ISIJ International, Vol. 45 (2005), No. 6, pp.

791–797

Phase Equilibria in the Cristobalite, Tridymite and Pyroxene

Primary Phase Fields in the MgO–“FeO”–SiO2 System in
Equilibrium with Metallic Iron

Shengqi CHEN, Evgueni JAK and Peter C. HAYES

Pyrometallurgy Research Centre, School of Engineering, The University of Queensland, Brisbane, Qld 4072, Australia.
E-mail: P.Hayes@minmet.uq.edu.au
(Received on October 15, 2004; accepted on December 21, 2004 )

Experimental studies have been carried out to determine the phase equilibria for the ternary MgO–
“FeO”–SiO2 in equilibrium with metallic iron. Liquidus isotherms have been determined in the tem-
perature range from 1 578 to 1 898K in the silica and pyroxene primary phase fields using high temperature
equilibration, modified quenching and electron probe X-ray microanalysis (EPMA) techniques.
KEY WORDS: slag; liquidus; phase equilibria; MgO–“FeO”–SiO2 system.

reported a calculated MgO–SiO2 phase diagram based on

1. Introduction
The slag system MgO–“FeO”–SiO2 forms the basis of a
number of important ceramic, metallurgical and igneous
rockforming systems. The only experimental liquidus data
for the MgO–“FeO”–SiO2 system found in the literature are
by Bowen and Schairer.1) Primary phase fields in the MgO–
“FeO”–SiO2 system include the silica (SiO2) poly- morphs
tridymite and cristobalite, magnesiowüstite (MgO–FeO),
olivine (Mg2SiO4–Fe2SiO4) and pyroxene (MgSiO3–
FeSiO3). No ternary compounds have been ob- served in
this pseudo-ternary system. The liquidus tempera- tures in
this system in equilibrium with metallic iron were measured
up to 1 773 K using equilibration/quenching tech- niques.1)
The liquidus and solid compositions were deter- mined by
optical metallography, chemical analysis and X-ray powder
diffraction. No liquidus data were obtained for the
temperatures higher than 1 773 K since metallic iron
envelopes (iron melting point 1 811 K2)) were employed in
their studies.1) The phase diagram of the MgO–“FeO”–SiO2
system in equilibrium with metallic iron redrawn by Muan
and Osborn3) based on Bowen and Schairer’s data is shown
in Fig. 1; the former did not add any new experimental data
on this system.
Liquidus and subliquidus experimental data in the MgO–
SiO2 system has been reported by a number of
researchers.4–6) The principal features of the MgO–SiO2
system include a region of liquid immiscibility on the sili-
ca-rich side, a congruently melting compound forsterite
(Mg2SiO4) and an incongruently melting compound
protoenstite (MgSiO3). The binary eutectic reaction in-
volves the liquid+MgO+Mg2SiO4 phases, and peritectic
reaction involves liquid+Mg2SiO4+MgSiO3 phases. A re-
cent experimental study7) has shown that there is no solid
solubility of SiO2 in MgO and has provided new data on the
liquidus in the silica primary phase field. Wu et al.8) have
optimised thermodynamic models describing the available
experimental data.
Liquidus and subliquidus equilibria in the “FeO”–SiO2
system in equilibrium with metallic iron have been system-
atically investigated by a number of researchers.4,9,10)
Congruently melting fayalite (Fe2SiO4) was found to be
the only binary compound in this system resulting in two
eutectic reactions, liquid+wustite+fayalite, and liquid+
fayalite+silica. Zhao11) reinvestigated this system and de-
termined the eutectic temperatures to be 1 461±2K and
1 462±2K respectively. No inflection in the liquidus of
wustite primary phase field was observed in this later inves-
tigation contrary to earlier findings.
Following assessments of the available thermodynamic
and phase equilibrium data the phase diagram of the MgO–
“FeO”–SiO2 system in equilibrium with metallic iron was
calculated by Wu et al.8) using the F*A*C*T thermo-
dynamic computer package.12) General agreement between
the predictions and the data were obtained in the silica pri-
mary phase field and the boundary between silica and py-
roxene primary phase fields. Critical reevaluation and reop-
timisation of the thermodynamic database for the MgO–
FeO-Fe2O3-SiO2 system have been performed by Degterov
et al.13) using the FactSage program.12) No direct
comparisons between calculations and the available experi-
mental data in the MgO–“FeO”–SiO2 system were reported
in either study.8,13) Karlemo and Taskinen14) predicted the
impact of silica on MgO–“FeO”–SiO2 slag by MTDATA
thermodynamic database and program,15) again no direct
comparisons between liquidus data and model predictions
were provided.
The objectives of the present study are to provide new
experimental liquidus and phase equilibria data in the silica
and pyroxene olivine primary phase fields of the MgO–
“FeO”–SiO2 system, to confirm the accuracy of the earlier

791 © 2005 ISIJ

 ISIJ International, Vol. 45 (2005), No. 6
Fig. 1. Phase diagram of MgO–“FeO”–SiO2 in equilibrium with metallic iron (Muan and Osborn3)).
study by Bowen and Schairer1) and to extend the investigat-
ed range of compositions and temperatures.
2. Experimental
The experimental procedure used in this investigation
has been described in detail in previous publications by the
authors.16–20) The slag samples were prepared from the fol-
lowing chemicals, obtained from Sigma-Aldrich Pty. Ltd.
(Australia): 99+% MgO, 99.9+% SiO2, 99.9+% iron pow-
der and 99.98+% Fe2O3. The chemicals were mixed in an
agate mortar to the required compositions. 15 % excess Fe
powder (<10 m m in size) was added to the slag samples to
ensure that free metallic iron is always present in and dis-
persed throughout the slag. About 0.3 g of the mixture was
pelletized and placed in molybdenum foil (over 99.9 % Mo,
0.025-mm thick) or iron foil (over 99.9 % Fe, 0.1-mm
thick) envelopes. Equilibration experiments were carried
out in a vertical lanthanum chromite furnace (PYROX,
France) having a recrystallised alumina reaction tube. The
furnace temperature was controlled within ±1 K. A B-type
Pt/30wt% Rh–Pt/13wt% Rh thermocouple was inserted into
the hot zone within the reaction tube. The thermocouple
was calibrated against a reference thermocouple supplied
by National Measurement Laboratory (CSIRO, Melbourne,
Australia). The overall temperature accuracy was estimated
to be within ±2 K.
The sample was first placed in a Mo dish suspended by a
Mo wire, and positioned in the low temperature zone at the
bottom end of the furnace, and the bottom end of the reac-
tion tube was sealed with a thin polymer film. Ultra high
purity Argon gas (total impuritiesŠ5 ppm, O2Š1 ppm) was
passed through the reaction tube for 30 min to remove air;
the sample was then raised into the hot zone adjacent to the
thermocouple. The experiments were carried out in two
steps. The first step was to completely premelt the sample;
© 2005 ISIJ
heating the slag above the liquidus temperature to ensure
the sample is homogeneous. The second step was to equili-
brate the sample at the desired temperature. After equilibra-
tion the suspension wire was released and the sample
quenched directly into stirred bath of cold water. Using this
rapid cooling approach the liquid molten oxide phase at the
equilibration temperature is retained as a glass of the same
composition at room temperature. The solid phases present
at the equilibration temperature are also retained in the
same crystal structure and compositions by rapid cooling to
room temperature.
The quenched samples were mounted and polished for
analysis. The polished samples were initially examined
using optical microscopy and scanning electron microscopy
(SEM) to identify the phases present, and energy dispersive
spectra (EDS) analysis were used for qualitative analysis.
The compositions of the phases were then measured using a
JEOL 8800L electron probe X-ray microanalyser (EPMA)
with wavelength dispersive spectrometers (WDS). An ac-
celerating voltage of 15 kV and a probe current of 15 nA
were used. Standards supplied by Charles M Taylor Co.
Stanford, CA were used in EPMA measurement. A wollas-
tonite (CaSiO3) standard (Ca=34.50, Si=24.18, and
O=41.32 wt%) was used for silicon, and a magnesia
(MgO) standard (Mg=60.31, and O=39.69 wt%) and a
hematite (Fe2O3) standard (Fe=69.94 and O=30.06 wt%)
for magnesium and iron determination respectively.
CaMoO4 standard (Ca=20.038, Mo=47.966, O=31.995
wt%) was used to inspect the Mo in the slag. The
Duncumb-Philibert ZAF correction procedure supplied
with the JEOL was applied. The average accuracy of the
EPMA measurements is within ±1 wt%. X-ray powder dif-
fraction (XRD) analysis was also used in some cases to
confirm the phase identification. XRD measurements were
carried out in a PHILIPS PW1130 X-ray diffractometer
with a graphite monochromator using Cu Ka radiation.
792
 ISIJ International, Vol. 45 (2005), No. 6

Table 1. EPMA measurements of the phases present in the Table 1. (Continued)

MgO–“FeO”–SiO2 system in equilibrium with metal-
lic iron.

The methodology and approach enabled accurate mea-

surement of the compositions of phases in samples. The
melting point of metallic iron is 1 811 K.2) It is therefore
not possible to use iron foil as an envelope to hold the slag
samples at higher temperatures. Mo foil (melting point
2 896 K2)) and wire were used in these higher temperature
studies. It was confirmed by EPMA measurement that there
is no molybdenum oxide present in the quenched slag sam-
ples. Accurate liquidus isotherms could be constructed
from experimental data by carrying out a series of experi-
ments at fixed temperatures.

3. Results and Discussion

3.1. Cristobalite and Tridymite Primary Phase Fields
Nearly one hundred compositions defining liquidus
isotherms between 1 623–1 898 K have been determined in
the tridymite and cristobalite primary phase fields of the
MgO–“FeO”–SiO2 system. EPMA measurements obtained
in this study are presented in Table 1. Each of the experi-
mental points reported is the average of five or more mea-
surements obtained using EPMA in different positions of

793 © 2005 ISIJ

 ISIJ International, Vol. 45 (2005), No. 6

phase in samples. The data are grouped by the primary
phase fields (cristobalite, tridymite and pyroxene) and by
the temperature of equilibration. Liquidus data have been
determined in the range 25–55 wt% “FeO” and 40–65 wt%
SiO2. It was found that there is no solid solution of MgO or
FeO in the cristobalite or tridymite phases.
Using these experimental data it is now possible to con-
struct the liquidus in the cristobalite and tridymite (SiO2)
primary phase fields as presented in Figs. 2 and 3. Liquidus
data along the binary MgO–SiO2 and “FeO”–SiO2 systems
were taken from Bowen and Andersen5) and Zhao11) respec-
tively. Solid lines were drawn according to the best fit to the
data obtained in the present experiments. It can be seen
from Fig. 2 that the shape of isotherms in the cristobalite
and tridymite primary phase fields are similar. The cristo-
balite/tridymite transition occurs at 1 738 K21) (see Fig. 2).
There appears to be no significant change in liquidus on
transition between these polymorphs. No solubility of MgO
in cristobalite and tridymite phases was found in this inves-
tigation while the solubility of “FeO” in cristobalite and
tridymite is insignificant and less than 0.9 wt%.
The experimental data at the low MgO concentration in
the tridymite and cristobalite primary phase fields are in
good agreement with the results reported by Bowen and
Schairer1) up to 1 773 K. It can be seen that three points at
1 773, 1 778 and 1 768 K from Bowen and Schairer’s result
are close the isotherms of 1773 K determined in present
experimental study. Present experimental points at 1 623,
1 673 and 1 723 K (Fig. 3) are consistent with the experi-
mental results from Bowen and Schairer1) at 1 643, 1 673
and 1 733 K respectively.
3.2. Pyroxene Primary Phase Field
Experiments were also carried out to determination the
liquidus and phase equilibria in the pyroxene primary phase
field and along the boundary line between pyroxene and

Fig. 2. Experimentally determined liquidus isotherms in the cristobalite primary phase field in the MgO–“FeO”–SiO2
system in equilibrium with metallic iron.

Fig. 3. Experimentally determined liquidus isotherms in the cristobalite and tridymite primary phase fields of the MgO–
“FeO”–SiO2 system in equilibrium with metallic iron. Crosses represent experimental data points by Bowen and
Schairer.1)

© 2005 ISIJ 794

 ISIJ International, Vol. 45 (2005), No. 6

Fig. 4. Experimentally determined liquidus surface in the MgO–“FeO”–SiO2 system in equilibrium with metallic iron in
the pyroxene and cristobalite/tridymite primary phase fields.

Fig. 5. Conjugate lines illustrating the phase equilibria between

liquid and pyroxene at various temperatures in pyroxene
primary phase field of the MgO–“FeO”–SiO2 system in
equilibrium with metallic iron.

cristobalite/tridymite primary phase fields between

1 578–1 798 K. The analysis of the phases observed in these
studies as measured using EPMA are given in Table 1.
Figure 4 shows present experimental results. The data from
Bowen and Schairer1) are also presented in Fig. 4 for com-
parison. It can be seen that the present experimental data at
the boundary line between cristobalite/tridymite and pyrox-
ene (1 673, 1 723 and 1 773 K) are in agreement with those
from Bowen and Schairer1) in pyroxene primary phase field
(cross point in Fig. 4 at 1 673, 1 703 and 1 768 K). The tem-
perature of the four phase equilibria SiO2-pyroxene-olivine-
liquid is higher than that reported by Bowen and Schairer1)
since no liquid phase was formed in the sample of this
composition equilibrated at 1 578 K (Table 1). The experi-
mentally determined conjugate lines between the liquid and
pyroxene phase at various temperatures are shown in Fig. 5.
3.3. Cristobalite/Tridymite Liquidus Temperatures
The liquidus temperatures for the MgO–“FeO”–SiO2 sys-
tem in equilibrium with metallic iron in the cristobalite/
tridymite primary phase fields based on experimental data
obtained in the present study in specific area are shown in
Figs. 6(a) and 6(b) as a function of the concentration of
FeO and SiO2 respectively. It can be seen (Fig. 6(a)) that
with increasing FeO concentration for fixed ratio of
Fig. 6. Predicted liquidus temperature of MgO–“FeO”–SiO2 sys-
tem in equilibrium with metallic iron in the cristobalite
primary field based on data obtained in the present study,
as functions of (a) FeO concentration at various SiO2/
MgO ratios, and (b) SiO2 concentration at various MgO/
FeO ratios.
SiO2/MgO, there is a gradual decrease of liquidus tempera-
ture of 10–12 K for each 1 wt% increase in FeO concentra-
tion. For a fixed MgO/FeO ratio an increase in SiO2 concen-
tration leads to an increase in liquidus temperature of ap-
proximately 23–35 K for each 1 wt% increase in SiO2 con-
centration (Fig. 6(b)) depending on the starting composi-
tion.
3.4. Industrial Applications
The principal components of slags obtained by the smelt-

795 © 2005 ISIJ

 ISIJ International, Vol. 45 (2005), No. 6
Fig. 7. The MgO–“FeO”–SiO2 system in equilibrium with metallic iron after Bowen and Schairer. 1) Slag compositions of
various nickel smelters.23) 1. Falconbridge-Dominicana, 2. Cerro Matoso, 3. Morro do Niquel, 4. PT Aneka
Tambang, 5. Hyuga, 6. SLN-Doniambo, 7. PT Inco, and 8. Exmibal.
ing slags of laterite ores to produce ferronickel are MgO,
“FeO” and SiO222); on this simplified basis the approximate
slag compositions used in various industrial operations are
shown in Fig. 7. In many cases this pseudo-ternary system
has been used as the basis for design of nickel smelting
processes.
Assuming the nickel smelter slag is in contact with the
Fe–Ni alloy (approximately 42 wt% Ni) at the operating
temperatures, the activity of iron in the system is approxi-
mately 0.6 relative to the pure liquid iron.24) As a first ap-
proximation the nickel smelting slags may be described in
the form of the MgO–“FeO”–SiO2 system in equilibrium
with metallic iron i.e. aFe=1.
In smelting practice changes in the SiO2/MgO ratio in
the feed at a particular operating temperature can lead to
conditions where the slag is above or below the liquidus.
Operating above the liquidus enhances the refractory disso-
lution. At conditions well below the liquidus the viscosity
of the slag increases rapidly. To compensate for these
movements in the liquidus the iron oxide concentration of
the feed can be adjusted. It is clear from the phase diagram
that the system is very sensitive to small changes to
SiO2/MgO ratio and iron oxide concentration of the feed,
hence the need to accurately characterise the liquidus sur-
face as a function of bulk slag composition, temperature
and oxygen partial pressure.
It can be seen from Fig. 7 that the liquidus temperatures
and phase equilibria in the cristobalite/tridymite and pyrox-
ene primary phase fields described in the present paper are
of particular relevance to the slags produced at the Cerro
Matoso (2), P. T. Aneka Tambang (4) and Morro de Niquel
(3) operations. Subsequent papers in this series will de-
scribe the influence of minor elements Al2O3 and Cr2O3 on
the liquidus temperatures in these industrially important
slag systems.
4. Conclusions
Liquidus isotherms in the range from 1 578 to 1 898 K in
the cristobalite/tridymite and pyroxene primary phase fields
© 2005 ISIJ
of the MgO–“FeO”–SiO2 system in equilibrium with metal-
lic iron have been accurately determined experimentally.
(1) The new experimental data in the cristobalite pri-
mary phase field at temperature higher than 1 773 K and
high FeO concentration regions represent conditions not
previously investigated in this system.
(2) The liquidus data obtained below 1 773 K and the
position of the cristobalite/tridymite and pyroxene primary
phase fields boundaries are generally in agreement with
those of Bowen and Schairer.1)
(3) At fixed ratio of SiO2/MgO there is a gradual de-
crease of liquidus temperature of approximately 10–12 K
for 1 wt% increase in FeO. For a fixed MgO/FeO ratio an
increase in SiO2 concentration leads to an increase in liq-
uidus temperature of approximately 23–35 K per 1 wt% in-
crease in SiO2.
Acknowledgements
The authors would like to thank BHP Billiton, QNI,
Cerro Matoso SA and Australian Research Council (ARC
Linkage Program) for providing the financial support to en-
able this research to be carried out, and to Centre for
Microstructure and Microanalysis (CMM), University of
Queensland for providing electron microscope facilities
that enabled the microanalytical measurements to be under-
taken.
REFERENCES
1) N. L. Bowen and J. F. Schairer: Am. J. Sci., 29 (1935), 151.
2) R. L. David: CRC Handbook of Chemistry and Physics, 84th Ed.,
CRC Press, Boca Raton, Florida, (2003–2004), 4–62, 4–70.
3) A. Muan and E. F. Osborn: J. Am. Ceram. Soc., 39 (1956), 121.
4) J. W. Greig: J. Am. Sci., 13 (1927), 1.
5) N. L. Bowen and O. Anderson: Am. J. Sci., 37 (1914), 487.
6) C. M. Schlaudt and D. M. Roy: J. Am. Ceram. Soc., 48 (1965), 248.
7) S. Chen: PhD Thesis, The University of Queensland, Brisbane,
Australia, (2004).
8) P. Wu, G. Eriksson, A. D. Pelton and M. Blander: ISIJ Int., 33
(1993), 26.
9) N. L. Bowen and J. F. Schairer: Am. J. Sci. Ser., 24 (1932), 177.
10) W. C. Allen and R. B. Snow: J. Am. Ceram. Soc., 38 (1955), 264.
11) B. Zhao: PhD Thesis, The University of Queensland, Brisbane,
Australia, (1999).
796
 ISIJ International, Vol. 45 (2005), No. 6

12) C. W. Bale, P. Chartrand, S.A. Degterov, G. Eriksson, K. Hack, R. B. (1999), 1017.

Mahfoud, J. Melancon, A.D. Pelton and S. Petersen: Calphad, 26 19) E. Jak, B. Zhao and P. C. Hayes: Metall. Mater. Trans. B, 31B
(2002), 189. (2000), 1195.
13) S. A. Decterov, I. H. Jung and A. D. Pelton: J. Am. Ceram. Soc., 85 20) E. Jak and P. C. Hayes: Proc of 7th Int. Conf. on Molten Slags Fluxes
(2002), 2903. and Salts, SAIMM, Johannesburg, (2004), 209.
14) B. J. W. Karlemo and P. A. Taskinen: Scand. J. Metall., 31 (2002), 21) E. Jak, S. Degterov, P. Wu and P. C. Hayes: Metall. Mater. Trans. B,
307. 28B (1997), 1011.
15) A. T. Dinsdale, S. M. Hodson, T. I. Barry and J. R. Taylor: Computer 22) S. Chen, E. Jak and P. C. Hayes: Proc of Int. Laterite Nickel Symp.,
Software in Chemical and Extractive Metallurgy, Pergamon Press, ed. by W. P. Imrie and D. M. Lane, Charlotte, TMS Warrendale, PA,
New York, (1989), 59. (2004), 503.
16) E. Jak, H. G. Lee and P. C. Hayes: Kor. IMMJ, 1 (1995), 1. 23) C. M. Diaz, C. A. Landolt, A. Vahed, A. E. M. Warner and J. C.
17) E. Jak, N. Liu and P. C. Hayes: Metall. Mater. Trans. B, 29B (1998), Taylor: J. Met., 40 (1988), 28.
541. 24) G. B. Belton and R. J. Fruehan: J. Phys. Chem., 71 (1967), 1403.
18) B. Zhao, E. Jak and P. C. Hayes: Metall. Mater. Trans. B, 30B

797 © 2005 ISIJ