Fuel Processing Technology 86 (2004) 33 – 47

www.elsevier.com/locate/fuproc

Reduction of iron oxide in carbon monoxide

atmosphere—reaction controlled kinetics
K. Mondal a, H. Lorethova b, E. Hippo a,
T. Wiltowski a,b, S.B. Lalvani c,*
a
Department of Mechanical Engineering and Energy Processes, Southern Illinois University,
Carbondale, IL 62901, USA
b
Coal Research Center, Southern Illinois University, Carbondale, IL 62901, USA
c
Paper Science and Engineering, Miami University, Oxford, OH 45069, USA
Accepted 1 December 2003

Abstract

Thermogravimetric analysis (TGA) of the reduction of Fe2O3 in a continuous stream of 100% CO

was conducted at temperatures ranging from 800 to 900 jC. X-ray diffraction analysis of solids
identified the presence of iron, graphite and a carbide of iron as the products of reactions. A kinetic
model based on the first-order irreversible rate kinetics was developed and fitted to the TGA data so
as to estimate the rate constants for each reduction reaction. The reaction pathways considered in this
analysis involved reduction of iron oxides, Boudouard reaction and iron-carbide formation. The rate
parameters were calculated and compared with data reported in literature.
D 2004 Elsevier B.V. All rights reserved.

Keywords: Iron oxide; Carbon monoxide; Reduction; Kinetics

1. Introduction

The objective of this research was to investigate the oxidation of CO present in syngas
to CO2. It is expected that subsequent sequestration of CO2 subsequently using an
appropriate adsorbent such as lime/dolomite would lead to a pure hydrogen gas stream
which has the potential for utilization in fuel cell applications. The conversion of CO
using Fe2O3 results in oxidation to CO2. This paper presents the evaluation of the kinetics

* Corresponding author.
E-mail address: lalvansb@muohio.edu (S.B. Lalvani).

0378-3820/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.fuproc.2003.12.009
34 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

of Fe2O3 reduction at high temperatures in a continuous stream of pure carbon monoxide.

In addition to the reduction to wusite and metallic iron, carbon monoxide disintegration
by the reverse Boudouard’s reaction [1,2] and carbidization [3 –5] are expected to occur
simultaneously. A thermodynamic analysis indicate that relatively low temperatures (less
than 1000 jC) and high CO concentrations along with the presence of metallic iron lead
to carbon deposition (Fig. 1). Increase in temperature and CO concentration leads to
reduction in iron oxides.
The reduction of iron oxide and various ores containing iron oxide have been studied
in the past [7 – 21]. Various controlling mechanisms have been suggested in their
research. Shimokawabe [6] suggested a random nucleation mechanism for the reduction
of hematite (Fe2O3) while phase boundary mechanism was evoked in Sastri et al.’s [8]
research. In 2001, Tiernan et al. [19] concluded that reduction of hematite to magnetite
(Fe3O4) was via phase boundary while that of magnetite to free iron was via random
nucleation. El-Geassy’s [13] work on the reduction of hematite at temperatures ranging
between 1173 and 1473 K concluded that the reduction of hematite to wusite (FeO) was
controlled by a mixed reaction mechanism in the early stages followed by interfacial
chemical reaction, while the reduction of wusite to iron was controlled by a mixed
chemical reaction. Similar conclusions were arrived at by Moon and Rhee [14] and Moon
et al. [15]. An experimental study on the kinetics of reduction of iron oxide in a fluidized
bed reactor [17] showed the existence of two stages of reduction. In the first stage, the
rate of reduction was controlled by mass transport in the gas phase and in the second
stage, the rate-controlling mechanism was chemical reduction within small grains of the
iron ore particles. Ishii et al. [9] found that the reduction of Fe2O3 was 10 times faster in
H2 atmosphere as compared to CO atmosphere. In 1997, Moon and Rhee [14] estimated
that the reaction rate for Fe2O3 reduction with CO was 2– 3 times lower than that with
H2. They estimated the value of the apparent activation energy for iron oxide reduction to
be 14.6 kJ/mol in a carbon monoxide atmosphere. In a later work [15], they estimated the
activation energy for the reduction reaction in CO to be 19.8 kJ/mol in pure CO stream,
the value of which increased with reducing CO partial pressure in a CO – H2 gas stream.
The value of the activation energy was found to be 42.1 kJ/mol in a 100% hydrogen

Fig. 1. Phase diagram for iron oxide reduction and Boudouard reaction. [6].
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 35

Table 1
Activation energies for iron oxide reduction reported in literature
Reference Activation energy Reducing
(kJ/mol) agent
Shimokawabe [7] 74 – 117 H2
Sastri et al. [8] 57 – 73 H2
Moon and Rhee [14] 35 H2
Moon and Rhee [14] 14.6 CO
Moon et al. [15] 42.1 H2
Moon et al. [15] 19.8 CO
Tiernan et al. [19] 96 – 106 H2

atmosphere. Table 1 lists the activation energies for iron oxide reduction in various
reducing atmospheres reported in the literature.
The focus of this research is to investigate these reaction conditions that result in the
oxidation of carbon monoxide to carbon dioxide. Carbon dioxide can subsequently be
removed by lime. A plot of the Gibbs free energy change with temperature is shown in
Fig. 2. The data show that the reduction of hematite to wusite is favored at high
temperatures. However, the capture of carbon dioxide via lime is generally favored at
temperatures less than 900 jC. Nonetheless, the temperature window of 725 – 900 jC

Fig. 2. Gibbs free energy vs. temperature.

36 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

affords thermodynamically favorable reaction conditions for both the oxidation of CO to

CO2 using iron oxide and the subsequent sequestration of CO2 by lime. In addition,
carbon monoxide disintegration and iron carbide formation is not thermodynamically
favored at temperatures greater than 725 jC. Hence, in this investigation, the temperature
range of 725 –900 jC was studied for the oxidation of CO. A kinetic model was
developed and the data from thermogravimetric experiments were fitted to the model to
extract the numerical values for the reaction rate parameters. X-ray diffraction analysis
was used to identify the various products of the reaction.

2. Experimental

Ferric oxide samples (both fine and coarse grade) of greater than 99% purity supplied
and certified by J.T. Baker was used in this study. The particle size distribution
measurements were performed using the Microtrac laser particle analyzer (Leeds and
Northrup Instruments, Model 7995-10). The results shown in Table 2 represent the
average of five measurements. The data show that the median sizes of the fine (powder)
and coarse (particle) samples were 1.22 and 58.6 Am, respectively. The average specific
surface areas of the fine and coarse samples, as determined by a BET analyzer, were
estimated to be 5.2 and 0.37 m2/cm3, respectively. The thermogravimetric analyses
(TGAs) of the gaseous reduction of the fine and coarse ferric oxide samples were
accomplished using a Perkin Elmer TGA-7 analyzer. The solid samples were preheated to
a desired temperature under an inert gas atmosphere of nitrogen, followed by the
introduction of carbon monoxide. Experiments were carried out at temperatures between
800 and 900 jC.

3. Kinetic model for reduction of iron oxide using carbon monoxide

The reduction of ferric oxide by carbon monoxide is assumed to take place via a series
pathway in which the products of the reactions are FeO, Fe and Fe3C. In addition, the
Boudouard reaction is involved to explain the formation of carbon and production of
carbon dioxide. Evidence of Fe3C was found by X-ray diffraction analysis of the products
of the reaction. Several researchers [3 –5,19] have identified various iron carbides being

Table 2
Particle size measurements
16% percentile 50% percentile 84% percentile Mean diameter Surface area Standard
(Am) (Am) (Am) (Am) (m2/cm3) deviation
Fe2O3 fine sample (powder)
0.81 1.22 1.69 1.2 5.2 0.44

Fe2O3 coarse sample (particles)

15.84 58.58 97.75 57.8 0.37 41
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 37

formed in such reducing atmospheres and the presence of carbon. The following
chemical reactions are proposed.
k1
Fe2 O3 ðAÞ þ 2CO ! 2FeOðBÞ þ 2CO2
k2
FeOðBÞ þ CO ! FeðCÞ þ CO2
k3
3FeðCÞ þ 2CO ! Fe3 CðDÞ þ CO2
k4
2CO !
Fe
CðD1 Þ þ CO2

The above reactions can be represented by a series irreversible reaction mechanism as

shown below:
k1 k2 k3
A !B!C !D
k
CO !
Fe
4
D1 ð1Þ

The concentration vs. time profiles are analytically derived for the following cases: (i) k1 p
k2 p k3, (ii) k1 = k2, (iii) k2 = k3, (iv) k1 = k3, and (v) k1 = k2 = k3. It is reasonable to assume
that the above mentioned reactions can be represented by pseudo first-order, irreversible
chemical rate expressions because the experiments are carried out under the continuous
flow of pure carbon monoxide. The following differential equations describe the mass
balance of the chemical species involved in Eq. (1) assuming first-order irreversible
kinetics:
dA
¼ k1 A ð2Þ
dt

dB
¼ 2k1 A  k2 B ð3Þ
dt

dC
¼ k2 B  k3 C ð4Þ
dt

dD 1
¼ k3 C ð5Þ
dt 3

dD1
¼ k4 ðC þ DÞ ð6Þ
dt
38 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

The overall mass balance [i.e., sum of Eqs. (2) – (6)] can be represented as follows:
2A þ B þ C þ 3D ¼ Ct ¼ 2Ao þ Bo þ Co þ 3Do ð7Þ

The solution to Eq. (2) is obtained by separating the variables and using the initial
condition: t = 0, A = Ao:

A ¼ Ao ek1 t ð8Þ

The substitution of A from Eq. (8) into Eq. (3) results in the following expression:
dB
þ k2 B ¼ k1 Ao ek1 t
dt

The solution to the above first-order, homogeneous differential equation is given by:
Z
B ¼ 2k1 Ao eak1 t eða1Þk1 t dt þ C1 eak2 t ð9Þ
where,
a ¼ k2 =k1 ð10Þ

and C1 is the constant of integration which is evaluated using the initial condition: t = 0,
B = Bo. Similarly, the solution to the differential Eq. (4) is given by:
Z
C ¼ ak1 ebk1 t Bebk1 t dt þ C2 ebk1 t ð11Þ
where
b ¼ k3 =k1 ð12Þ

and C2 is the constant of integration which is evaluated using the initial condition: t = 0,
C = Co. Closed form solutions for B and C are reported in literature for the case when k1 p
k2 p k3, i.e., a p b p 1. In Table 3, we have reported the solution for kinetic concentration
of species involved for all possible scenarios, namely:

(i) a p 1 and
(ii) a = 1.

For case (i), three possibilities exist: (a) a p b, b p 1; (b) b = 1, and (c) a = b. For case (ii),
the following two possibilities were considered: (a) b p 1 and (b) a = b. The concentra-
tion of species D is obtained from Eq. (7). Finally, the concentration of species D1 is
obtained by integrating Eq. (6). Considering case I-a (Table 3), we obtain the following
expression for D1.
2ab1 Ao ½b  3 C1 b1 ½b  3a
D1 ¼ ðD1 Þo þ   ðek1 t  1Þ þ 
ða  1Þðb  1Þ 3 ðb  aÞ 3a
2b1
 ðeak1 t  1Þ þ C2 ðek1 t  1Þ þ b1 k1 ðC2 Þ1 t ð13Þ
3b
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 39

Table 3
Expressions for concentration of B and C
CASE B C C1 C2
(I) a p 1
(a) b p 1, 2 2aAo aC1 ak1 t 2Ao 2aAo
ap b Ao ek1 t ek1 t þ e Bo  Co 
a1 ða  1Þðb  1Þ ba a1 ða  1Þðb  1Þ
þ C1 eak1 t þ C2 ebk1 t aC1

ba
(b) b = 1; Same as above 2ak1 Ao tek1 t aC1 ak1 t Same as aC1
 e above Co 
a1 a1 1a
k1 t
þ C2 e

(c) a = b, Same as above 2aAo k1 t Same as 2aAo

bp 1 e þ ak1 C1 teak1 t above Co 
ða  1Þ ða  1Þ2
þ C2 eak1 t
(II) a = 1
(a) b p 1   Bo
2k1 Ao tek1 t 2Ao k1 t 1
Co þ
2Ao

C1
e k1 t 
þ C1 e k1 t b1 b1 ðb  1Þ2 b1
C1 k1 t
þ e þ C2 ebk1 t
b1

(b) a = b Same as above k1 Bo Co

ð2k1 Ao t þ C1 Þtek1 t þ C2 ek1 t
2
a = k2/k1, b = k3/k1.

where,

b1 ¼ k4 =k1

and

2Ao Bo Co
ðC2 Þ1 ¼ Do þ þ þ :
3 3 3

The total mass of the solids, W, is given by

W ¼ A MA þ B MB þ C MC þ D MD þ D1 MD1 ð14Þ

where, M stand for the molecular (or atomic) weight. Substituting D from Eq (7) into the
above equation we find

W ¼ AðMA  2=3MD Þ þ BðMB  1=3MD Þ þ CðMC  1=3MD Þ

þ 1=3ð2 Ao þ Bo þ Co ÞMD þ Do MD þ D1 MD1 : ð15Þ

Because, A, B, C and D1 are known as functions of time and from the mass of solids vs.
the time data, the three rate constants can be evaluated.
40 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

Fig. 3. X-ray diffraction pattern for the as-received iron oxide samples—fine (powder) and coarse (particle).

Fig. 4. X-ray diffraction pattern for the reduced iron oxide samples in pure CO atmosphere at 850 jC—fine
(powder) and coarse (particle).
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 41

4. Results and discussion

The size distributions of both the iron oxide samples used are provided in Table 2.
The X-ray diffraction pattern of the as-received iron oxide samples and that of the
products resulting from the TGA experiment in 100% CO are provided in Figs. 3 and
4, respectively. It is seen from these figures that the feed sample contained only a-
Fe2O3. The fine (powder) and the coarse (particle) samples are observed to have the
same phase structure. The X-ray diffraction pattern of the reduced product (Fig. 4),
when the fine sample was used, showed the presence of Fe3C, graphite and a-Fe. This
validates the inclusion of the formation of iron carbide in our kinetic model. The data
on the mass vs. time obtained at temperatures 800, 850 and 900 jC when experiments
were conducted using the fine (powder) sample are shown in Figs. 5 –7. The weight
loss data obtained for the particle sample using the same experimental conditions are
shown in Fig. 8 for comparison. It is observed from the data that the weight of the
solid sample decreases sharply for the first 2 –5 min followed by a continuous
increase. The initial loss of mass is attributed to the reduction of iron oxide by CO.
The subsequent increase in the mass of the solid is attributed to carbon deposition and
iron carbide formation as evidenced by the X-ray diffraction (Fig. 4). While both

30

Isothermal - 800 °C
25

20
Weight (gms)

15

10
Experimental
Theoretical
5

0
0 10 20 30 40 50 60 70 80
Time (mins)

Fig. 5. Experimental and theoretical TGA curves obtained via reduction of fine iron oxide sample in pure CO
stream at 800 jC.
42 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

Fig. 6. Experimental and theoretical TGA curves obtained via reduction of fine iron oxide sample in pure CO
stream at 850 jC.

samples showed increase in weight in the latter part of the experiments, the rate of
increase is observed to be higher in the case of the fine sample (as expected). In fact,
the experiment with the coarse sample at 800 jC showed a very modest increase in
weight even after 3 h. The theoretical weight loss of Fe2O3 to Fe is 32.5%. In the fine
sample, we observe a maximum weight loss of f 30%. The maximum weight loss is
noted to decrease with temperature. The difference between the theoretical value of the
weight loss and the actual experimental data is attributed to the simultaneous carbon
deposition and carbide formation. Both the occurrence of Boudouard reaction [1– 3]
and carbide formation [1,4,5] have been reported in the literature. In the case of the
coarse (particle) samples (Fig. 8), the maximum weight loss observed was less than
30% (23 –27%). Due to the larger particle size, the rate of reaction could be limited in
part by diffusion of CO onto the reactive sites. In addition, diffusion of CO2 from
reactive sites to the bulk could further reduce the overall reaction rate.
Eq. (15) was fitted to closely mimic the experimental data of the powdered sample
and the rate constants inherent in the model were evaluated. The model calculated data
are provided in Figs. 5 – 7. It is seen that the data and the theoretical model are in very
good agreement. Excellent match in the calculated and observed mass of solids is
observed for experiments conducted at 800 and 900 jC. The r2 values for the regression
were found to be greater than 0.995 in both the cases. The agreement in the data is fairly
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 43

30

Isothermal - 900 °C

25

20
Weight (gms)

15

Experimental
Theoretical
10

0
0 10 20 30 40 50 60 70 80
Time (min)

Fig. 7. Experimental and theoretical TGA curves obtained via reduction of fine iron oxide sample in pure CO
stream at 900 jC.

good (r2 = 0.99) for the experiment conducted at 850 jC for most part except for
reaction times of reaction greater than 60 min. Although magnetite (FeO.Fe2O3) may be
an intermediate product in the reduction of Fe2O3 to FeO, its role was not considered.
Magentite is a chemical mixture of Fe2O3 and FeO. Experimental evidence from the
TGA data (Fig. 9) shows that for shorter reaction times at lower temperatures ( < 600
jC), a sudden sharp drop in weight (up to approximately 3% weight loss) is observed
which is then followed by a relatively gradual weight change rate. The initial reaction
mechanism may involve the formation of magnetite. However, at higher temperatures
(>600 jC) used in our study for the evaluation of rate constants, no such change in the
rate of weight loss was observed. The plots of the natural log of the rate constants for
each reaction vs. temperature are presented in Fig. 10. It is seen from the figure that
reduction rate constant for hematite is consistently greater than that of wusite in the
temperature range of this study. The rate of constant for the Boudouard reaction is
greater that of the reduction of wusite and the formation of iron carbide. The formation
of iron carbide has the slowest reaction rate constant of all the reactions considered in
the temperature range of 800 –900 jC. The Arrhenius equation was used to obtain
activation energy and frequency factor for the four rate constants. The data is provided
44 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

Fig. 8. Experimental TGA curves obtained via reduction of coarse iron oxide samples in pure CO at 800, 850 and
900 jC.

Fig. 9. Experimental TGA curves obtained via reduction of fine iron oxide sample in pure CO at temperatures
ranging from 400 to 900 jC.
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 45

Fig. 10. Variation of rate constant with temperature for the reactions in iron(III) oxide reduction in CO.

in Table 4. The data show that the activation energies for Fe2O3 reduction agree fairly
well with the ones (Moon and Rhee [14], Moon et al. [15]) reported in the literature
(Table 1).
Experiments were also carried out using various CO concentrations on fine iron
oxide (powder) samples at a temperature of 850 jC. Required CO partial pressures
were obtained by diluting the CO stream with helium. Fig. 11 shows the effect of CO
partial pressure on the weight loss kinetics in the powdered sample for 30%, 40%, and
50% CO. It is seen in the figure that increasing the CO percent increases the rate of
weight gain after maximum weight loss has occurred as well as the onset of weight
gain. The data indicate that the rate of carbon monoxide disintegration and carbide
formation is significantly slower at lower CO concentrations. In addition, the rate of
weight loss in the initial minutes of the experiment was found to increase with
increasing CO partial pressures which indicates the greater than zero order dependence
of the rate of reaction on CO concentration at lower partial pressures. The data also
show that the controlling mechanism for powdered iron oxide reduction is reaction
controlled. The X-ray diffraction pattern of the product of reduction using 40% and
50% CO is shown in Fig. 12. No iron carbide phase was observed in the products
obtained from experiments conducted at CO partial pressures less than 50%.

Table 4
Estimated activation energies and frequency factors for the various reactions involved in iron oxide reduction with
100% CO gas stream
E (kJ mol-1) k0 (min-1)
k1 9.97 4.82
k2 14.13 7.44
k3 14.65 0.082
k4 10.60 0.44
46 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47

Fig. 11. Experimental TGA curves obtained via reduction of fine iron oxide sample in 30%, 40% and 50% CO at
850 jC.

Fig. 12. X-ray diffraction pattern for the reduced iron oxide samples in 40% and 50% CO atmosphere at 850 jC.
 K. Mondal et al. / Fuel Processing Technology 86 (2004) 33–47 47

Nomenclature
A, B, C, D, D1 Concentrations of reactive species (mol vol-1)
Ao, Bo, Co, Do, D1,0 Initial concentrations, (mol vol-1)
Ct Total concentration, (mol vol-1)
E Activation energy (kJ mol-1)
M Molecular (atomic) weight
k1, k2, k3 Rate constants, (min-1)
ko Frequency factor (min-1)
t time (min)
W mass (g)
a k2/k1
b k3/k1
b1 k4/k1

Acknowledgements

The authors would like to thank the Department of Energy for the financial support for
the project and Mr. Gerald Fink for XRD.

References

[1] O. Boudouard, Ann. Chim. Phys. 24 (1901) 85.

[2] Y.T. Chein, Y.-C. Ko, Am. Ceram. Soc. Bull. 62 (1983) 779.
[3] J. Fienman, Direct reduction and smelting processes, Iron Making Volume, AISE Steel Foundation, Pitts-
burg, PA, 1999, p. 763.
[4] B.H., Davis, E., Iglesia, Technology development for iron and cobalt Fischer – Tropsch catalysts, Quarterly
Report-DE-FC26-98FT40308, April 30, 1999.
[5] S. Sawai, Y. Iguchi, S. Hayashi, J. Iron Steel Inst. Jpn. 84 (1998) 844.
[6] W. Backloh, E. Spetzler, Archiv für Eisenhüttenwesen 11 (1940) 447.
[7] M. Shimokawabe, Thermochim. Acta 28 (1979) 287.
[8] M.V.C. Sastri, R.P. Viswanath, B. Viswanath, Int. J. Hydrogen Energy 7 (1982) 951.
[9] K. Ishii, T. Akiyama, Y. Kashiwaya, S. Kondo, Memoirs of the Faculty of Engineering, vol. 17, Hokkaido
University, Sapporo, Japan, 1986, p. 1.
[10] O.J. Wimmers, P. Arnoldy, J.A. Moulijn, J. Phys. Chem. 90 (1986) 1331.
[11] J. Jankowski, A. Sadowski, Iron Making Steel Making 23 (1996) 479.
[12] Y.K. Rao, P. Pichestapong, in: K.E. Spear (Ed.), Proc. of 9th Int. Conf. on High Temperature Materials
Chemistry, Electrochemical Society, Pennington, 1997, p. 211.
[13] A.-H.A. El-Geassy, Scand. J. Metal. 27 (1998) 205.
[14] I.J. Moon, C.H. Rhee, Proc of the 1997 TMS Annual Meeting, Feb 9 – 13, Orlando, FL, Minerals Metals and
Materials Society, Warrendale, PA, USA, 1997, p. 649.
[15] I.J. Moon, C.H. Rhee, D.J. Min, Steel Res. 69 (1998) 302.
[16] H.W. Kang, W.S. Chung, T. Murayama, ISIJ Int. 38 (1998) 109.
[17] A. Habermann, F. Winter, H. Hofbauer, J. Zirngast, J.L. Schenk, ISIJ Int. 40 (2000) 935.
[18] I.V. Murin, V.M. Smirnov, G.P. Voronkov, V.G. Semenov, B.M. Sinelnikiv, Solid State Ionics 133 (2000) 203.
[19] M.J. Tiernan, P.A. Barnes, G.M.B. Parkes, J. Phys. Chem., B 105 (2001) 220.
[20] Y. Xiao, M.A. Reuter, L. Holappa, Proceedings of the Ninth International Ferroalloys Congress and the
Manganese 2001 Health Issues Symposium, Ferroalloy Association, Quebec City, Canada, June, 2001, p. 147.
[21] T. Mattison, A. Lyngfelt, P. Cho, Fuel 80 (2001) 1953.