Communications
Thermodynamics on the Formation of
Spinel Nonmetallic Inclusion in Liquid
Steel
HIROYASU ITOH, MITSUTAKA HINO, and
SHIRO BAN-YA
The nonmetallic spinel inclusion, MgOzAl2O3, has a high
melting point and produces an undeformed C-type inclusion
in steel products. Therefore, it is very harmful in the pro-
duction of high grade wire, spring, and bearing steels. Thus,
it is necessary to develop a methodology by which we can
predict the occurrence of such defects.
The thermodynamics of the formation of spinel nonme-
tallic inclusions during co-deoxidation with aluminum and
magnesium is discussed in the present work. Previous lit-
erature thermodynamic values for strong deoxidizers such
as aluminum, magnesium, and calcium are not in good
agreement with the observed results and measurements.
This discrepancy between predicted and measured results
is due to an incomplete expression of the activities of dis-
solved oxygen and deoxidizers in current literature.
Namely, only first-order interaction parameters are avail-
able despite the very strong interactions among the dis-
solved elements.
In the present work, the deoxidation equilibrium with
magnesium in liquid iron was studied in order to evaluate
the activities of oxygen and dissolved magnesium and, in-
clude in the thermodynamic description, the first- and sec-
ond-order interaction parameters, including the
cross-product terms. A dolomite crucible was used to en-
hance the magnesium content in liquid iron in the present
experiment on thermodynamics of spinel formation with
reference to the study on desulfurization with a dolomite
crucible undertaken by Ototani et al.[1] The assessment of
aluminum-oxygen equilibrium in liquid iron reported by
other investigators was also reviewed to develop values for
the second-order interaction parameters. Finally, thermo-
dynamic data on calcium deoxidation equilibrium were ob-
tained.
The iron sample was melted by a high frequency induc-
tion furnace. Iron of about 90 grams was charged into a
crucible. The added weight of a deoxidizer such as mag-
nesium and calcium was 1 mass pct of the weight of iron.
The experimental procedure is briefly shown as follows.
After the iron melted under a gas flow of a mixture of
Ar-50 vol pct H2, hydrogen was impinged onto the melt
surface to reduce dissolved oxygen content until it was less
HIROYASU ITOH, formerly Research Student, Department of
Metallurgy, Graduate School of Engineering, Tohoku University, is
Researcher, Kuzuu Works, Yoshizawa Lime Industry, Tochigi-ken 327-
05, Japan. MITSUTAKA HINO, Professor, is with the Department of
Metallurgy, Graduate School of Engineering, Tohoku University,
Miyagi-ken 980-77, Japan. SHIRO BAN-YA, President, Akita National
College of Technology Iijima, Akita-ken 011, Japan, is also Professor
Emeritus, Department of Metallurgy, Graduate School of Engineering,
Tohoku University.
Manuscript submitted March 11, 1997.
METALLURGICAL AND MATERIALS TRANSACTIONS B
Fig. 1—Deoxidation equilibrium with Mg in liquid iron at 1873 K.
than 20 mass ppm in liquid iron, and the temperature was
kept at 1873 K for 4 hours. Then the atmosphere in the
furnace was converted to a Ar-2 vol pct H2 gas mixture.
The experimental temperature of liquid iron was kept at
1873 or 2023 K, and then the deoxidizer wrapped with iron
film, which was kept in the branch of the gas inlet tube
made by glass, was dropped into liquid iron by a magnet.
The iron sample was quenched by helium in the furnace
after reaching deoxidation equilibrium and was cut and pol-
ished for chemical analysis. The infrared absorbable
method was applied for quantitative analysis of the dis-
solved oxygen content. The quantitative analysis of dis-
solved magnesium, aluminum, and calcium contents was
made by induction coupled plasma spectroscopy. The con-
firmation of stably formed oxide phases was done using the
XRD method.
The deoxidation equilibrium of magnesium is repre-
sented by Eq. [1], and the equilibrium constant, KMg, is
expressed by Eq. [2]:
MgO (s) 5 Mg 1 O [1]
KMg 5 aMgzaO /aMgO [2]
where aMg and aO are the activities of magnesium and ox-
ygen, respectively, and their standard states are their infin-
itude dilute solution in liquid steel.
As a magnesia crucible was used in the present work,
aMgO 5 1. Therefore, Eq. [2] is transformed into Eq. [3].
KMg 5 aMgzaO 5 fMg[pct Mg]zfO[pct O] [3]
When activity coefficients were arranged by Wagner’s
relation, we introduced the first- and second-order interac-
tion parameters, including cross-product terms. Eventually,
the equilibrium constant is derived as in Eq. [4] by means
of Lupis’s relation:[2]
log KMg 5 log K'Mg
1 eO(Mg) ([pct Mg] 1 1.52[pct O]
2 0.0252[pct Mg][pct O]) 1 eO(O)[pct O] [4]
1 rO(Mg) ([pct Mg]2 1 3.04[pct Mg][pct O])
1 rMg(O)([pct O]2 1 1.32[pct Mg][pct O])
Equation [4], including the first- and second-order inter-
action parameters, is considered to be the fundamental
equation of magnesium deoxidation equilibrium in liquid
iron.
VOLUME 28B, OCTOBER 1997—953
 Table I. Deoxidation Equilibria of Magnesium, Calcium, and Aluminum in Liquid Iron

Mg Deoxidation Ca Deoxidation Al Deoxidation

Reaction MgO (s) 5 Mg 1 O CaO (s) 5 Ca 1 O Al2O3 (s) 5 2Al 1 3O
Equilibrium constant log KMg 5 24.28 2 4,700/T log KCa 5 23.292 2 7,220/T log KAl 5 11.62 2 45,300/T
First-order eO(O) 5 0.76 2 1,750/T [5] eO(O) 5 0.76 2 1,750/T [5] eO(O) 5 0.76 2 1,750/T [5]
interaction eO(Mg) 5 630 2 1.705 3 106/T eO(Ca) 5 627 2 1.755 3 106/T eO(Al) 5 1.90 2 5,750/T
parameter eMg(O) 5 958 2 2.592 3 106/T eCa(O) 5 1570 2 4.405 3 106/T eAl(O) 5 3.21 2 9,720/T
eMg(Mg) 5 0 eCa(Ca) 5 0 eAl(Al) 5 80.5/T[5]
Second-order rO(Mg) 5 70,500 2 1.696 3 108/T rO(Ca) 5 36,100 2 1.013 3 108/T rO(Al) 5 0.0033 2 25.0/T
interaction rMg(O) 5 21.904 3 106 1 4.222 3 109/T rCa(O) 5 22.596 3 106 1 6.080 3 109/T rAl(O) 5 2107 1 275,000/T
parameter rO(Mg,O) 5 22.513 3 106 1 5.573 3 109/T rO(Ca,O) 5 22.077 3 106 1 4.864 3 109/T rO(Al,O) 5 127.3 1 327,300/T
rMg(O,Mg) 5 2.143 3 105 2 5.156 3 108/T rCa(O,Ca) 5 1.809 3 105 2 5.075 3 108/T rAl(O,Al) 5 20.021 2 13.78/T

Fig. 2—Deoxidation equilibrium with Ca in liquid iron at 1873 K.

The magnesium deoxidation equilibrium relation that

was obtained at 1873 K in our work[3,4] is illustrated in
Figure 1, and the numerical values are expressed with Lu-
pis’s relation in Table I.
The present values do not show a discrepancy with the
experimental data, as shown in Figure 1. The introduction
of the first- and second-order interaction parameters is a
reasonable manner by which we can express numerically
the experimental data on magnesium deoxidation equilib-
rium.
The equilibrium constant and interaction parameters on
calcium deoxidation were also determined in the present
work. As a result, the thermodynamic values were deter-
mined, as shown in Table I.
The equilibrium constant and interaction parameters on
aluminum deoxidation were determined based on the pre-
vious literature.[5–8] These are also shown in Table I.
The deoxidation equilibrium relation concerning calcium
and aluminum at 1873 K is shown in Figures 2 and 3,
respectively. The obtained equilibrium relation between
calcium and oxygen is comparable with the experimental
data in the present work and also in the previous studies
by other investigators, as shown in Figure 2. The equilib-
rium relation assessed in the present work corresponds with
Fig. 3—Deoxidation equilibrium with Al in liquid iron at 1873 K.
the experimental data, especially that of Rohde et al.[6] and
Jacobson and Mehrotra,[7] as shown in Figure 3. The ther-
modynamic values obtained at 1873 K in the present work
are summarized in Table II.[3,4]
When liquid iron is deoxidized in MgO lined ladles with
aluminum, MgO, MgOzAl2O3, and/or Al2O3 appear as deox-
idation products in melt at steelmaking temperatures. The
stability diagram of MgO, MgOzAl2O3, and Al2O3 at steel-
making temperature as a function of mass percent of mag-
nesium and aluminum in liquid iron is shown in Figure 4.
The thermodynamic values obtained in the present work
and the oxide activities of aMgO, aMgOzAl2O3, and aAl2O3, assum-
ing the ideal solid solution in the system MgO-Al2O3, are
used to evaluate the phase boundaries. Calculation of the
boundary lines between MgO and MgOzAl2O3 and between
MgOzAl2O3 and Al2O3 are based on Eqs. [5] through [8].
MgO (s) 1 2Al 1 3O 5 MgOzAl2O3 (s) [5]
DG75 5 2887,960 1 210.88T (J/mol) [6]

954—VOLUME 28B, OCTOBER 1997 METALLURGICAL AND MATERIALS TRANSACTIONS B

 Table II. Thermodynamic Values Determined in the Present Work at 1873 K

Reaction log KM e O(M) eM

(O)
e (M)
M e (O)
O r (O)
M r O(M) r (M,O)
O rM
(M,O)
r (Mg,Al)
O r Mg
(Al,O)
r Al
(Mg,O)
r O(Al,Ca) r (Ca,Mg)
O

MgO (s) 5 Mg 1 O 26.80 2280 2430 0 20.17 350,000 220,000 462,000 261,000
CaO (s) 5 Ca 1 O 27.15 2310 2780 0 20.17 650,000 218,000 520,000 290,000
Al2O3 (s) 5 2Al 1 3O 212.57 21.17 21.98 0.04 20.17 40 20.01 47 20.028
MgOzAl2O3 (s) 5
MgO (s) 1 Al2O3 (s) 21.18 2150 2230 2260 0 0

Fig. 5—Stability diagram of MgO, MgOzAl2O3, and CaOz2Al2O3, and iso-

Fig. 4—Stability diagram of MgO, MgOzAl2O3, and Al2O3 inclusions, and oxygen contour lines at 1873 K with experimental data, where [mass ppm
iso-oxygen contour lines calculated at 1873 K with experimental data. Ca] 5 1.

Al2O3 (s) 1 Mg 1 O 5 MgOzAl2O3 (s) [7]

DG77 5 2110,720 2 93.51T (J/mol)
The free energy changes of Eqs. [5] and [7] are derived
from those of magnesium deoxidation and aluminum deox-
idation, which are evaluated in the present work, and of the
formation of MgOzAl2O3 (s) from MgO (s) and Al2O3 (s),
as reported by Knacke et al.[9]
MgO (s) 1 Al2O3 (s) 5 MgOzAl2O3 (s)
DG79 5 220,740 2 11.57T (J/mol)
The phase boundary of MgO-MgOzAl2O3-Al2O3 and the
experimental results, including the present work and that of
others, are shown at 1873 K in Figure 4, in which the is-
oconcentration lines of dissolved oxygen in liquid iron are
illustrated. The following thermodynamic values obtained
at 1873 K in the present work are used in Figure 4 for the
calculation of activities of dissolved elements in liquid iron.
rO(Mg,Al) 5 2150
rMg(O,Al) 5 2230
rAl(O,Mg) 5 2260 [13]
[8] The chemistry and type of oxides, which were confirmed
by XRD, formed between the quenched iron sample and
the crucible are shown in Figure 4 with the quantitative
values of dissolved magnesium, aluminum, and oxygen
contents in the sample. The observed results show a good
agreement with the evaluated stability phase diagram.
Moreover, the chemically analytical concentration of oxy-
[9] gen showed a good agreement with the calculated iso-ox-
ygen contour lines.
[10] MgOzAl2O3 equilibrates with MgO and CaOz2Al2O3 at
steelmaking temperatures when liquid iron containing a few
mass parts per million of calcium coexists with the oxide
system. Therefore, the phase boundary between MgO and
MgOzAl2O3 was determined in the aforementioned manner
as the co-deoxidation of the Al-Mg-Ca-O system. The
phase boundary between MgOzAl2O3 and CaOz2Al2O3 was
evaluated by Eq. [14] in the present work, where the free
energy change of Eq. [14] is derived from those of mag-
nesium deoxidation and calcium deoxidation, which are
[11] evaluated in the present work, and of the formation of
CaOz2Al2O3 (s) from CaO (s) and Al2O3 (s), as reported by
[12] Nagata et al.[10]

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 28B, OCTOBER 1997—955

 2Mg 1 O 1 CaOz2Al2O3 (s)
[14]
5 2MgOzAl2O3 (s) 1 Ca
DG714 5 298,870 2 149.82T (J/mol) [15]
CaO (s) 1 2Al2O3 (s) 5 CaOz2Al2O3 (s) [16]
DG716 5 15,650 1 25.82T (J/mol) [17]
These equilibrium oxides, however, are harmful inclusions
in produced steel, so the stability area of liquid nonmetallic
inclusion formed by deoxidation has to be shown from the
viewpoint of shape control of nonmetallic inclusions.
The activity of each constituent in the CaO-Al2O3-MgO
system, such as aAl2O3, aMgO, or aCaO, was quoted in the cal-
culation from our previous study,[11,12] where the regular
solution model was applied to evaluate their activities. The
calculated stability phase diagram is shown in Figure 5. The
evaluated liquid phase area agreed well with the liquidus
lines observed in our previous work.[11,12]
The main findings of this study on the thermodynamics
of spinel nonmetallic inclusion formation in liquid iron can
be summarized as follows.
(1) The equilibrium constants and interaction parameters
on calcium, magnesium, and aluminum deoxidation
in liquid iron at steelmaking temperature must be
expressed by the first-and second-order interaction
parameters, including cross-product terms.
(2) Stability phase diagrams of MgO-MgOzAl2O3-Al2O3
and MgO-MgOzAl2O3-CaOz2Al2O3 were drawn at
steelmaking temperatures as a function of dissolved
magnesium, aluminum, calcium, and oxygen con-
tents in liquid iron.
(3) The liquid oxide phase area in the stability phase
diagram of the MgO-Al2O3-CaO ternary system was
determined at steelmaking temperatures.
This study was sponsored by a Grant-in-Aid for Iron and
Steel Research from The Iron and Steel Institute of Japan
in 1994 and 1995. The authors are grateful to Yoshizawa
Lime Industry for providing one of the authors (Y1) with
financial support during his stay at Tohoku University.
They wish to thank former undergraduate students, Messrs.
H. Suzuki (now with Chuo Spring Co., Ltd.) and T. Suzuki
(now with MEMC), from Tohoku University for their ef-
fective assistance in making the experiments.
REFERENCES
1. T. Ototani, Y. Kataura, and T. Degawa: Tetsu-to-Hagané, 1975, vol.
61, pp. 2167-81.
2. C.H. Lupis: Chemical Thermodynamics of Materials, North-Holland,
Amsterdam, 1983, pp. 254 and 256.
3. Y. Itoh, M. Hino, and S. Ban-ya: CAMP-ISIJ, 1995, vol. 8, pp. 75-
78; Y. Itoh: Ph.D. Theses, Tohoku University, Sendai, 1996.
4. M. Hino, H. Itoh, and S. Ban-ya: Japan-US Joint Seminar, Clean
Steel for the 21st Century: Fundamental Issues, Futtsu, Chiba, Japan,
Apr. 25–27, 1996, The Japan Society for The Promotion of Science
and The National Science Foundation, Tokyo, 1996, pp. 125-30.
5. Steelmaking Data Sourcebook, The Japan Society for the Promotion
956—VOLUME 28B, OCTOBER 1997
of Science, The 19th Committee on Steelmaking, Gordon and Breach
Science Publishers, New York, NY, 1988, p. 45.
6. L.E. Rohde, A. Choudhury, and M. Wahlster: Arch. Eisenhuttenwes.,
1971, vol. 42, pp. 165-67.
7. N.S. Jacobson and G.M. Mehrotra: Metall. Trans. B, 1993, vol. 24B,
pp. 484-86.
8. N.A. Gokcen and J. Chipman: J. Met., 1953, vol. 5, pp. 173-78.
9. O. Knacke, O. Kubaschewski, and K. Hesselmann: Thermochemical
Properties of Inorganic Substances II, 2nd ed., Springer-Verlag,
Berlin, 1991, p. 1171.
10. K. Nagata, J. Tanabe, and K.S. Goto: Tetsu-to-Hagané, 1989, vol. 75,
pp. 2023-30.
11. M. Hino and S. Ban-ya: 14th PTD Conf. Proc., Orlando, FL, 1995,
vol. 14, pp. 47-52.
12. M. Hino, S. Kinoshita, Y. Ehara, H. Itoh, and S. Ban-ya: Proc. 5th
Int. Conf. on Molten Slags, Fluxes and Salts ’97, Sydney, 1997,
ISS-AIME, Warrendale, PA, 1997, pp. 53-57; Y. Ehara: Masters
Theses, Tohoku University, Sendai, 1994.
Thermodynamics of Calcium and
Oxygen in Molten Ti3Al
YOSHINAO KOBAYASHI and FUMITAKA
TSUKIHASHI
The solubility of Ca and O in molten Ti3Al equilibrated
with solid CaO has been measured at 2003 and 2053 K.
The equilibrium constant of reaction CaO (s) 5 Ca (mass
pct in Ti3Al) 1 O (mass pct in Ti3Al) and the interaction
parameter between Ca and O were determined. The stan-
dard Gibbs energy of reaction was obtained as follows:
DG7 5 258,000 2 98.0T J/mol (2003 to 2053 K)
The deoxidation of Ti3Al using Ca bearing fluxes is dis-
cussed in comparison with that of other Ti-Al alloys.
The improvement of mechanical properties of titanium-
aluminum alloy is of growing importance in the aircraft and
space industry. Eliminating oxygen from the alloy is im-
portant for practical applications, because the oxygen de-
grades the mechanical and physical properties. Calcium has
been considered to be a promising reagent for the deoxi-
dation of refractory metals because of its strong affinity for
oxygen. Thermodynamic properties of calcium and oxygen
in molten Ti and Ti-Al alloys have been investigated[1] us-
ing a chemical equilibrium technique. In this study, ther-
modynamic properties of calcium and oxygen in molten
Ti3Al equilibrated with solid CaO at 2003 and 2053 K were
observed to discuss the deoxidation of titanium-aluminum
alloys using calcium-based fluxes.
The experimental procedure is the same as that described
in a previous article.[1] Samples were a mixture of Ti and
TiAl or Al with the composition of Ti3Al, as shown in
Table I. Five grams of sample were equilibrated in a CaO
crucible (22-mm o.d., 18-mm i.d., and 60-mm height) man-
ufactured by Yoshizawa Lime Industry Co., Ltd, Tokyo,
Japan. A CaO lid was adhered with zirconia cement on the
top of the crucible to prevent the evaporation loss of cal-
YOSHINAO KOBAYASHI, Graduate Student, and FUMITAKA
TSUKIHASHI, Associate Professor, are with the Department of
Metallurgy, The University of Tokyo, Tokyo 113, Japan.
Manuscript submitted October 30, 1996.
METALLURGICAL AND MATERIALS TRANSACTIONS B