FeOOHa
Fe3O4 → Fe, the precursor of Fe2O3 was Ferrihydritea
Fe3O4 → Fe, 5% H2 in Hea
[6]
[8]
[8]
[9]
[10]
a
70.4
Fe3O4 → Fe, 5% H2 in N2
Fe3O4 → Fe
250–450
250–450
580 and 720
77.3
85.7
72.0
Fe3O4 → FeO, hydrogen–argon mixture (10% H2)a
FeO → Fe, hydrogen–argon mixture (10% H2)a
Wu¨stite → Fe
[7]
[7]
[11]
Presence of water vapor
in the reducing gas
mixture
465, 485 and 505
52.7
Pure Fe2O3 with 4% water vapor (w.v.)
[5]
465, 485 and 505
465, 485 and 505
465, 485 and 505
290–480
55.2
58.9
53.1
124.0
172.0
Pure Fe2O3 with 7.5% w.v.
Fe2O3ore with 2% w.v.
[5]
[5]
[5]
[6]
[6]
Fe2O3 ore with 5% w.v.
Fe2O3 → Fe3O4, 67% H2 (H2–Ar), ≈3% H20a
290–480
Fe3O4 → Fe metal, 67% H2 (H2–Ar), ≈3% H2Oa
Impurities
460–500
460–500
460–500
109.9
107.8
129.2
Fe2O3 and MgO
Fe2O3 and A12O3 or In2O3 or Li2O
Fe2O3 and TiO2
[4]
[4]
[4]
CO
700–1150
72.3
Fe2O3 → Fe, iron pellets cont. 97.2% of iron oxides
[12]
a
Temperature-programmed reduction.

A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
91
Table 2
Controlling mechanisms of the iron oxides reduction
Gas
H2
Solid
Temperature range (◦C)
Kinetics controlling mechanism
Ref.
Pure Fe2O3
Pure Fe2O3 with 2.5–7.5% water
vapor
460–500
460–500
Topochemical reaction at the interface gas/solid l − (1 − X)1/3
Increased rate of reduction attributed to a hydrogen spill-over
effect: the chemisorbed hydrogen atoms activate the surface
migration
[4]
[4]
Pure Fe2O3 + foreign metal
oxides (A12O3 . . .) or hematite
ore
Pure Fe2O3 + fresh metal
powders
460–500
Retard of the reduction kinetics due to structural factor:
topochemical reaction and different mechanism involving the
mixed oxides (FeAl2O4 . . .) formed at the surface of Fe2O3
Increased rate of reduction
[4]
460–500
[4]
[5]
Pure Fe2O3 with water vapor
465, 485 and 505
The rate increases with water vapor content between 2.5 and
7.5%: this is attributed to H spill-over. The rate determining
step is identified to be the desorption of H2O
Three-dimensional nucleation model according to
Avrami–Erofeyev
Self-catalyzed nucleation (autocatalysis): nuclei catalyze
further nuclei formation, due to branching of nuclei or to the
catalytic role of the Fe metal in H2 dissociation. H2O might
assist during the acceleration by assuring fast hydrogen
spill-over
67% H2/Ar
␣-Fe2O3 (Merck, pro analysis)
290–480
290–480
[6]
[6]
␣-Fe2O3 (Merck, pro analysis)
Addition of 3% H20 to the
reducing gas
5% H2/He
5% H2/N2
Fe2O3 (99.98%, Aldrich)
250–610
250–610
The prereduction step (Fe2O3 to Fe3O4) was described by an
“nth-order” expression
[9]
[9]
Fe2O3 (99.98%, Aldrich) or
Fe3O4 (99.997%, Alfa
Chemicals)
Iron oxide prepared by
precipitation of
The reduction of Fe3O4 followed a nucleation or autocatalytic
mechanism. Metal nuclei formed are believed to dissociate
and activate dihydrogen molecules leading to autocatalysis
Fe2O3 → Fe3O4 unimolecular model [a first order reaction:
r = A exp[−E/RT] (1 − α)]
300–900
300–900
<500
[10]
[10]
[13]
Fe(NO3)3·9H2O
Fe3O4 → Fe two-dimensional nucleation according to
Avrami–Erofeyev model
H2
Ferric oxide pellets
For degree of reduction = x < 0.01: thin layer at the pellet
surface has been reduced: chemisorption of H2 on the oxygen
vacancies
CO, H2
Iron ore: 80% Fe2O3, 18%
FeO–globular pellets
700–1150
Chemical reaction: the advance of reaction front during
reduction in 1 − (1 − R)1/3
[12]
One of the most important parameters of the hematite reduc-
tion is the apparent activation energy as it defines the reactor
dimensions and the energy consumption. Literature survey indi-
cates that its value depends on the starting raw material, nature
of reducing gas, temperature range, reaction step, presence of
water vapor in the gas mixture, impurities, physical shape,
etc.
Table 1 confirms that the apparent activation energy (Ea)
depend on these factors. Ea varies from 18 to 246 kJ/mol. Such
range integrates the kinetic of different controlling steps and
indirectly the reactivity of solids. Moreover, Table 1 suggests
that Ea obtained by reducing iron oxides with CO is comparable
to that of H2 [4,12]. This is also confirmed by other studies [14].
Table 2 groups the controlling mechanism suggested by different
Table 3
Suggested mathematical modeling of reaction kinetics
Equation
Shape factor, FP
Mechanism
Ref.
[15]
kt = 1 − (1 − X)1/F
kt = X
(v)
(vi)
(vii)
(viii)
(ix)
General equation
P
1
2
3
1
2
3
3
Phase-boundary-controlled reaction (infinite slabs)
Phase-boundary-controlled reaction (contracting cylinder)
Phase-boundary-controlled reaction (contracting sphere)
One-dimensional diffusion
Two-dimensional diffusion
Three-dimensional diffusion
kt = 1 − (1 − X)1/2
kt = 1 − (1 − X)1/3
kt = X2
kt = X + (1 − X) ln(1 − X)
(x)
kt = 1 − 3(1 − X)2/3 + 2(1 − X)(xi)
kt = [1 − (2/3)X] − (1 − X)2/3 (xii)
Three-dimensional diffusion (Ginstling–Brounshtein equation)
[16]
2
kt = [1 − (1 − X)1/3
kt = [− ln(1 − X)]
]
(xiii)
(xiv)
(xv)
3
1
2
3
Three-dimensional diffusion (Jander equation)
[16]
[16]
[16]
[16]
Random nucleation; unimolecular decay law (first-order)
Two-dimensional growth of nuclei (Avrami–Erofeyev equation)
Three-dimensional growth of nuclei (Avrami–Erofeyev equation)
kt = [− ln(1 − X)]1/2
kt = [− ln(1 − X)]1/3
(xvi)
Where k is the constant, t the reduction time (min), X the extent of reduction (X = 0 at the beginning of the reduction and X = 1 at the end of reduction) and FP the
particle shape factor (1 for infinite slabs, 2 for long cylinders and 3 for spheres).

92
A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
authors for the iron oxides at different temperature ranges. It is
clear that the controlling mechanism is related to the starting
raw material, gas composition, temperature range, etc. Table 3
presents the suggestedequationsused forthemathematicalmod-
eling of the kinetic data of the reaction. Eqs. (v)–(xvi) cover most
of the probable mechanisms that control the gas–solid reactions.
These equations will be used for the mathematical modeling of
our kinetic data. The best fitting equation(s) will be chosen to
eventually determine the controlling mechanism.
3. Material and experimental procedure
Hematite used in this study had a Fe2O3 content higher than
99.8% supplied by Merck. Impurities are essentially traces of
Ca2+, Mg2+, Cl− and SO42−. The specific surface area of Fe2O3
was 0.51 m2/g. The porosity volume is 3.3 cm3/g. The SEM indi-
cates that the grain size of the hematite is about 1–2 ␮m.
Thermogravimetric (TG) tests were performed using 100 mg
of sample and a CAHN microbalance (Fig. 2). It has a sensibility
of 20 ␮g. The sample is reduced in gold boat having a surface
area of about 2.2 cm2. The sample was first heated in a nitrogen
flow to the desired temperature. N2 was then replaced by pre-
heated H2 at the initial reduction time. The average velocity of
H2 was about 3 cm/s and the flow rate was about 100 1/h. A data
acquisition system records the sample weight and temperature
as a function of time. Hematite, as well as the reduction prod-
ucts, were examined by X-rays diffraction (XRD) and scanning
electron microscope (SEM).
Fig. 2. Thermogravimetric apparatus. (1) Quartz reactor, (2) thermocouple, (3)
sample boat, (4) furnace and (5) counterweight.
Fig. 3. Evolution of reduction extent percentage vs. reaction time for the reduction of hematite with hydrogen.

A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
93
4. Results
Fig. 3 groups the isotherms of hematite reduction versus the
reaction time at different temperatures. As can be observed, most
of the isotherms have a plateau at about 11% of reduction extent
percentage ‘R%’ that corresponds to the reduction of hematite
to magnetite. One may underline that the time for full reduction
of hematite is about 1500, 40 and 5 min at temperatures 237, 349
and 472 ◦C, respectively. This suggests that the apparent activa-
tion energy ‘Ea’ is relatively important. Fig. 4 is the Arrhenius
diagram obtained by using the data of Fig. 3.
This figure shows that the value of Ea for the reduction of
hematite to magnetite is about 76 kJ/mol. The reduction of mag-
netite to iron is characterized by an apparent activation energy of
88 and 39 kJ/mol at temperatures lower and higher than 417 ◦C,
respectively.
Fig. 5 groups the mathematical modeling of the experimental
data during the reduction step of Fe3O4–Fe with H2 at 233, 330
and 472 ◦C. The models tested for the temperatures generates
correlation coefficient that varies from 0.9921 to 0.9993. It is
almost arbitrary to choice between these models. However, it
seems that the best fitting of data obtained at 233 and 330 ◦C
is that of Eqs. (xv) and (xvi). These equations suggest that the
reaction is controlled by the two- and three-dimensional growth
of nuclei. This mechanism is in agreement with the value of Ea
at temperatures < 417 ◦C. At 472 ◦C, the best fitting of experi-
mental data is obtained using Eq. (vi). This equation suggests
that reduction of magnetite at this temperature is controlled by
phase boundary reaction (infinite slabs).
Fig. 4. Arrhenius diagram for the reduction of hematite by hydrogen. (a) Reduc-
tion extent ≤ 11% and (b) reduction extent ≥ 11%.
Fig. 6 exhibits the morphological aspects of the reduction
products at different temperatures using the SEM. Pictures A–C
indicate that the average grain size is almost constant. At tem-
peratures higher than 425 ◦C, pictures D and E show gradual
sintering of grains. This could explain the modification of Ea
and the change in the mechanism controlling the reaction rate.
However, Eq. (vi) could be explained by gas starvation. For this
Fig. 5. Mathematical modeling of experimental data of Fig. 3. (—) Correlation coefficients > 0.995 and (- - -) correlation coefficients < 0.995

94
A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
Fig. 6. Effect of reduction temperature on the morphological aspects of reduction products using hydrogen (X = 6000 600).
reason, reduction of hematite is studied using a gas mixture com-
posed of 10% H2 and 90% N2.
than Tt. Again, it is clear that sintering of particles is performed
at temperature higher than 415 ◦C.
Fig. 7 traces the evolution of R% as function of time at dif-
ferent temperatures for the reduction of hematite with the above
mentioned gas mixture. At 586 ◦C, the curve shows two plateaus
at R% of about 11 and 33. These values correspond to the reduc-
tion of Fe2O3 into Fe3O4 and Fe3O4 into Fe(1−x)O, respectively.
At lower temperatures, only the plateau at 11% reduction
extent is observed that corresponds to the reduction of hematite
to magnetite. Fig. 8 shows the Arrhenius plot of Fig. 7 data.
Again, a change of Ea occurs at about 415 ◦C confirming results
ofthehydrogenreductionofhematite(Fig. 4). Ontheotherhand,
the value of Ea decreases from about 103 to 36 kJ/mol as the
reduction temperature exceeds that of the transition temperature
(Tt) 415 ◦C. Mathematical analysis of Fig. 7 data confirms that
the reaction rate, at temperature lower than Tt, is controlled by
the two- and three-dimensional growth of nuclei and by phase
boundary reaction at higher temperature.
It is well known that the hydrogen reduction of iron oxides
produces a compact iron layer [6,8,10,12]. Consequently, the
decrease of the apparent activation energy could be attributed
to the formation of this compact layer of iron that slow down
the reaction rate leading to lower Ea. To check this hypothesis,
experimentation of the hematite reduction with carbon monox-
ide is performed.
Fig. 10 shows the evolution of the reaction extent ver-
sus time for the reduction of hematite by CO at different
temperatures. One may underline that the weight increase cor-
responds to the formation of iron carbide. Only data of the
initial reaction rate are used to calculate the evolution of
the reaction rate as function of 1/T (K). Results are grouped
in Fig. 11. This figure shows that Ea for the reduction of
Fe2O3 to Fe3O4 is about 114 kJ/mol. While the reduction
of magnetite to iron is characterized by an Ea of 114 and
40 kJ/mol at temperatures lower and higher than 439 ◦C, respec-
tively.
Fig. 9 reflects the morphological aspects of hematite reduc-
tion with the gas mixture H2–N2 at temperature lower and higher

A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
95
Fig. 7. Evolution of reduction extent percentage vs. reaction time for the reduction of hematite with hydrogen–nitrogen.
Fig. 12 shows the evolution of the morphological aspects of
the reaction products of the reduction of Fe2O3 by CO. It seems
that sintering of the particles is avoided in this case. One may
come to the conclusion that the compact iron layer could be
responsible for the modification of the reaction rate expressed
by the decrease of Ea. However, Fig. 11 also shows a transi-
tion temperature of about 439 ◦C. Consequently, the hypothesis
about the influence of the compact layer of Fe0 on decrease of
Ea can be abandoned. Thus, it is probable that other physico-
chemical phenomenon is responsible of this evolution.
To check this hypothesis, a X-rays furnace is used for the in
situ reduction of hematite with hydrogen. This furnace allows
data collection of formed solids for θ = 0–60◦. The goniometer
speed was adapted to the reaction kinetics at different temper-
atures and varied from θ equal 0.25–1◦/min. Details of this
furnace are grouped in Ref. [17].
Eq. (xvii) summarizes the obtained results, from X-ray
diffraction, using gas mixtures containing H2 and N2. While
results at temperatures lower than 450 ◦C and higher than
570 ◦C are expected, that obtained in the temperature range
◦
¨
of 450–570 C revels the presence of wustite. Moreover,
¨
wustite’s crystal parameter, in this temperature range, was
˚
about 4.33 A corresponding to almost stoechiometric FeO.
¨
One may underline that the presence of wustite at tempera-
tures lower than 570 ◦C have been detected by other authors
[18,19].
(xvii)
Consequently, the decrease of Ea around Tt could be
explained by the different paths during the magnetite reduction.
At temperatures lower than Tt reduction of magnetite produces
directly metallic iron. Between Tt and 570 ◦C, reduction of
¨
Fe3O4 leads to the formation of wustite and metallic iron. At
Fig. 8. Arrhenius diagram for the reduction of hematite by H2–N2. (a) Reduction
extent ≤ 11% and (b) reduction extent ≥ 11%.
temperatures higher than 570 ◦C, reduction of magnetite to Fe0

96
A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
Fig. 9. Effect of reduction temperature on the morphological aspects of reduction products using a gas mixture H2–N2.
¨
proceeds through the intermediate formation of wustite. One
5. Discussion
¨
may correlate Tt to the formation of wustite. This imposes the
question why Fe(1−x)O appears in this temperature range? Ten-
As indicated in Table 1, Ea depend on the raw material and
its purity and the physical state of this raw material. Moreover,
for the same solid, different experimental conditions leads to
different values of Ea. Several authors confirm these conclusions
as indicated in Table 4. Results obtained in this work agreed
with those of Refs. [9,10,20,23,24]. The suggested mechanisms
of reduction of hematite match with those of Refs. [6,9,10].
On the other hand, the evolution of Ea versus the temperature
shows, in some cases, an important decrease of the apparent
activation energy around a transition temperature as shown in
Table 5. Several authors suggested that such evolution is due to
change of the gas–solid reaction from the chemical regime at low
temperature to mixed or diffusion regime at high temperatures
[20,21,26–29,39,40].
tative response will be discussed in Section 5.
Fig. 13 groups the data of the reduction of Fe3O4 by H2
and H2 + N2. This figure confirms that the values of Ea and Tt
are comparables. Consequently, the hypothesis of gas starvation
can be neglected. However, Tt for the reduction of Fe3O4 by
CO is equal to 439 ◦C. One may underline that this is probably
due to the cementation of Fe0. Consequently, the data used for
the calculation of Ea as well as the transition temperature are
limited to 20% of the reaction extent generated by the reduction
of magnetite to iron using CO. As a result, the real reaction
rate of the reduction of Fe3O4 by CO could be higher than the
obtained values. Consequently, the value of Ea should be higher
than the reported one and Tt should be lower than the calculated
one. Fig. 14 compares data obtained by the reduction of Fe3O4
by H2 and CO. This figure emphasizes that the rate of reduction
of magnetite with hydrogen is higher than that obtained using
CO.
However, this evolution can also be attributed to Hedvall
effect [33–35] that relates the change of reactivity of solids
to first and second order transitions. In some cases Tt corre-
sponds to a magnetic transformation such as Curie temperature

A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
97
Fig. 10. Evolution of reaction extent percentage vs. reaction time for the reduction of hematite with CO.
(575 ◦C for Fe3O4 [36]) or phase transition such as that of ␣-
1. sample weight from 10 to 100 mg;
2. concentration of H2 in the reducing gas mixture from 10 to
100%;
3. nature of the reducing gas: H2 or CO;
4. gas velocity from about 2 to 4 cm/s;
5. boat’s material (Cu or Au);
Fe to ␥-Fe (906 ◦C) [37,38]. However, most of Tt of Table 5
does not correspond to any physical transitions of the concerned
solids.
To elucidate the decrease of Ea around the transition tempera-
ture, the following experimental parameters have been modified:
6. experimental method TGA and in situ XRD.
Performing the above mentioned modifications of experimental
parametersleadstoalmostthesameresults. InsituXRDbetween
◦
¨
300 and 700 C revels the presence of stoechiometric wustite
at about 450 ◦C. In this case, one may suggest that Tt is more
or less independent of all the above-mentioned experimental
¨
parameters and could be related to the presence of wustite or
other physico-chemical phenomenon.
Naeser and Scholz [41] examined the effect of grinding on
the reduction of hematite by hydrogen. They observed that the
reduction of the ungrounded sample starts at 420 ◦C. While
that of the ground one starts at 340 ◦C. This is probably due
to increase of the solids’ defects through mechanical treat-
ment. Papanastassiou and Bitsianes [27] studied the oxidation
of magnetite to hematite and found that Ea decreases sharply
at temperature higher than 400 ◦C. Schrader and Vogelsberger
[42] indicate the presence of two exothermic peaks around 250
and 400 ◦C during the oxidation of magnetite to hematite. They
attributed these peaks to the oxidation of Fe3O4 to ␥-Fe2O3
and transformation of ␥-Fe2O3 to ␣-Fe2O3, respectively. They
also indicate that the second peak’s temperature is variable. It
depends on the initial hematite sample’s state of deformation.
Fig. 11. Arrhenius diagram for the reduction of hematite by CO. (a) Reaction
extent ≤ 11% and (b) reaction extent ≥ 11%.

98
A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
Fig. 12. Effect of reduction temperature on the morphological aspects of reduction products using CO (X = 7000 700).
Table 4
Ea in relation with the starting material, temperature range and the reducing gas mixture
Ref.
Step
Temperature range (◦C)
Gas
Ea1 (kJ/mol)
Tt (◦C)
Ea2 (kJ/mol)
[20]
Fe2O3 → Fe
400–1000
400–1000
400–1050
400–900
250–400
250–400
250–400
254–298
210–254
290–360
220–683
337–604
265–482
220–683
337–604
265–482
H2
H2
H2–N2
H2
H2
H2 or CO
H2
H2
CO
H2
H2
H2–N2
CO
H2
H2–N2
CO
51.0
96.1
62.3
61.4
62.7
108.7
108.7
125.4
125.4
137.9
75.9
94.8
114.1
88.0
575
575
550
550–600
38.0
48.9
64.0
13.4
[20]a
Fe2O3 → Fe
[21]
[22]
[23]
[23]
[23]
[24]
[24]
[25]
This work
This work
This work
This work
This work
This work
Fe2O3 → Fe
Fe3O4 → Fe
Fe2O3 → Fe2O3 + Fe3O4 + Fe
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe2O3 → Fe3O4
Fe3O4 → Fe
417
415
439
39
36
40
Fe3O4 → Fe
103.0
114.0
Fe3O4 → Fe
a
Natural single crystals (Sweden).

A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
99
Table 5
Evolution of Ea around Tt for different solid–gas reactions
a
c
Ref.
[26]
[20]
Starting solid
WO2.9
Process
Method
TGA
Product
W
Eal
Tt (◦C)b
Ea2
Reduction
94.0
625
37.6
Fe2O3
Fe2O3
Reduction
Reduction
TGA
TGA
Fe
Fe
96.1
50.2
575
575
48.9
37.6
[22]
[27]
Fe3O4
Reduction
TGA
Fe
61.4
550
13.4
d
Fe3O4
Fe3O4
Oxidation
Oxidation
TGA
TGA
Fe2O3
Fe2O3
164.7
143.4
390
420
10.0
10.5
e
[28]
[29]
[30]
[31]
[32]
Fe2O3
Fe
Nb
Cu2O
Cu
Reduction
Chlorination
Oxidation
Oxidation
Sulfidation
TGA
TGA
TGA
TGA
TGA
Fe3O4
FeCl2
NbO2
CuO
61.2
96.1
∼96.0
∼54.3
71.1
441
∼400
400
7.3
∼21.0
∼21.0
<54.3
34.3
∼765
Cu2S
∼400
a
Ea1: apparent activation energy (Ea) for T < Tt.
Tt: transition temperature.
b
c
d
e
Ea2: Ea for T > Tt.
Missouri–magnetite concentrate.
Taconite concentrate.
Gillot et al. [43] confirmed the existence of the two peaks and
show that their temperature is function of the grains’ size. They
indicate that the temperature of the transition ␥-Fe2O3 to ␣-
Fe2O3 decreases ◦from 470 ◦C, for solids of average grain size
˚
of 600 A, to 380 C for solids having an average grain size of
˚
1400 A.
All these authors suggest that the reactivity of solids is influ-
enced by their imperfections. Fagherazzi and Lanzavecchia [44]
have confirmed this hypothesis as they indicated that magnetite
annealed at 500 ◦C gets rid of its defects and part of its microde-
formation.
Fox and Soria-Ruiz [45] studied the effect of dislocations
movement on the kinetics of thermal decomposition of freshly
cleaved calcite crystal. They found two patterns of behavior at
temperatures lower and higher than 427 ◦C. They attributed this
change to the fast annealing of crystal’s defects at temperatures
higher than 427 ◦C.
Fig. 14. Arrhenius diagrams for the reduction of hematite by H2 and CO for a
reduction extent ≥ 11% corresponding to the reduction of Fe3O4 → Fe0.
Dabosi [46], Venturello et al. [47], Hu [48] and Leslie [49]
observed a reduction of the dislocation density of metallic iron
in the temperature range of 400–450 ◦C. They attributed this
phenomenon to the recovery and/or re-crystallization of iron.
In absence of studies on behavior of dislocations in iron
oxides in function of temperature and with respect to results
of several authors [41–49], one may speculate that Tt could be
related to the annealing of the magnetite defects around 420 ◦C.
¨
Consequently, the formation of stoechiometric wustite, as inter-
mediate in the reduction of Fe3O4 to Fe0, is probably due to
upgrading of the crystal structure.
6. Conclusions
Results of the reduction of iron oxides with hydrogen in the
temperature range of 200–680 ◦C leads to the following conclu-
sions:
Fig. 13. Arrhenius diagrams for the reduction of hematite by H2 and H2–N2 for
a reduction extent ≥ 11% corresponding to the reduction of Fe3O4 → Fe0.

100
A. Pineau et al. / Thermochimica Acta 447 (2006) 89–100
1. The reduction of hematite in magnetite by H2 is characterized
by an apparent activation energy of about 76 kJ/mol.
[9] M.J. Tierman, P.A. Barnes, G.M.B. Parkes, J. Phys. Chem. B 105 (2001)
220.
[10] H.Y. Lin, Y.W. Chen, C. Li., Thermochim. Acta 400 (1–2) (2003) 61.
[11] A. Jess, H. Depner, Steel Res. 69 (3) (1998) 77.
[12] E. Mazanek, M. Wyderko, Met.-Odlew. Met. (1974) 55.
[13] M.H.A. Elhamid, M.M. Khader, A.E. Mahgoub, B.E.L. Anadouli, B.G.
Ateya, J. Solid State Chem. 123 (2) (1996) 249.
[14] (a) N. Rudenko, S.T. Rostovtsev., Izv. V.U.Z. 4 (1959) 3;
(b) O.L. Koxtelov, S.T. Rostovtsev., Sta. 3 (1965) 209.
[15] J. Szekely, J.W. Evans, H.Y. Sohn., Gas-Solid Reactions, Academic
Press, New York, 1976, p. 115, 232.
[16] A. Ortega, Thermochim. Acta 284 (1996) 379.
[17] N. Gerard, J. Phys. E Sci. Instrum. 5 (1972) 524.
[18] V.P. Romanov, P.A. Tatsienko, L.F. Checherskaya, G.P. Basantsev, V.D.
Checherskii, Izv. Akad. Nauk SSSR, Neorganischeskie Materialy 7 (3)
(1971) 450.
2. The reductionpath of magnetitetoiron isfunction of thereac-
tion temperature. At temperatures lower than 420 ◦C, Fe3O4
is reduced directly to iron. At 450 < T < 570 ◦◦C, magnetite
¨
and wustite are present with iron. At T > 570 C, magnetite
¨
is fully reduced to wustite before its reduction to Fe.
3. The apparent activation energy for the reduction of Fe3O4 by
H2 decreases from 88 to 39 kJ/mol for temperatures lower
and higher than 420 ◦C.
4. Mathematical modeling of experimental data suggest that
the controlling mechanism is the two- and three-dimensional
growth of nuclei and by phase boundary reaction at higher
temperature.
[19] V.P. Romanov, L.F. Checherskaya, P.A. Tatsienko, Phys. Stat. Sol. A 15
(1973) 721.
5. The decrease of Ea around 420 ◦C could be attributed to the
annealing of defects of the crystalline structure of magnetite.
6. The reduction rate of magnetite with hydrogen is higher than
that obtained with CO.
[20] N.B. Gray, J. Henderson, TMS-AIME 236 (1966) 1213.
[21] W.M.M.C. Kewan, TMS-AIME 218 (1960) 2.
[22] J.M. Quets, M.E. Wadsworth, J.R. Lewis, TMS-AIME 218 (1960) 545.
[23] U. Colombo, F. Gazzarrini, G. Lanzaveccha, Mater. Sci. Ing. 2 (1967)
125.
[24] V.A. Roiter, V.A. Yuza, A.N. Kuzentsov, Zhur. Fiz. Khim. 25 (8) (1951)
960.
[25] D. Dobovisek, B. Korousic, Rudarsko-Metalurski Zbornik (2) (1968)
127.
7. The morphological study of the reduced samples of iron
oxides by H2 shows agglomeration of the reaction product at
temperatures higher than 420 ◦C. Reduction of these oxides
with CO does not generate compact iron layer.
[26] L.C. Dufour. Thesis, Dijon, 1965.
In spite of results obtained in this work, low temperature reduc-
tion of iron oxides with hydrogen is handicapped by lack of data
about the behavior of point and linear defects of iron oxides
as function of temperature that control its reactivity, the high
energy consumption and cost of hydrogen production.
[27] D. Papanastassiou, G. Bitsianes, Met. Trans. 4 (1973) 487.
[28] J.P. Hansen, G. Bitsianes, T.L. Joseph, B.F. Coke Oven and Raw Mater.
Conference (1960) 185.
[29] R.J. Fruehan, Met. Trans. 3 (1972) 2585.
[30] E.A. Gulbransen, K.F. Andrew, J. Electrochem. Soc. A (1958) 105.
[31] K. Hauffe, P. Kofstad, Z. Elektrochem. 59 (1955) 399.
[32] A. Galerie, M. Caillet, J. Besson, in: J. Wood, et al. (Eds.), 8th ISRS,
Gothenburg 14-19/6/1976, Plenum Press, 1977.
Acknowledgments
[33] J.A. Hedvall, Z. Anorg, Allgem. Chem. 61 (1924) 135.
[34] (a) H. Forestier, R. Lille, Comp. Rend. 204 (1937) 265, 1254;
(b) H. Forestier, R. Lille, Comp. Rend. 208 (1939) 891.
[35] H. Forestier, M. Daire, Akad. der Wissenschaftenundder Lit. (7) (1966)
705.
[36] A. Chakraborty, J. Magn. Magn. Mater. 204 (1999) 57.
[37] I. Gaballah, F. Jeannot, Ch. Gleitzer, C.R. Acad. Sci. Ser. C 280 (1975)
697.
Part of this work was performed in ‘Laboratoire de Chimie
du Solide’ of the University of Nancy, France. The authors thank
Dr. Ch. Gleitzer for his help and discussions. They also indebted
to Mrs. Ch. Richard for her kind help in technical and adminis-
tration work.
[38] I. Gaballah, C. Gleitzer, J. Aubry, in: J. Wood, et al. (Eds.), Reactivity
of Solids, Plenum Press, 1977, p. 391.
References
[39] E. Wicke, Chemie–Ingenieur Technic 29 (1957) 305.
[40] L. Bonnetain, G. Hoyant, Les Carbones, Masson, Paris, 1965, tome II,
290 pp.
[41] G. Naeser, W. Scholz, Kolloid-Zeitschrift 156 (1) (1958) 1.
[42] R. Schrader, W. Vogelsberger, Z. Chem. 9 (9) (1969) 354.
[43] B. Gillot, J. Tyranowicz, A. Rousset, Mater. Res. Bull. 10 (1975) 775.
[44] G. Fagherazzi, G. Lanzavecchia, Mater. Sci. Eng. 5 (2) (1970) 63.
[45] P.G. Fox, J. Soria-Ruiz, Proc. R. Soc. Lond. A 314 (1970) 429.
[46] F. Dabosi. Thesis, Paris, 1962.
[47] C. Venturello, C. Antonione, F. Bonaccorso, Trans. AIME 227 (1963)
1433.
[48] H. Hu, Recovery and Recrystallization of Metals, Interscience, New
York, 1963, p. 311.
[1] Anonymous, Hydrogen production, in: US Climate Change Technology
Program—Technology Options for the Near and Long Term, November,
2003, p. 75.
[2] L. Barreto, A. Makihira, K. Riahi, Int. J. Hydrogen Energy 28 (2003)
267.
[3] A. Carlson, Energy Policy 31 (2003) 951.
[4] M.V.C. Sastri, R.P. Viswanath, B. Viswanathan, Adv. Hydrogen Energy
2 (1981) 1279 (Hydrogen Energy Prog. 3).
[5] B. Viswanathan, R.P. Viswanath, M.V.C. Sastri, Trans. Indian Inst. Met-
als 32 (4) (1979) 313.
[6] O.J. Wimmers, P. Arnoldy, J.A. Moulijn, J. Phys. Chem. 90 (1986) 1331.
[7] G. Munteanu, L. Ilieva, D. Andreeva, Thermochim. Acta 291 (1997)
171.
[8] G. Munteanu, L. Ilieva, D. Andreeva, Thermochim. Acta 329 (2) (1999)
157.
[49] W.C. Leslie, Recrystallization, grain growth and textures, Am. Soc. Met.
(1967) 59.