A COMPARISON OF THE KINETICS OF THE CO-CO2 REACTION WITH STEELMAKING 393
Introduction
The reaction of iron oxide in the liquid state with solid
carbon or carbon dissolved in iron has had increasing
attention paid to it in recent years. The reaction is the basis
of iron or semi-steel production in the new generation of
ironmaking processes, also called ‘Direct Smelting’. The
reaction also plays a major role in increasing the efficiency
of the electric arc furnace by slag foaming. In addition, in
modern copper making processes, at the ‘slag cleaning’
stage, reduction of magnetite in the slag with carbon is
employed to lower viscosity of slag. With more fluidity,
separation of entrapped matte and metal from slag is
facilitated. Because of the practical importance of the
reaction, a considerable number of studies have been
reported on both fundamentals and applied aspects of
reaction of iron oxide in the liquid state with carbon. It has
been demonstrated1–5 that the reaction proceeds through an
intermediate gas layer around carbon particles. This layer is
mostly comprised of CO and CO2. The overall reaction
involves the following individual steps in series1–4:
Diffusion of FeOx (Fe+2/Fe+3 and O-2 ions) from the bulk
slag to the slag-gas interface;
Interfacial chemical reaction at the slag-gas interface,
[1]
[2]
Diffusion of CO2 away from the slag-gas interface toward
the gas-carbon interface.
Chemical reaction at the carbon-gas interface,
[3]
Diffusion of CO away from the gas-carbon interface to
the gas-slag interface.
Numerous kinetics studies have been carried out on the
overall reaction rate as well as individual reaction steps.
Inconsistencies have been observed among the results of
different investigations for the values of reaction rate,
controlling steps and mechanism of reaction5–8,10.
In the present work, the rate of dissociation of CO2 on
different types of slag has been studied. The effect of iron
oxide content, oxidation state and temperature, on the rate
constant has been investigated. In addition, the effect of
copper oxide addition to the slag has been examined.
Experimental
Isotope exchange method
In this work, the isotope exchange method has been
employed for the measurements of rate of reaction. Many
studies have been made to measure the interfacial rate of
CO2 dissociation on the surface of different metals11–12 and
slags13–17 by isotope exchange method. The overall
exchange reaction is given by:
[4]
where * symbol, represents the labelled carbon. This can be
either the radioisotope 14C or stable isotope 13C.
Equation [5] was employed by Cramb and Belton12 to
determine the rate constant (ka) of CO2 dissociation when
14CO2 is the tracer isotope;
[5]
where A is the reaction area; T, absolute temperature; β, the
equilibrium CO2/CO ratio; V°, the volume flowrate of the
˙
/
V
ART
n
P P eq
k
CO CO
a
1
1
1 1
1 14 13+ − ( )
=β
* *CO CO CO CO2 2+ = +
CO C CO2 2+ =
2 3 2 23 2 2 2 2Fe O CO Fe O CO
+ − + −( ) + = ( ) +, ,
Fe O CO Fe CO2 2 2
+ −( ) + = +,
BARATI, M., CHEN, E., and COLEY, K. A comparison of the kinetics of the CO-CO2 reaction with steelmaking and copper making slags. VII International
Conference on Molten Slags Fluxes and Salts, The South African Institute of Mining and Metallurgy, 2004.
A comparison of the kinetics of the CO-CO2 reaction with
steelmaking and copper making slags
M. BARATI, E. CHEN, and K. COLEY
McMaster University, Department of Materials Science and Engineering, Hamilton, Ontario, Canada
The reaction between carbon and slags is of significant importance in direct smelting of iron ores,
EAF slag foaming and copper slag cleaning processes. This reaction is known to occur via
gaseous intermediates and it is important to have fundamental data to understand the relative
importance of each reaction step. In this study, the rate of CO2 dissociation on different slag
surfaces is measured using an isotope exchange technique. It was found that increasing iron oxide
content up to 30 wt per cent does not have a significant effect on the rate constant while increasing
from this limit, increases the rate constant dramatically. Activation energy of CO2 dissociation on
slags of unit basicity and different iron oxide contents is in the range of 180±40 kJ/mol. It is found
that rate constant dependence on oxygen potential in the slag phase is increasing with iron oxide
concentration. In the case of copper making slags, the influence of copper oxide is found to
increase the magnitude of apparent rate constant, while decreasing the activation energy
considerably.
Key words: Gas-slag reaction, CO2 dissociation, copper slag, kinetics, isotope exchange
Administrator
高亮
MOLTEN SLAGS FLUXES AND SALTS394
gas mixture; P14CO, the partial pressure of tagged CO in the
reacted gases and (P14CO)eq is the partial pressure of
labelled CO if complete equilibrium were to be achieved.
In the present study, the stable isotope 13C (13CO2) was
employed as the tracer isotope. By taking full account of
the natural abundance of 13C, Equation [5] was modified as
follows:
[6]
where all meanings for terms in Equation [5] are valid for
this equation with the new term P13°CO being the partial
pressure of 13CO in the inlet gases, i.e. the partial pressure
of natural 13CO.
Experimental procedure
A slag sample weighting 0.5 to 1 gram was placed in a
platinum, alumina or zirconia crucible in the vertical
resistance heated tube furnace. This arrangement is shown
in Figure 1. Once the desired temperature was achieved,
CO-CO2 gas mixture was passed over the slag. Gas
flowrates were adjusted to yield desired CO2/CO ratio and
flow rate. The overall gas flowrate was kept high enough to
be in the regime of interfacial chemical control. Each slag
sample was equilibrated for at least 1 hour under controlled
CO2/CO ratio then labelled gas was introduced. The
labelled gas contained 10 mol per cent 13CO2, but was
diluted on mixing. An Isoprime isotope ratio mass
spectrometer (Micromass) was employed to measure the
concentration of labelled species in both ingoing and
outgoing gas samples.
Results
CaO-SiO2-FeOx Slags
Experiments were conducted to measure the oxidation rate
of CaO-SiO2-FeOx slags with unit basicity (CaO/SiO2=1.0
molar) and iron oxide content varying from zero to 90 wt
per cent. The effect of oxidation state of the slag (fixed by
the CO2/CO ratio and temperature) on the magnitude of rate
constant was examined for each slag composition.
Effect of iron oxide content
Figure 2 shows the dependence of apparent rate constant,
ka, on the iron oxide content in the slag. A very sharp
increase in the ka is seen with ‘FeOx’ content increasing
above 30 and up to 90 per cent. Data from several other
studies are included for comparison.
Effect of oxidation state of the melt
Dependence of ka on the applied CO2/CO ratio for some of
the studied slags is presented in Figure 3. It is seen that log
(ka) decreases linearly with increasing log(CO2/CO).
Effect of temperature
Figure 4 shows the temperature dependence of ka with
CO2/CO=1.0 for slags of unit basicity and different iron
oxide contents. Activation energy of 180 ± 40 kJ/mol is
obtained for different iron oxide content.
SiO2-Al2O3-FeOx -(Cu2O) Slags
To determine the effect of copper, rate constant
measurements at varying CO2/CO and temperature were
˙
/
/
V
ART
n
P P
P P eq
kCO CO
CO CO
eq
a
1
1
1
1
1
13 13
13 13+
− ( )
− ( )
=
°
β
Figure 1. Schematic diagram of experimental apparatus
Figure 2. Apparent rate constant as a function of FeOx content
for slags equilibrated with CO2/CO =1.0
Figure 3. Apparent rate constant as a function of CO2/CO ratio
for different slags at 1873 K
Lo
g
(K
a,
mo
l/c
m2
.
s.
a
tm
A COMPARISON OF THE KINETICS OF THE CO-CO2 REACTION WITH STEELMAKING 395
conducted on two slags: copper free slag with 60 wt per
cent FeOx, 30 wt per cent SiO2, 10 wt per cent Al2O3 and a
slag, with the same wt ratios of FeOx to SiO2 to Al2O3, plus
6 wt per cent Cu2O. The dependence of ka on CO2/CO ratio
for both slags is provided in Figure 5. An increase in the ka
is seen while the slope of lines remained essentially
constant.
Results of present and previous studies on temperature
dependence of ka are presented in Figure 6. Addition of
copper to slag has lowered the activation energy from 190
to 100 kJ/mol.
Discussion
Rate of oxidation of iron oxide containing slags
Effect of oxidation state of the slag
The relationship between the rate constant of CO-CO2
reaction with slags and the oxidation state of slag can be
simply expressed as
[7]
where aO is the imposed CO2/CO, ka° is temperature
dependent constant for any slag composition and n
characterizes the degree of dependence of rate constant to
the oxidation state of the slag.
k k aa a O
n
= ( )° −
Figure 4. Arrhenius plot of the apparent rate constant measured at CO2/CO =1.0 for some slags
Figure 5. Dependence of apparent rate constant on CO2/CO for
two types of slags, with copper and without copper,
both at 1583 K
Figure 6. Arrhenius plot of the apparent rate constant, measured
at CO2/CO =1.0
� 85 wt pct FeOx
� 50 wt pct FeOx
� 30 wt pct FeOx
CaO/SiO=1.0
Linear (85 wt pct)
FeO2/CO=1
Linear (50 wt pct)
Linear (30 wt pct
FeOx)
Lo
g
(K
a,
mo
l/c
m2
.
s.
a
tm
Lo
g
(K
a,
mo
l/c
m2
.
s.
a
tm
Lo
g
(K
a,
mo
l/c
m2
.
s.
a
tm
MOLTEN SLAGS FLUXES AND SALTS396
Variation of n for different iron oxide content is provided
in Figure 7. Despite small discrepancies, it is shown that
increasing FeOx from zero to 90 wt per cent, increases n
from approximately 0.5 to 1.0.
While many investigators13,16–17 have found an inverse
relationship (ka = ka°(aO)-1) for the rate of reaction on iron
oxide rich or pure iron oxide condense phases, Sun19
showed that the first order rate constant for CO2
dissociation on CaO-SiO2-Al2O3 slags containing about 1
per cent FeOx decreased with aO according to relationship
ka = ka°(aO)-0.5. El-Rahaiby et al.16 have also shown that for
an equimolar CaO-SiO2-FeOx slag, ka = ka°(aO)-0.75). It is
thus apparent that, as FeOx decreases, ka dependence on aO
decreases. From results of this study, as seen in Figure 7, it
may be concluded that the rate constant dependence on the
state of oxidation is variable depending on the total iron
oxide content in the slag. Increasing the iron oxide content
of the slag, increases the value of n to unity for iron-rich
slags.
A mechanism, involving transfer of two negative charges
(electrons) to the adsorbed CO2 would lead us to expect a
value of 1 for n. For a detailed review, the reader is referred
to an article by Sasaki et al.13. Briefly, based on this
mechanism, for a given number of reaction sites, the
reaction rate is proportional to the surface concentration of
weakly adsorbed CO22- ions. The ions are formed through
following reaction:
[8]
The oxidation state of the slag, affects the electronic
defect structure, thereby affecting the electrochemical
potential of electrons which finally influence the
concentrations of CO22-. This mechanism explains the value
of -1 for n. However, it fails to explain the values of n less
than 1. There are two possible explanations for the
deviation of n from unity in some slag systems, particularly
at low iron oxide contents;
Firstly, if adsorbed ions on the slag surface, are produced
in the form of singly charged ions (CO2-), instead of doubly
charged CO2 ions, ( CO22- ), value of n, should be 0.5 in
ideal slags. This value is very close to low-iron content
slags. On the other hand, value of n in the iron-rich end of
diagram, is close to 1. Therefore, it can be explained that
increasing the iron oxide content of the slag, the dominant
mechanism changes from formation of singly charged to
doubly charged CO2 ions.
The second probable reason for changing the value of n
with FeOx content may be variation in the electronic
structure of slag with variation in iron oxide content. For
pure iron oxide, when doubly charged CO2 is adsorbed on
the surface, the ‘ideal value’ of n=1 is confirmed by this
charge transfer model. However, addition of other oxides to
the iron oxide can cause variations in its electronic
properties thus deviation in the ‘ideal’ behaviour.
Effect of iron oxide content
Results of present and some of previous studies have been
gathered in Figure 2. An exponential trend is observed in all
separate studies. Increasing iron oxide content up to
approximately 30 per cent does not have a significant effect
on the ka. On the other hand, increasing ‘FeOx’ content,
from this limit up to 100 wt per cent increases the rate
dramatically. The lower temperature studies of Li et al.6
and Sun19 deviate at higher FeOx contents. This can be
attributed to higher temperature in the present study.
Assuming a constant activation energy at different iron
oxide contents, the observed divergence of the two sets of
data is close to the expected. As seen, the agreement with
the study of Mori et al.17 for the experiments conducted at
the same temperature as the present work is extremely
good.
As noted in previous section, based on charge transfer
mechanism, electronic properties of slag can affect the rate
of reaction significantly. Several measurements on
electrical conductivity of slag20–23 show that increasing iron
oxide content of the slag, increases the electrical
conductivity. Li et al.6 gathered data on the electrical
conductivity of iron oxide melts and showed that electrical
conductivity as a function of FeOx content is analogous to
the variations of ka. They concluded that electrical
properties of slag are most likely to be the reason for such
trend in dependence of ka on FeOx content.
Effect of copper oxide addition
As shown in Figure 5, addition of 6 wt per cent Cu2O to the
slag at 1583K, increases ka by a factor of 3, while there is
no change in the CO2/CO dependency, with a slope of -0.4
for both lines. The effect of copper could be explained in
terms of basicity. In general, the apparent rate constant at a
given oxygen potential increases with the basicity of the
melt16. It is reasonable that oxygen transfer is more rapid on
a more extensively reduced surface, because of the readier
dissociation of CO224. From thermodynamic data, it is
known that Cu+ has a higher reducibility than Fe3+.
The arguments, whether considered in terms of basicity
or extent of reduction, depends on the presence of copper
ions in the slag. However, from thermodynamic data, it is
expected that in the present work a significant proportion of
copper oxide is reduced to copper metal. Therefore,
increase in the ka may have arisen from different
phenomena. One possibility is suspension of small copper
droplets at the slag/gas interface. Another possible effect is
the presence of Cu+ ions in the slag, as described earlier. At
this stage, the role of each phenomenon has not been
clarified but an investigation of this is ongoing.
The strong influence of copper on the apparent rate
constant is also illustrated in Figure 6. This figure shows
again that the reduction rate is more rapid on the surface of
a copper slag than on that of copper-free slag. The
CO e CO
ad ad2 2
22( )
−
( )
−+ =
Figure 7. Dependence of n on the iron oxide content in the melt
with CO2/CO=1.0
A COMPARISON OF THE KINETICS OF THE CO-CO2 REACTION WITH STEELMAKING 397
calculated activation energy is 190 kJ/mole for slag without
Cu2O. However, it is only 100 kJ/mole for slag with Cu2O
Results of other studies on similar slags are also shown in
Figure 6. The activation energy for copper free slag is very
close to El-Rahaiby’s16 result (196 kJ/mol) for a slag with a
similar composition. Compared with Utigard’s18 results, the
biggest difference for the rate constants is 1 order within
the experimental temperature range after adding Cu2O. This
might be because that there is 0.5 per cent S in Utigard’s
slag.
Conclusion
• For CO2/CO exchange in CaO-SiO2-FeOx-Al2O3-
(Cu2O) slags, the rate constant decreases with oxygen
potential. This decrease is consistent with a charge
transfer mechanism and a rate determining step
involving CO22n- formation/dissociation where 0.5 ≤ n
≤ 1.
• The rate constant for a given temperature is represented
by the equation
for an iron silicate slag with and without copper, n=0.4
for CaO-SiO2-FeOx slags, and n is a function of total
iron oxide content. For a CaO/SiO2=1 slag, n = 0.0048
(%FeOx)+0.469
• The activation energy was found to be independent of
total iron oxide content but was dependent on copper
presence in slag. For copper free slags, activation
energy was 180kJ/mol. Addition of 6 wt per cent
copper oxide, decreased activation energy to 100
kJ/mol.
• For CaO-SiO2-FeOx slags, the rate constant increases
significantly with total iron content, when iron oxide
increases from approximately 30 wt per cent.
Acknowledgement
The authors acknowledge the financial supports provided
by ‘Steel Research Center, McMaster University’ for Mr.
M. Barati and by ‘Centre for Chemical Process Metallurgy’
for Ms. E. Chen. The authors are grateful to Dr. F. Ji for his
invaluable assistance in arranging the experimental setup.
References
1. SARMA B., CRAMB A.W., and FRUEHAN R.J.
Reduction of FeO in smelting slags by solid carbon:
experimental results. Metall. Trans. B. vol. 27, 1996,
pp. 717–731.
2. SIDDIQI, N., BHOI, B., PARAMGURU, R.K.,
SAHAJWALLA, V., and OSTROVSKI, O. Slag-
graphite wettability and reaction kinetics, part 1
kinetics and mechanism of molten FeO reduction
reaction., Ironmaking and Steelmaking vol. 27, no. 6,
2000, pp. 437–441.
3. MIN, D.J. and FRUEHAN, R.J. Rate of reduction of
FeO in slags by FeO-C drops Metall. Trans. B.
vol. 23, 1992, pp. 29–37.
4. MIN, D.J., HAN, J.W., and CHUNG, W.S. A study of
the reaction rate of FeO in slag by solid carbon,
Metall. Trans. B. vol. 30, 1999, pp. 215–221.
5. SHEIKHSHAHAB BAFGHI, M., FUKUDA, M.,
ITO, Y., YAMADA, S., and SANO, M. Effect of CO
gas formation on reduction rate of iron oxide in
molten slag by graphite, ISIJ Int. vol. 33, no. 11,
1993, pp. 1125–1130.
6. LI, Y. and RATCHEV, I.P. Rate of interfacial
reaction between molten CaO-SiO2-Al2O3-FeOx and
CO-CO2, Metall. Trans. B. vol. 33, 2002, pp.
651–660.
7. JI, F., BARATI, M., COLEY, K., and IRONS, G.A. A
kinetic study of carbon injection into electric arc
furnace slags, 7th International Conference on Molten
Slags, Fluxes and Salts, Cape Town, South Africa,
25-28 January 2004.
8. STORY, S.R., SARMA, B., FRUEHAN, R.J.,
CRAMB, A.W., and BELTON, G.R. Reduction of
FeO in Smelting Slags: Re-Examination of the
Influence of the Gas-Carbon Reaction, Metall. Trans.
B. vol. 29, 1998, pp. 829–832.
9. MIN, D.J., HAN, J.W., and CHUNG, W.S. A Study
of the Reaction rate of FeO in slag by solid carbon,
Metall. Trans. B. vol. 30, 1999, pp. 215–221.
10. SHALIMOV, M.P., BORONESHKOV, V.N., and
LYAMKIN, S.A. Mechanism and kinetics of FeO-
SiO2 melts with carbon, Russian Metall., vol. 3, 1980,
pp. 31–34.
11. CRAMB, A.W., GRAHAM, W.R., and BELTON,
G.R. The interfacial kinetics of the reaction of CO2
with nickel. Part1: the 14CO2-CO exchange reaction
and the influence of sulphur, Metall. Trans. B. vol. 9,
1978, pp. 623–629.
12. CRAMB, A.W. and BELTON, G.R. Studies of the
interfacial kinetics of the reaction of CO2 with liquid
iron by the 14CO2-CO isotopic exchange reaction
Metall. Trans. B., vol. 12, 1981, pp. 699–704.
13. SASAKI, Y., HARA, S., GASKELL, D.R., and
BELTON, G.R. Isotope exchange studies of the rate
of dissociation of CO2 on liquid iron oxides and CaO-
saturated calcium ferrites. Metall. Trans. B., vol. 15,
1999, pp. 563–571.
14. SUN, S. and BELTON, G.R. The effect of surfactants
on the interfacial rates of reaction of CO2 and CO
with liquid iron oxide , Metall. Trans. B., vol. 29,
1998, pp. 137–145.
15. SUN, S., SASAKI, Y., and BELTON, G.R. On the
interfacial rate of reaction of CO2 with a calcium
ferrite melt, Met