Liquidus Surface of FeO-Fe2O3-SiO2-CaO Slags at Constant CO2/CO Ratios

7/17/2019 Liquidus Surface of FeO-Fe2O3-SiO2-CaO Slags at Constant CO2/CO Ratios

http://slidepdf.com/reader/full/liquidus-surface-of-feo-fe2o3-sio2-cao-slags-at-constant-co2co-ratios 1/6

Liquidus Surface of FeO-Fe2O3-SiO2-CaO Slags at Constant CO2/CO Ratios

Florian Kongoli1* and Akira Yazawa2

1FLOGEN Technologies Inc., www.flogen.com, 5757 Decelles Ave., Suite 511, Montreal, Quebec, H3S 2C3, Canada

2Tohoku University, Sendai 981-0934 Japan

Liquidus surface of FeO-Fe2O3-SiO2-CaO slags is an important parameter in various smelting and converting processes. It helps not only

to optimize the slag chemistry of current processes and their fluxing strategies but also to determine the availability of new slags for more

advanced technologies. In our previous publications, the liquidus surface of some multicomponentiron oxide slags has been quantified at several

constant oxygen potentials and the eff ect of the latter, ignored until that moment, was quantified along with the eff ect of some minor

components. In this work, the liquidus surface of someiron oxide slags is quantified at constant CO2/COratios. This is a new convenientway for

the quantitative description of the slag liquidus surface and the e ff ect of several fluxes, especially in those processes, such as slag solidification,

where oxygen potential changes continuously. This type of diagram also describes more dynamically the eff ect of oxygen potential, clarifies the

relation between CO2/CO ratio and oxygen potential in terms of the liquidus surface (not widely understood by metallurgists today) and reduces

the gap between laboratory work and industrial experience.

(Received June 30, 2003; Accepted August 14, 2003)

Keywords: liquidus, FeOx-SiO2 based slag, iron oxides, oxygen potential, CO2/CO, slag solidification, smelting, converting

1. Introduction

Iron oxide slags are the most commonly used slags in

sulfide smelting and steel making. They usually contain silica

and lime as well as other minor oxides, which are introduced

through raw materials, fluxes, dissolved refractories etc.

Liquidus surface of these slags constitutes an important

parameter for the sulfide smelting and converting processes.

It helps not only to optimize the slag chemistry of the currentprocesses and the fluxing strategies, but also to determine the

availability of new slags for more advanced technologies.

In our previous work the liquidus surface of some iron

oxide slags has been quantified at low oxygen potentials,

characteristic of reductive processes1–3)

and at intermediate

oxygen potentials, characteristic of oxidative processes4–7)

such as direct smelting and continuous converting. This was

carried out by the means of a new type of multicomponent

phase diagrams1) at constant oxygen potentials and deducted

from the use of a new thermophysicochemical model.

Through a series of these diagrams, the important eff ect of

oxygen potential on the liquidus surface of multicomponentslags, ignored until that moment, was quantified along with

the eff ect of some minor components. Considerable con-

fusion found in literature about the eff ect of some minor

components was also clarified. Among others, it was found

that this eff ect could be fundamentally diff erent in reductive

and oxidative conditions. However, confusion still exists,

especially for those particular processes in which oxygen

potential changes dynamically mainly as a result of the

continuous cooling of the slag and the use of coke in the

process. An example of these processes is the settling phase

of matte smelting which is a subsequent sub-process of matte

oxidative smelting and/or slag solidification in which

temperature drops continuously from around 1573 to

1423 K and sometimes coke breeze is used in the last stage.

In these processes, as well as in some others, contradictions

are often found between the microscopic results of the

laboratory quenching measurements and slowly cooled

solidified smelting slags from the industrial practice. Contra-

dictory assertions are also given about the eff ect of minor

components in these processes. It seems that the sensitivity of

the slag liquidus temperature toward changes of the oxygen

potentials has been ignored and this is believed to be the

reason of the above confusion.Following a previous proposal,8) the purpose of this work

is to quantitatively describe the eff ect of the dynamic changes

of oxygen potential at these particular processes through a

new type of phase diagrams at constant CO2/CO ratios,

based on the previous model. This will not only help clarify

the above confusion but will also shed light in the under-

standing of the slag solidification process and the solidified

slag mineralogy, which are recently becoming important in

the environmental point of view.

The partial pressures of oxygen throughout the article are

given as dimensionless ones defined by pO2 = (PO2

)/

(101325 Pa).

2. Variation of PO2 and CO2/CO during Slow Cooling

As stated in previous work 4–7)

oxygen potentials during

matte smelting and blister making converting at 1573 K can

be respectively approximated as 108 and 106. This is

illustrated in Fig. 1, which gives the calculated equilibrium

oxygen and sulfur pressures during oxidative copper smelt-

ing. However, the oxygen potentials of the slag might drop

below 109 during matte smelting near the solidification

temperature or in the reductive slag-cleaning furnace. The

gradual cooling at a lower temperature and the use of coke

breeze during settling change continuously the oxygen

potential. Based on the previous article8)

the variations of

pO2 and CO2/CO during cooling are given below in Figs. 2

to 4.*Corresponding author, E-mail: [email protected]

Materials Transactions, Vol. 44, No. 10 (2003) pp. 2130 to 2135#2003 The Japan Institute of Metals



Figure 2 describes the variation of pO2 with temperature

when CO2/CO is kept constant and the variation of CO2/CO

with temperature when pO2 is kept constant. It can be seen

that a constant CO2/CO ratio describes more concisely the

continuous change of oxygen potential during equilibrium

cooling since it includes in itself many variations of pO2

depending on temperature.

Figure 3 describes the variation of the activity of FeO(l)

with temperature at constant pO2 and CO2/CO. It can be seen

that the eff ect of temperature on the activity of FeO(l) is

much less pronounced at constant CO2/CO ratio compared to

constant pO2 ratio.

Figure 4 describes the variation of pO2 and CO2/CO ratio

according to two separate equilibrium reactions and during

the equilibrium cooling of a slag ‘‘X’’ of composition

46.3 mass% FeO, 6.7 mass% Fe2O3, 37mass% SiO2 and

10 mass% CaO from 1623 K to 1373 K. It is shown that while

oxygen potential changes considerably during cooling from

around 107 to 1010, the resulting CO2/CO ratio is almost

constant. Taking into account this fact, the phase diagrams at

constant CO2/CO ratio seem to be an interesting new

alternative in the quantification of the liquidus surface of a

multicomponent slag at those processes such the continuous

cooling and solidification where the oxygen potentials

change dynamically.

Some examples of this new type of diagrams at constant

CO2/CO are given below. They have been constructed by

FLOGENTM software9)

through a new thermophysicochem-

ical model,4) which was already verified against all available

experimental data on liquidus temperatures as well as other

thermodynamic properties at several oxygen potentials.

Fig. 1 Variations of equilibrium oxygen and sulfur pressures during

oxidative matte smelting and converting. 1000 moles CuFeS2 concentrate

is assumed to be oxidized with air or oxygen enriched air at 1473.15 K and

1573.15K. ( : 1573 K Air blow; -- - - - - 1573K, 40%O2 blow; –– – –– –1473 K, Air blow).

Fig. 2 Relation between PO2 a nd CO2/CO ratio according to

2CO(g)+O2(g) = 2CO2(g) at constant CO2/CO (dashed lines) and

constant PO2 (solid lines).

Fig. 4 Variation of pO2 and CO2/CO ratio with temperature according to

3FeO(l)+O2(g) = Fe3O4(s) and 3FeO(l)+CO2(g) = Fe3O4(s)+ CO(g) at

a constant aFeO(l) and during the equilibrium cooling of a real slag ‘‘X’’ of

composition 46.3mass% FeO, 6.7mass% Fe2O3, 37mass% SiO2 and

10 mass% CaO from 1623 to 1373K.

Fig. 3 Eff ect of temperature on the activity of FeO(l) coexisting with

Fe3O4(s) at fixed PO2 (dashed curves) according to 3FeO(l)+O2(g) =

Fe3O4(s) and at fixed CO2/CO (solid curves) according to

3FeO(l)+CO2(g) = Fe3O4(s)+CO(g).

Liquidus Surface of FeO-Fe2O3-SiO2-CaO Slags at Constant CO2/CO Ratios 2131



3. Polythermal Projection Diagrams at Constant CO2/

CO Ratio

Figure 5 presents the liquidus surface of FeO-Fe2O3-SiO2-

CaO slag at a constant ratio of log(CO2/CO) = 2 by the

means of a new format of multicomponent phase diagrams

whose basis has been previously described.1)

This is in fact

the slag liquidus surface during slow equilibrium cooling

from 1623 to 1373 K where the oxygen potentials change

continuously from about 105 to around 108. It can be

seen that at these conditions several primary phases are

present, i.e., magnetite (spinel), alpha-Ca2SiO4, Ca3Si2O7,

wollastonite/pseudo-wollastonite and silica and each of themhas its own characteristics. Some important points can be

easily made from this diagram. First, during continuous

cooling of these slags, the ‘‘magnetite’’ spinel phase is the

dominant primary precipitate phase. There is no stable

olivine or ‘‘fayalite’’ at these conditions, which suggest that

the name ‘‘fayalite slag’’ is not adequate at this particular

case. A meaningful name would be ‘‘magnetite’’ or ‘spinel’

slag if the primary precipitate phase is to be used to name the

slag. At these conditions the mineralogy of a slow cooled

solidified slag with an overall liquid composition of Fe/

SiO2=1.1 and CaO=10 mass% (point X in the diagram)

would mostly contain primary magnetite. Second, lime doesnot decrease, but instead, increases the liquidus temperature

of the slag at spinel saturation area and consequently

increases the risk of the magnetite (spinel) precipitation

which makes lime not a good flux in terms of the liquidus

temperature. This eff ect is much more pronounced in this

kind of diagrams compared to the diagrams at constant pO2

presented previously4)

and this reflects the sensitivity of the

liquidus temperature toward changes on the oxygen poten-

tials. It should be mentioned however that lime decreases the

liquidus temperature only in the region of newly proposed

FCS slag5,6) where it may be used as a good flux. Third, at

constant CaO an increase of Fe/SiO2 ratio in the ‘magnetite’

saturation area would increase the risk of ‘magnetite’precipitation. This eff ect is more pronounced in this diagram

compared to the one given at constant pO2, which again

reflects the sensitivity of the liquidus temperature toward

changes on the oxygen potentials.

Figure 6 presents the liquidus surface of the same slag at

log(CO2/CO)=1. Again this represents the liquidus surface

of this slag during continuous cooling from 1623 to 1373 K

where the oxygen potentials change continuously from about

107 to around 1010. It can be seen that besides the 5

primary phases mentioned above, two new stable phases are

present at these particular conditions i.e. olivine and wustite.

Contrary to the previous case, lime in small and limited

amounts decreases the slag liquidus temperature in the

olivine saturation area but increases it in the spinel(magne-

tite), wustite, and wollastonite saturation areas. At theseconditions the mineralogy of a slow cooled solidified slag

with an overall liquid composition of Fe/SiO2=1.1 and

CaO=10 mass% (point X in the diagram) would mostly

contain primary olivine.

Figure 7 gives the liquidus temperature of the slag at

log(CO2/CO)=0.3 which corresponds to the continuous

equilibrium cooling of the slag from 1623 to 1373 K where

the oxygen potentials drop continuously from about 108 to

around 1012. In this case spinel is not anymore a primary

Fig. 5 Liquidus surface of FeO-Fe2O3-SiO2-CaO slag at log(CO2/

CO)= 2.


CO)= 1.


CO)= 0.3.

2132 F. Kongoli and A. Yazawa



stable phase, but all other primary phases are present i.e.

olivine, wustite, alpha-Ca2SiO4, Ca3Si2O7, wollastonite/

pseudo-wollastonite and silica. It can be seen that lime

decreases the liquidus temperature of these slags in the

olivine saturation area but increases it in almost all other

areas. At these conditions the mineralogy of a slow cooled

solidified slag with an overall liquid composition of Fe/

SiO2=1.1 and CaO=10 mass% (point X in the diagram)

would mostly contain primary olivine.

4. Isothermal Diagrams at Constant CO2/CO Ratio

Figure 8 describes the eff ect of CO2/CO ratio on the liquidregions of the current slag at 1448 K. The liquid regions for

constant pO2 have also been given for comparison. It can be

seen that, as expected, decreasing the CO2/CO ratio or the

pO2 increases the liquid region at this particular temperature

especially in the olivine, wustite and spinel surface. The slag

‘‘X’’ with an overall liquid composition of Fe/SiO2=1.1 and

CaO=10mass% would be completely liquid only at

log(CO2/CO) of 1 and 0.3 as well as at pO2 of 1010.

Figure 9 also describes the eff ect of CO2/CO ratio on the

liquid regions of the current slag at 1498 K. The liquid

regions for constant pO2 have also been given for compar-

ison. In this case also, a decrease of CO2/CO ratio or pO2

increases the liquid region. The slag ‘‘X’’ with an overall

liquid composition of Fe/SiO2=1.1 and CaO=10 mass%

would be completely liquid only at log(CO2/CO) of 1 and 0.3

as well as at pO2 of 108 and 1010. It should also noted that

the liquidus curve at log(CO2/CO)=0.3 coincides with the

one at PO2 ¼ 1010 since at this temperature these values

correspond to each other. This can be easily understood from

Table 1 that gives the relationship between the log(pO2)

values at several constant log(CO2/CO) constant ratios atvarious temperatures for the reaction 2CO2(g) =

2CO(g)+O2(g). In our particular case at log(CO2/CO) of

0.3 the corresponding value of oxygen potential is log(pO2) of

10:06. This explains the overlapping of liquidus curves at

both above-mentioned conditions.

5. Quenching and Slow Cooling

As mentioned above, contradictions are often found

among the microscopic results of laboratory quenching

measurements and slowly cooled solidified smelting slags

from industrial practice. For instance, in matte smelting,quenching experimental measurements show that ‘magnetite’

is normally the primary precipitate solid phase of the

quenched samples within the glass phase (the finely crystal-

line structure, representing frozen liquid). However, micro-

scopic examinations of relatively big amounts of solidified

slag from this process, especially from settling stage, show

olivine as the dominant crystallized phase. The fluxing eff ect

of lime in both cases is also a subject of confusion. These

disagreements can now be explained and clarified in the light

of the present work.

Figure 10 gives the liquidus temperature of the FeO-

Fe2O3-SiO2-CaO slag at Fe/SiO2=1.1 and at diff erent

constant values of CO2/CO and pO2 as well as at ironsaturation. If the above-mentioned slag ‘‘X’’ (Fe/SiO2=1.1,

CaO=10 mass%) is kept at constant oxygen potentials, as it is

almost the case of matte smelting and continuous converting

Fig. 8 Liquidus regions of FeO-Fe2O3-SiO2-CaO slag at 1448K and at

various constant values of CO2/CO and oxygen potentials.

Fig. 9 Liquidus regions of FeO-Fe2O3-SiO2-CaO slag at 1498K and at

various constant values of CO2/CO and oxygen potentials.

Table 1 The relationship between logðpO2Þ and log(CO2/CO) ratios at

various temperatures for the reaction 2CO2(g) = 2CO(g)+O2(g)

T /K DeltaH DeltaS DeltaG K Log (K) Log(PO2)

(kJ) (J/K) (kJ) Log(CO2/CO)

2 1 0.3

1373 562.100 171.038 327.239 3.61E-13 12:44 8:44 10:44 11:84

1398 561.864 170.868 322.965 8.71E-13 12:06 8:06 10:06 11:46

1423 561.626 170.699 318.696 2.04E-12 11:69 7:69 9:69 11:09

1448 561.385 170.532 314.430 4.62E-12 11:34 7:34 9:34 10:74

1473 561.143 170.366 310.169 1.02E-11 10:99 6:99 8:99 10:39

1498 560.899 170.201 305.912 2.19E-11 10:66 6:66 8:66 10:06

1523 560.653 170.038 301.659 4.58E-11 10:34 6:34 8:34 9:74

1548 560.405 169.877 297.410 9.36E-11 10:03 6:03 8:03 9:43

1573 560.155 169.717 293.165 1.87E-10 9:73 5:73 7:73 9:13

1598 559.904 169.559 288.924 3.65E-10 9:44 5:44 7:44 8:84

1623 559.652 169.402 284.687 6.99E-10 9:16 5:16 7:16 8:56




(108 or 106) the primary precipitate phase is ‘‘magnetite’’

(spinel) within the glassy phase representing liquid. If the

slag ‘‘X’’ is equilibrated at 1433 K and pO2 ¼ 108 and then

quenched, the microscopic examination will reveal only

‘magnetite’(spinel) as a precipitate phase besides the glass. In

this case lime is not a good flux in terms of liquidus

temperature since it increases it and consequently increases

the risk of magnetite precipitation. If the same slag ‘‘X’’ is

slow cooled, as it is the case of certain relatively big amount

of slags in the industrial practice or in the settling phase of

matte smelting, the primary precipitate phase would now be

olivine. As it can be seen in the upper part of Fig. 4, duringthe equilibrium cooling of the slag X, log(CO2/CO) stays

almost constant around the value of 1 and at these conditions

the primary precipitate phase is olivine, as shown in Fig. 10.

Although the slow cooling of the industrial slag and it

solidification is not a truly equilibrium process the diagrams

at constant CO2/CO ratio are the best approximation of these

processes. The microscopic examination of many solidified

slags confirms this conclusion since it reveals that many of

these slags consist of mainly olivine. In this case lime in

limited amounts would be a good flux in terms of the liquidus

temperature, especially in the settling phase of matte

smelting where the temperature may reach 1473 K andsometimes coke breeze is used in the process. In Fig. 10 it is

also worth noting that at iron saturation the primary

precipitate phase is still olivine although at this particular

conditions oxygen potential stays almost constant around the

value of 1011 or 1012. Quenching of the slag X in iron

crucible from a holding temperature of 1360 K would

produce only olivine as primary precipitate crystals within

the field of frozen liquid.

In this light, it can be said that there is a fundamental

diff erence between quenching and slow equilibrium cooling

of an iron oxide slag at intermediate oxygen potentials. The

diff erence could be in the primary precipitate phases, on the

eff ect of minor components, in the mineralogical composi-

tion of the solidified slag, etc. This diff erence has not been

always understood and one of the reasons for that is that the

experiments have always been carried out in equilibrium with

metallic iron and in air where the oxygen potential does not

change (in air) or change only slightly (in equilibrium with

metallic iron). This is clearly reflected on the diff erence that

exists between the diagrams at constant oxygen potentials4)

and those at constant CO2/CO, as given in this work.

6. Conclusions

The liquidus surface of FeO-Fe2O3-SiO2-CaO slag was

quantified through a new type of phase diagrams at constant

CO2/CO ratios. It was shown that this is a convenient

quantitative way for the description of the slag liquidus

surface and the eff ect of minor components in those

processes, such as slag solidification, where the oxygen

potential changes continuously. It describes more dynam-

ically the eff ect of oxygen potential and clarifies the relation

between CO2/CO ratio and oxygen potential in terms of the

liquidus surface. The analysis of the variation of pO2 and

CO2/CO showed that during slow equilibrium coolingalthough the oxygen potential changes continuously, CO2/

CO ratio stays almost constant. Consequently, this new type

of diagrams at constant CO2/CO ratio also simulates the slow

cooling process of the industrial slag.

The eff ect of CO2/CO ratios and oxygen potentials on the

liquidus temperature were also quantified and the contra-

dictions often found between the microscopic results of

laboratory quenching measurements and slowly cooled

solidified smelting slags from industrial practice were

clarified. It was shown that there is a fundamental diff erence

between the quenching and slow equilibrium cooling of an

iron oxide slag at intermediate oxygen potentials in terms of

the primary precipitate phases, the eff ect of minor compo-nents, the mineralogical composition of the solidified slag,

etc. The eff ect of lime on the liquidus temperature was also

quantified at both processes for FeO-Fe2O3-SiO2-CaO

system. However, the presence of other minor components

may alter this eff ect. The constructed diagrams shed light in

the understanding of the slag solidification process and the

solidified slag mineralogy, which are recently becoming

important in the environmental point of view. The diagrams

at constant CO2/CO and pO2 are complements of each other

and both help in the quantification of the liquidus surface of

slags in several metallurgical processes.

Acknowledgment

The authors wish to thank Mitsubishi Materials Corpo-

ration and Sumitomo Metal Mining Co. Ltd. for the financial

support.

REFERENCES

1) F. Kongoli and I. McBow: EPD Congress 2000, ed. by P. Taylor, (The

Minerals, Metals and Materials Society, Warrendale, PA, 2000) pp. 97-

108.

2) F. Kongoli and I. McBow: Copper 99, Vol. VI: Smelting, Technology

Development, Process Modeling and Fundamentals. ed. by C. Diaz, C.

Landolt and T. Utigard, (The Minerals, Metals and Materials Society,

Warrendale, PA, USA, 1999) pp 613-625.

3) F. Kongoli, M. Kozlowski, R. A. Berryman and N. M. Stubina: James M.

Toguri Symposium on the Fundamentals of Metallurgical Processing,

Fig. 10 Liquidus temperature of the FeO-Fe2O3-SiO2-CaO slag at Fe/

SiO2=1.1 and at diff erent constant values of CO2/CO and pO2 and at iron

saturation.

2134 F. Kongoli and A. Yazawa



ed. by G. Kaiura, C. Pickles, T. Utigard and A. Vahed, (The Canadian

Institute of Mining, Metallurgy and Petroleum, Ottawa, 2000) pp 107-

121.

4) F. Kongoli and A. Yazawa: Metall. Mater. Trans. B 32 (2001) 583-592.

5) A. Yazawa and F. Kongoli: High Temperature Materials and Processes

20 (2001) 201-207.

6) F. Kongoli and A. Yazawa: James M. Toguri Symposium on the

Fundamentals of Metallurgical Processing, ed. by G. Kaiura, C. Pickles,T. Utigard and A. Vahed, (The Canadian Institute of Mining, Metallurgy

and Petroleum, Ottawa, 2000) pp 365-377.

7) A. Yazawa and F. Kongoli: Proceedings of the Sixth International

Conference on Molten Slags, Fluxes and Salts, Stockholm, Sweden -

Helsinki, Finland, 2000, ed. by S. Seetharamanand D. Sichen, (Trita Met

85, 2000), CD-Rom, 016.pdf.

8) A. Yazawa: Tetsu-to-Hagane 86 (2000), 1-10.

9) F. Kongoli, I. McBow and S. Llubani: FLOGEN Technologies Inc.

(www.flogen.com), 1999-2003.


Liquidus Surface of FeO-Fe2O3-SiO2-CaO Slags at Constant CO2/CO Ratios

Documents