Composition Changes of Inclusions by Reaction with Slag and Refractory: A Review

Abstract

Precise control on inclusions is of great importance for improving steel quality. In secondary refining, three types of inclusion are generally observed in Al-killed steel: Al₂O₃, MgO·Al₂O₃ spinel, and CaO–Al₂O₃ type. Many researchers have reported on inclusions transformed in the routine of Al₂O₃ → MgO·Al₂O₃ spinel → CaO–Al₂O₃ during secondary refining in Al-killed steel. The deoxidizer of Al is intentionally added to the steel, and Al₂O₃ inclusions are formed as a deoxidation product. However, MgO·Al₂O₃ spinel and CaO–Al₂O₃-type inclusions have been observed even without any intentional Mg or Ca addition. Therefore, it is important to clarify the source of Mg and Ca causing MA spinel and CaO–Al₂O₃-type inclusions formation, in order to control the compositions of the inclusions. Regarding to this phenomenon, a good review was published by Park and Todoroki in 2010. Several studies have since been conducted. This paper summarized the research activities on the composition changes of inclusions during secondary refining from the aspect of thermodynamics and kinetics.

1. Introduction

Inclusions play a crucial role in determining steel quality because their properties, such as plasticity and thermal expansion, differ from those of a steel matrix: they usually act as a stress raiser and a source of cracks.^1,2) In general, inclusions are harmful for steel quality, except in certain circumstances where inclusions with extremely small sizes can be utilized as nucleation sites for phase transformation.^3,4) In addition, the solid inclusion at steelmaking temperatures, such as Al₂O₃, MgO·Al₂O₃ spinel (MA spinel), or CaS, is easy to sinter with the Al₂O₃–C nozzle material, causes clogging, and disturbs the casting process.^5,6) Therefore, inclusions have harmful effects both on steel products and operation.

Generally, inclusions form in two ways. In the first case, inclusions are formed by entrapped slag droplets, refractory particles, or the contamination of steel; these are known as exogenous inclusions. In the other case, inclusions are originally formed as deoxidation products; these are known as endogenous inclusions. The formation of exogenous inclusions can be suppressed by optimizing the refractory lining, optimizing the meniscus flow of steel, inhibiting reoxidation, etc. On the other hand, as endogenous inclusions are formed by the intentional addition of a deoxidizer, such as Al or Si, to steel, it is impossible to suppress their formation. Therefore, by enhancing the flotation through control of the fluid flow in the ladle, tundish, and mold, large-size inclusions can be eliminated. In some grades of steel, the composition of inclusions must be controlled in addition to their number and size. In this case, the control of their formation reaction is considered to be one of the most important measures.

During secondary refining, three types of inclusion are generally observed in Al-killed steel: Al₂O₃, MA spinel, and CaO–Al₂O₃ type. Different types of inclusions have different physical properties and affect the steel quality in different ways.

Al₂O₃ has the properties of high melting temperature (2326 K), high hardness (386 GPa), and a large contact angle with molten steel (136°).^7,8) Therefore, Al₂O₃ inclusions are easy to sinter with each other and form clusters of large sizes. Al₂O₃-type inclusions are harmful to the surface quality and fatigue resistance.⁹⁾ In addition, clogging by Al₂O₃ inclusions during casting has been widely reported.⁵⁾

The similar crystal structure of MA spinel to that of Al₂O₃ can explain its relative high hardness (273 GPa), high melting point (2408 K), and large contact angle with molten steel (139.7°).^10,11) MA spinel inclusions with high hardness can be a detriment to the quality of the final product, including the surface quality and fatigue resistance.⁶⁾ In addition, MA spinel inclusions are easy to sinter with the nozzle material, which is mainly made of Al₂O₃.⁶⁾

The system of CaO–Al₂O₃ has some stoichiometric compounds, which can be called calcium aluminates.⁵⁾ The phases C₁₂A₇ and CA (C=CaO and A=Al₂O₃) have the lowest melting points and exist as liquid inclusions at steelmaking temperatures. As a low-melting-point inclusion is deformable at a hot rolling temperature, these inclusions are called soft inclusions, compared with solid inclusions that do not deform significantly and are called hard inclusions. These soft CaO–Al₂O₃-type inclusions are usually the objective of inclusion control in some grades of steel in which hard inclusions must be avoided. However, soft CaO–Al₂O₃-type inclusions are easy to elongate during rolling, and inclusions with small sizes affect the fatigue resistance.¹²⁾ In addition, CaO–Al₂O₃-type inclusions are easy to dissolve into a water solution and act as the start point of corrosion.^13,14)

Many researchers have reported inclusions transformed in the routine of Al₂O₃ → MA spinel → CaO–Al₂O₃ during secondary refining in Al-killed steel.^{15,16,17,18,19,20)} The deoxidizer of Al is intentionally added to steel, and Al₂O₃ inclusions are formed as deoxidation products. However, MA spinel and CaO–Al₂O₃-type inclusions have been observed even without the intentional addition of Mg and Ca during secondary refining. Therefore, it is necessary to clarify the source of Mg and Ca for the formation of MA spinel and CaO–Al₂O₃-type inclusions in order to control the composition of the inclusions.

With regard to this phenomenon, a good review was published by Park and Todoroki in 2010.⁶⁾ Many studies have since been conducted. Therefore, this paper summarizes the research activities on the composition changes of inclusions during secondary refining from the viewpoints of thermodynamics and kinetics.

2. Formation of MA Spinel

2.1. Composition Change from Al₂O₃ to MA

Several studies have been conducted to clarify the formation mechanism of MA spinel inclusions. To simplify, MA spinel inclusions can form in two ways. In the first case, MA spinel is caused by entrapped slag droplets or ladle glaze; these are known as exogenous inclusions. In the other case, MA spinel is formed by chemical reactions within the steel and existing Al₂O₃ inclusions; these are known as endogenous inclusions.

The formation of MA spinel inclusions was reported in 1982 by Sunami et al.²¹⁾ for Si-killed steel during the EAF-LF-CC process. They concluded that this phase precipitated from the entrapped slag droplets of the CaO–SiO₂–Al₂O₃–MgO system. As the temperature decreased during the LF-CC process, MA spinel could be formed by the phase separation from this oxide system. This phenomenon was also observed and confirmed in the production of Al-killed steel by Park et al.²²⁾ and Ehara et al.^23,24) Ehara et al.²⁴⁾ investigated the precipitation behavior of MA spinel from CaO–SiO₂–Al₂O₃–MgO slag during cooling from 1773 to 1273 K at a cooling rate of 0.0167 K/s, and the precipitated phase of MA spinel was observed. Son et al.²⁵⁾ and Du et al.^26,27) studied the reactions between glazed refractory and Al-deoxidized steel. They found that the refining slag that adhered on the surface of the refractory (ladle glaze) separated into MA spinel and CaO–Al₂O₃ phase during cooling. They considered that the formation of MA spinel and CaO–Al₂O₃-type inclusions was caused by the entrapped ladle glaze.

In general, the formation of MA spinel by the exogenous process occurred with CaO–Al₂O₃-type or CaO–SiO₂-type inclusions simultaneously. However, in some cases of industrial production, only MA spinel and Al₂O₃ inclusions were detected several minutes after Al deoxidation during secondary refining. Therefore, the MA spinel formed via an endogenous process, has to be considered.

In the following text, the latest publications regarding MA spinel inclusions generated through the endogenous process are summarized. Thermodynamic data for the formation of MA spinel are discussed regarding the equilibrium constant of the Mg–O reaction. The Mg source for the dissolved Mg in steel is then explained, including MgO-containing slag and the MgO-based refractory. Furthermore, the formation reaction of MA spinel is shown, and the formation kinetics of MA spinel are discussed. Finally, proposed countermeasures for the control of MA spinel are summarized.

2.2. Thermodynamic Data for Formation of MA Spinel

The thermodynamics for the formation of MA spinel inclusions are composed of the following reactions:   

$$\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)=2\left[\mathrm{A}\mathrm{l}\right]+3\left[\mathrm{O}\right]$$(1)

  

$$\mathrm{M}\mathrm{g}\mathrm{O}\left(\mathrm{s}\right)=\left[\mathrm{M}\mathrm{g}\right]+\left[\mathrm{O}\right]$$(2)

  

$$\mathrm{M}\mathrm{g}\mathrm{O}\left(\mathrm{s}\right)+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)=\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)$$(3)

For the equilibrium constant of Eq. (1), K_Al, the results obtained by different studies showed small differences, and the recommended value of the Japan Society for Promotion of Science (JSPS) was proposed.²⁸⁾ However, the value of the JSPS recommendation has a narrow range of Al content (0.002 < [%Al] < 0.08) for application. Itoh et al.²⁹⁾ proposed a value of K_Al, which was determined by various data together with JSPS and a wider range of Al content ([%Al] < 10), as shown in Eq. (4).   

$$log{K}_{\mathrm{A}\mathrm{l}}=-\frac{45\mathrm{\hspace{0.17em} }300}{T}+11.62$$(4) ²⁹⁾

For the equilibrium constant of Eq. (2), K_Mg, the measured values after 1980 are summarized in Table 1.^{30,31,32,33,34,35,36,37,38,39,40)} The relation of Mg content and O content is shown in Fig. 1.⁴⁰⁾ For this calculation, the interaction parameters listed in Table 2 were used.^{30,31,32,33,34,35,36,37,40)} The results obtained by different studies varied widely, and the value of the JSPS recommendation was not published.

Table 1. List of equilibrium constants for Mg–O reaction.

Years	Researcher	logKMg at 1873 K	Determining method
1980	A. P. Gorobetz[30]	−9.24	–
1985	I. S. Kulikov[31]	−8.54	–
1986	M. Nadif et al.[32]	−5.7	MgO crucible, solubility product using EMF oxygen sensor
1994	R. Inoue et al.[33]	−7.8	1823–1923 K, MgO crucible, Mg alloy, and CaO–Al2O3–MgO slag
1996	H. Ohta et al.[34]	−7.86	1873 K, MgO crucible, CaO–SiO2–Al2O3–MgO slag
1997	H. Itoh et al.[35]	−6.8	1823–1923 K, MgO crucible
1997	Q. Han et al.[36]	−6.03	1823–1923 K, MgO crucible, Mg vapor
2000	J. D. Seo et al.[37]	−7.21	1873 K, MgO crucible, Mg vapor
2003	W. G. Seo et al.[38]	−7.24	1873 K, MgO and Al2O3 crucible, Mg alloy (steel equilibrated with MgO∙Al2O3 spinel)
2011	J. Gran et al.[39]	−8.07	1823 K, MgO crucible, Fe and Mg–Al alloy equilibrium

Fig. 1.

Equilibrium relation between Mg content and O content.⁴⁰⁾

Table 2. Interaction parameters to calculate Fig. 1.

Researcher	$e_{\text{O}}^{\text{Mg}}$	$e_{\text{Mg}}^{\text{O}}$	$e_{\text{O}}^{\text{O}}$	${\gamma }_{\text{O}}^{\text{Mg}}$	${\gamma }_{\text{Mg}}^{\text{O}}$	${\gamma }_{\text{Mg}}^{\text{O},\text{Mg}}$	${\gamma }_{\text{O}}^{\text{Mg},\text{O}}$
A. P. Gorobetz[30]	−250	−380	−0.17
I. S. Kulikov[31]	−160	−240	−0.17
R. Inoue et al.[33]	−190	−290	−0.17
H. Ohta et al.[34]	−300	−460	−0.17	16000	37000	48000	48000
H. Itoh et al.[35]	−280	−430	−0.17	−20000	350000	−61000	462000
Q. Han et al.[36]	−115	−175	−0.17
J. D. Seo et al.[37]	−370	−560	−0.2	5900	145000	17940	191400

The method to write a stability phase diagram of MgO–MA spinel-Al₂O₃ was given elsewhere.⁴¹⁾ To calculate the stability diagram, interaction coefficients are needed; these are listed in Table 3.^{29,35,40,41,42,43)} In Table 3, all data without notations are from ref. [40]. The calculated stability diagram for an Fe-11mass%Cr–Al system by different values of K_Mg is shown in Fig. 2. The calculated critical content of Mg for the formation of MA spinel and MgO varied greatly based on different values of K_Mg. Therefore, there is no recommendation value for the Mg–O reaction, and a more accurate value of K_Mg is needed in future work.

Table 3. List of interaction coefficients to calculate Fig. 2.

$e_{\text{i}}^{\text{j}}$	j i	Mg	Al	O	Cr
Mg		0	−0.27[42]	−3	0.022[42]
Al		−0.3[42]	0.0429	−1.98	0.0122[43]
O		−1.98	−1.17	−0.174	−0.052
${\gamma }_{\text{Al}}^{\text{O}}={39.82}^{\left[29\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{Al}}^{\text{Al},\text{O}}=-{0.02836}^{\left[29\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{Al}}^{\text{Mg},\text{O}}=-{260}^{\left[41\right]};$ ${\gamma }_{\text{Mg}}^{\text{O}}=350\mathrm{\hspace{0.17em} }{000}^{\left[35\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{Mg}}^{\text{Mg},\text{O}}=-61\mathrm{\hspace{0.17em} }{000}^{\left[35\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{Mg}}^{\text{Al},\text{O}}=-{230}^{\left[41\right]};$ ${\gamma }_{\text{O}}^{\text{Al}}=-{0.01}^{\left[29\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{O}}^{\text{Al},\text{O}}=-{302.046}^{\left[29\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{O}}^{\text{Cr}}=-0.000576;$ ${\gamma }_{\text{O}}^{\text{Mg}}=-20\mathrm{\hspace{0.17em} }{000}^{\left[35\right]};\mathrm{\hspace{0.17em} } {\gamma }_{\text{O}}^{\text{Mg},\text{O}}=462\mathrm{\hspace{0.17em} }{000}^{\left[35\right]};$

Fig. 2.

Calculated MgO/MA spinel/Al₂O₃ stability diagram at 1873 K.

For reaction (3), the value of K_spinel was only reported by Fujii et al.,⁴⁴⁾ as shown in Eq. (5).   

$$log{K}_{\mathrm{s}\mathrm{p}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{l}}=\frac{1\mathrm{\hspace{0.17em} }086}{T}+0.82$$(5) ⁴⁴⁾

2.3. Mg Sources for Formation of MA Spinel

As mentioned previously, Al₂O₃ inclusions are formed as deoxidation products, while MA spinel is formed spontaneously without the intentional addition of Mg. Recently, many studies were conducted to clarify the Mg sources for the formation of MA spinel, as shown in Table 4.^{15,45,46,47,48,49,50,51,52,53,54,55)} The Mg sources are mainly divided into two types: MgO-containing slag and MgO-based refractory. The supply mechanism of Mg is shown in reaction (6).   

$$4{\left(\mathrm{M}\mathrm{g}\mathrm{O}\right)}_{\mathrm{s}\mathrm{l}\mathrm{a}\mathrm{g}\mathrm{\hspace{0.17em} } \mathrm{o}\mathrm{r}\mathrm{\hspace{0.17em} } \mathrm{r}\mathrm{e}\mathrm{f}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{r}\mathrm{y}}+2\left[\mathrm{A}\mathrm{l}\right]=\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3+3\left[\mathrm{M}\mathrm{g}\right]$$(6)

Table 4. Studies related to Mg source for MA spinel inclusion formation.

	Years	Researcher	Mg source	Experimental variable
Reaction with slag	1998	T. Nishi et al.[45]	CaO–SiO2–Al2O3–MgO slag (MgO crucible)	Reaction time, CaO/SiO2 (1.2–2.4)
	2000	G. Okuyama et al.[46]	CaO–SiO2–Al2O3–MgO slag (MgO crucible)	Reaction time, CaO/SiO2 (1.8–11), CaO/Al2O3 (1.0–2.3)
	2003	H. Todoroki et al.[15,47]	CaO–SiO2–Al2O3–MgO–CaF2 slag (MgO crucible)	Reaction time, SiO2 content (0–10 mass%)
	2008	M. Jiang et al.[48]	CaO–SiO2–Al2O3–MgO slag (MgO crucible)	Slag composition
	2014	A. Harada et al.[49]	CaO–Al2O3–MgO slag (MgO crucible)	Reaction time with and without slag
	2016	C. Liu et al.[50]	CaO–Al2O3–MgO slag (Al2O3 crucible)	Reaction time
Reaction with Refractory	1996	V. Brabie et al.[51]	MgO–C refractory	–
	2006	S. Jansson et al.[52]	MgO–C refractory	Temperature, reaction time, Al content, rotation speed
	2014	A. Harada et al.[49]	Pure MgO refractory	Reaction time, rotation speed
	2016	C. Liu et al.[53]	MgO–C refractory	Reaction time
	2018	C. Liu et al.[54]	MgO–C refractory	Reaction time, Al content in steel, C content in refractory
	2018	C. Liu et al.[55]	MgO–Cr2O3 refractory	Reaction time, Al and Cr content in steel

Nishi et al.⁴⁵⁾ and Okuyama et al.⁴⁶⁾ investigated the supplying behavior of Mg from refining slag to Al-killed steel. They found that the Mg content in steel increased with the reaction time, as shown in Fig. 3,⁴⁶⁾ and MA spinel was observed in the steel after the reaction. Some researchers^{15,47,48,49,50)} conducted similar experiments with slag of different compositions, and the supplying behavior of Mg from slag was confirmed. However, these experiments were conducted using a MgO crucible, which is a potential Mg source for the formation of MA spinel; therefore, the ability of Mg supply from slag could not be explained exclusively. Harada et al.⁴⁹⁾ studied the supplying behavior of Mg using a MgO crucible with and without slag. The supply of Mg from both the crucible and slag was confirmed. Liu et al.⁵⁰⁾ studied using an Al₂O₃ crucible in which the supply of Mg from the crucible was strictly controlled. They found that the dissolved Mg content in steel gradually increased with the reaction time. From the above analysis, it was concluded that MgO in slag was reduced by Al in steel and supplied Mg to steel.

Fig. 3.

Changes in steel composition after reaction between slag and steel.⁴⁶⁾

During the past two decades, the supplying behavior of Mg from MgO-based refractory has gained great attention. Brabie et al.⁵¹⁾ and Jansson et al.⁵²⁾ investigated the mechanism of the reaction between MgO–C refractory and Al-killed steel. They found that after the reaction, inclusions of MgO and MA spinel formed in steel, and they concluded that MgO–C refractory acted as the Mg source for the formation of MA spinel. However, the information about the Mg content in steel was limited because they mainly focused on the dissolution behavior of the refractory. Harada et al.⁴⁹⁾ investigated the dissolution behavior of Mg into steel from a pure MgO crucible and found that the Mg content in steel increased gradually with the reaction time, as shown in Fig. 4.⁴⁹⁾ In addition, on the surface of the immersed MgO rod, a layer of MA spinel formed. Liu et al.⁵³⁾ studied the dissolution behavior of Mg from commercial MgO–C refractory into Al-killed steel. They found that the Mg content in steel increased gradually and then reached a steady state. From the above analysis, both the MgO-containing refining slag and MgO-based refractory act as Mg sources for the formation of MA spinel.

Fig. 4.

Changes in steel composition after MgO refractory-steel reaction.⁴⁹⁾

Recently, the dissolution behavior of Mg from MgO-based refractory that is employed in industrial production was examined by Liu et al.^54,55) They immersed a rod made with commercial MgO–Cr₂O₃ refractory or commercial MgO–C refractory into Al-killed steel using an Al₂O₃ crucible.

In the case of MgO–Cr₂O₃ refractory, the Cr₂O₃ in the refractory was reduced by the Al in the steel, and an MgO·Al₂O₃ spinel layer was generated on the surface. The surface of the formed layer consisted of Al₂O₃-saturated MgO·Al₂O₃ spinel, and the interface between the MgO–Cr₂O₃ refractory and the formed layer consisted of MgO-saturated MgO·Al₂O₃ spinel, as shown in Fig. 5.⁵⁵⁾ The Mg content after immersion for 60 min was lower than 1 ppm independent of the Al content in steel, but the Al₂O₃ initially present in the steel did not transform into MgO·Al₂O₃ spinel. They concluded that the formation of Al₂O₃-saturated spinel on the surface suppressed both the dissolution of Mg and the composition transfer of inclusions. By the same procedure, they immersed MgO–C refractory by changing its C content into molten steel with various Al content. In this case, Mg gradually dissolved into the steel, and MgO·Al₂O₃ spinel inclusions were generated regardless of the C content in the refractory and the Al content in the steel. At the interface between the refractory and steel, a MgO·Al₂O₃ spinel layer was partially formed, but at the other interface, the MgO was directly in contact with the steel. The MgO in the refractory was reduced by both the Al in the steel and the C in the refractory, independently, as shown in Fig. 6.⁵⁴⁾

Fig. 5.

Composition of surface layer formed on immersed refractory.⁵⁴⁾

Fig. 6.

Variation of Mg content in steel by changes in C content in refractory and Al content in steel.⁵⁴⁾ Horizontal axis of lower figure shows carbon content in refractory, and other axis shows solute content in steel. (Number after “G” indicates level of carbon content in refractory, and “High” and “Low” indicate levels of Al content in steel.)

2.4. Formation Reaction for MA Spinel in Steel

After the dissolution of Mg from the MgO source into the steel, the Mg content in the steel increased, and MA spinel was formed. Harada et al.⁴⁹⁾ summarized the formation reaction of MA spinel as follows:   

$$\left[\mathrm{M}\mathrm{g}\right]+\left[\mathrm{O}\right]+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3=\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$$(7)

  

$$\mathrm{M}\mathrm{g}\mathrm{O}+2\left[\mathrm{A}\mathrm{l}\right]+3\left[\mathrm{O}\right]=\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$$(8)

  

$$\left[\mathrm{M}\mathrm{g}\right]+2\left[\mathrm{A}\mathrm{l}\right]+4\left[\mathrm{O}\right]=\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$$(9)

Based on thermodynamic calculations, only reaction (8) had a large negative value, as shown in Fig. 7.⁴⁹⁾ However, the spinel formation by this reaction did not occur easily after the formation of the spinel layer, as the direct contact between the MgO and molten steel was restricted by this layer.

Fig. 7.

Standard free-energy change in formation calculated for each reaction.⁴⁹⁾

In the actual refining process, inclusions of pure MgO are normally not observed in the steel prior to the addition of Al. Therefore, in addition to reaction (8), reaction (10) should be considered.   

$$3\left[\mathrm{M}\mathrm{g}\right]+4\left(\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\right)=3\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3+2\left[\mathrm{A}\mathrm{l}\right]$$(10)

Liu et al.⁵³⁾ confirmed by experiment that reaction (8) and reaction (10) both occurred and generated MA spinel at steelmaking temperatures. In their experiment, the MgO content in the inclusions gradually increased, and the initial Al₂O₃ finally changed into MA spinel, as shown in Fig. 8.⁵³⁾ Based on kinetic experiments, they concluded that reaction (10) was the preferential reaction for the formation of MA spinel in industrial production. Wang et al.⁵⁶⁾ also found that during LF refining, only Al₂O₃ inclusions were observed just after Al deoxidation. During a brief period of time, the number of Al₂O₃ inclusions decreased dramatically, while that of the MA spinel subsequently increased sharply. In their research, the increase in the mass ratio of increased MA spinel and reduced Al₂O₃ inclusions correlated with reaction (10). Therefore, they concluded that reaction (10) was the predominant mechanism for the formation of MA spinel. From the above analysis, it is concluded that MA spinel is mainly formed by reaction (10) by the transformation of Al₂O₃.

Fig. 8.

Change in MgO content of inclusions in steel with reaction time.⁵³⁾

2.5. Kinetics of Formation of MA Spinel

As mentioned above, the MA spinel was mainly formed by the transformation of Al₂O₃. For the formation of MA spinel in steel, two steps are considered. In the first step, Mg is supplied from MgO sources to the steel. In the second step, the dissolved Mg reacts with the existing Al₂O₃ inclusions and forms MA spinel.

Many studies on the kinetics of the Mg supply step have been conducted. Okuyama et al.⁴⁶⁾ investigated the variation in the Mg content in steel that was reacted with slag using an induction furnace. Based on a theoretical analysis of the experimental results, they concluded that the Mg transfer in the steel was the rate-controlling step for the supply of Mg, as shown in Fig. 9.⁴⁶⁾ Harada et al.⁴⁹⁾ studied the supply rate of Mg to steel from MgO refractory; this was accomplished by immersing and rotating a cylinder of sintered MgO in steel. They also concluded that the Mg transfer in the steel was the rate-controlling step for the Mg supply. Liu et al.^50,53) investigated the dissolution of Mg from MgO refractory, MgO–C refractory, and MgO-containing slag to Al-killed steel. Their experimental results correlated with a model calculation, which assumed that the Mg transfer in the steel was the rate-controlling step. From the works mentioned above, it is widely accepted that Mg transfer in steel acts as the rate-controlling step for the step of Mg supply regardless of the MgO sources.

Fig. 9.

Observed changes in Mg content in metal after slag/steel reaction in comparison with calculated ones.⁴⁶⁾

Both MgO-containing refining slag and MgO-based refractory can supply Mg to Al-killed steel, and many studies have been conducted to compare the differences in the supply rate of Mg from various MgO sources. Harada et al.⁵⁷⁾ found that the effect of the reaction between steel/slag on the Mg content in steel was larger than that between steel/refractory, and the contribution ratio of the steel/slag reaction was almost 99% in their model calculation, as shown in Fig. 10.⁵⁷⁾ Liu et al.⁵⁰⁾ compared the dissolution of Mg from MgO refractory, MgO–C refractory, and MgO-containing slag to Al-killed steel. In their calculations, the reaction between steel/refractory had a contribution ratio of about 50% for the Mg content in steel.

Fig. 10.

Effect of each source on supplement of [Mg] and formation of spinel.⁵⁷⁾

With regard to the second step for the formation of MA spinel, i.e., the Al₂O₃ transformation, few studies have been conducted, and all researchers assumed that the steel was in equilibrium with the inclusions. However, Liu et al.⁵⁰⁾ observed that the inclusion of MA spinel had cores with low MgO content and outer layers with high MgO content, as shown in Fig. 11.⁵³⁾ This result indicated that a MgO gradient existed inside the solid inclusions, which contradicted the assumption that the inclusions were in equilibrium with the steel.

Fig. 11.

Inclusion of (a) high MgO, (b) spinel, and (c) spinel surrounded by outer layer of high MgO content.⁵³⁾ (Online version in color.)

Galindo et al.⁵⁸⁾ developed a kinetic model for the transformation of Al₂O₃ into spinel. In their model, the mass transfer at the boundary layer of the steel and the cationic diffusion at the formed spinel layer were considered. The diffusion in the formed spinel layer was calculated using the Wagner-Schmalzried theory, and the transfer rate in the steel was estimated using experimental data for the measured reaction rate between the molten slag and steel. The authors concluded that the mass transfer in steel was the rate-controlling step for the transformation of Al₂O₃ inclusions. However, their conclusion has not been confirmed by experimental results.

Recently, Liu et al. investigated the transformation rate from Al₂O₃ to MgO·Al₂O₃ spinel.⁵⁹⁾ In this experiment, an Al₂O₃ rod was immersed into Mg-containing molten steel, and the rate of formation of the spinel layer on the surface of the rod was measured. The rate of the transformation from Al₂O₃ to MgO·Al₂O₃ spinel was rapid when there was sufficient Mg in the steel, as shown in Fig. 12.⁵⁹⁾ By evaluating the apparent activation energy, the rate-controlling step was determined to be the MgO diffusion in the formed MgO·Al₂O₃ spinel layer. They calculated the time required for the Al₂O₃ inclusions to completely transform to spinel. They used the diffusion coefficient of MgO in the spinel layer estimated by their result, and concluded that it took just 3 s for a typical Al₂O₃ inclusion (5 μm in radius, spherical shape) at 1873 K.

Fig. 12.

Changes in thickness of spinel layer with immersion time.⁵⁹⁾

2.6. Countermeasures to Control Formation of Spinel Inclusions

To suppress the formation of spinel inclusions, the slag composition should be strictly controlled, and suitable refractory material should be chosen.

In order to control the Mg supply from slag, one possible method is to decrease the MgO activity (a_MgO) in the slag by decreasing the MgO content (X_MgO) and the activity coefficient of MgO (γ_MgO).

One of the origins of MgO in slag is BOF slag, which is poured into the ladle during tapping. Another source is intentionally added flux to decrease the erosion of the MgO-based refractory. If the MgO content in slag is decreased, then the erosion of the refractory by the slag is enhanced. Therefore, it is important to balance the merits and demerits of this action.

When the ratio of CaO/SiO₂ or CaO/Al₂O₃ in the slag decreases, the a_MgO decreases by the decrease in γ_MgO, as shown in Fig. 13.⁴⁶⁾ Nishi et al.⁴⁵⁾ found that the MgO content of the observed MA spinel increased significantly with an increase in the CaO/SiO₂ ratio in the slag. In research conducted by Okuyama et al.,⁴⁶⁾ the MgO content in inclusions decreased when the CaO/SiO₂ or CaO/Al₂O₃ ratios of the refining slag were reduced.

Fig. 13.

Calculated relations between slag basicity, CaO/Al₂O₃, and activity of MgO in slag.⁴⁶⁾

Another method to suppress the formation of spinel inclusions is to suppress the reaction between MgO in slag and Al in steel by increasing the FeO or MnO content in the slag, which preferentially reacts with the Al in steel. The FeO or MnO in slag originates from the BOF slag. Several researchers^60,61,62,63) found that the MgO content in inclusions increased as the FeO content in the slag decreased. When the FeO + MnO content was higher than 2–3 mass%, the MgO content in the inclusions was almost lower than 5 mass%. However, when the FeO or MnO content in the top slag increased, the steel cleanliness decreased.⁶⁴⁾ Therefore, it is important to consider the optimum conditions of this action.

Refining slag supplies larger amount of Mg than MgO-based refractory because of its higher dissolution rate. However, MgO-based refractory is able to supply enough Mg to promote the formation of spinel inclusions during industrial production. In order to suppress the formation of MA spinel, it is better to use Al₂O₃-based refractory as the lining material of the ladle. However, the MgO-based refractory has the merit of a low price and better resistance for thermal shock and erosion. Therefore, it is not easy to change the refractory materials.

3. Formation of CaO–Al₂O₃ Type Inclusions

3.1. Composition Change from Al₂O₃ to CaO–Al₂O₃ System via MA

As described in the previous section, many researchers reported that inclusions transformed into CaO–Al₂O₃ systems via MA spinel from Al₂O₃ during secondary refining in Al-killed steel.^{15,16,17,18,19,20)} Todoroki et al.¹⁵⁾ studied the changes in composition of the inclusions in Al-killed stainless steel by a reaction with the top slag of a CaO–Al₂O₃–MgO–F system. They found that Al₂O₃ or MA spinel was formed just after the addition of Al. The amounts of MgO and CaO in the slag were then reduced, and the concentrations of Mg and Ca in the steel increased. CaO–Al₂O₃-type inclusions formed in the final stage of treatment, when the Al content was extremely high. A similar phenomenon in which an Al₂O₃ inclusion was changed into a CaO–Al₂O₃-type inclusion by a reaction between Al in the steel and CaO-containing slag was also observed.^{16,17,18,19,20)} Harada et al.⁵⁷⁾ conducted a kinetic analysis of the compositional changes in inclusions through a reaction with slag using a laboratory-scale furnace. They found that a CaO–Al₂O₃ inclusion containing MgO was observed during the experiment and concluded that the origin of the CaO–Al₂O₃-type inclusion was entrapped slag droplets in which unstable oxides (i.e., SiO₂, FeO, and MnO) were reduced from the presence of Al in the steel.

In the following text, the latest publications regarding CaO–Al₂O₃-type inclusions are summarized. First, thermodynamic data related to the formation of CaO–Al₂O₃-type inclusions are discussed, followed by the formation mechanism. Finally, countermeasures for the control of CaO–Al₂O₃-type inclusions are reviewed.

3.2. Thermodynamic Data for Formation of CaO–Al₂O₃-type Inclusion

Thermodynamic data for the formation of CaO–Al₂O₃-type inclusions consist of the following three reactions:   

$$\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)=2\left[\mathrm{A}\mathrm{l}\right]+3\left[\mathrm{O}\right]$$(1)

  

$$\mathrm{C}\mathrm{a}\mathrm{O}\left(\mathrm{s}\right)=\left[\mathrm{C}\mathrm{a}\right]+\left[\mathrm{O}\right]$$(11)

  

$$m\text{CaO}\left(\mathrm{s}\right)+n{\text{Al}}_2{\text{O}}_3\left(\mathrm{s}\right)=m\text{CaO}\cdot n{\text{Al}}_2{\text{O}}_3\left(\mathrm{s}\right)$$(12)

Various values of the equilibrium constant of reaction (11), published by different researchers and recommended by JSPS, are listed in Table 5.^{28,30,65,66,67,68,69,70,71,72,73)} By ref. [40], the calculated relationship between [%Ca] and [%O] using an interaction coefficient sometimes reveals an oval shape, as shown in Fig. 14. At a given content of [%Ca], two values of [%O] are calculated in equilibrium conditions.

Table 5. List of equilibrium constants for Ca–O reaction.

Years	Researchers	lgKCa at 1873 K	Determining method
1970	S. Kobayashi et al.[65]	−9.82	1823 K, MgO crucible with CaO lining, Ca–Ar gas bubble
1975	T. Ototani et al.[66]	−8.23	1873 K, CaO crucible, pure Ca metal
1980	S. Gustafsson et al.[67]	−5.8	CaO crucible
1986	M. Nadif et al.[32]	−6.05	CaO crucible
1988	Q. Han et al.[68]	−8.26	1873 K, CaO–CaS crucible, Ca vapor
1989	T. Wakasugi et al.[69]	−9.04	–
1994	S. Cho et al.[70]	−10.22	1873 K, Al2O3 and CaO crucible, pure Ca metal
1997	H. Itoh et al.[71]	−7.15	1873 K, 2023 K, CaO crucible, pure Ca metal
1999	K. Ogawa et al.[72]	−7.6	1873 K, CaO crucible
2011	I. Seki et al.[73]	−4.38	1873 K, CaO, Al2O3, CaO–ZrO2 crucible, CaO–Al2O3–ZrO2 slag, and Ca metal
1968	JSPS[28]	−9.08	1873 K

Fig. 14.

Calculated relationships between [Ca] and [O].⁴⁰⁾

The standard Gibbs energies for the formation of compounds in a CaO–Al₂O₃ system were measured by Galvanic cells using 4CaO·P₂O₅ solid electrolyte by Nagata et al.⁷⁴⁾ The results are as follows:   

$$\begin{array}{l}\text{CaO}\left(\text{s}\right)+6{\text{Al}}_2{\text{O}}_3\left(\text{s}\right)=\text{CaO}\cdot 6{\text{Al}}_2{\text{O}}_3\left(\text{s}\right)\\ \mathrm{\Delta }{G}_{\text{f}-\text{CA}6}^0=-16\mathrm{\hspace{0.17em} }380-37.58T\end{array}$$(13)

  

$$\begin{array}{l}\text{CaO}\left(\text{s}\right)+2{\text{Al}}_2{\text{O}}_3\left(\text{s}\right)=\text{CaO}\cdot 2{\text{Al}}_2{\text{O}}_3\left(\text{s}\right)\\ \mathrm{\Delta }{G}_{\text{f}-\text{CA}2}^0=-15\mathrm{\hspace{0.17em} }650-25.82T\end{array}$$(14)

  

$$\begin{array}{l}\text{CaO}\left(\text{s}\right)+{\text{Al}}_2{\text{O}}_3\left(\text{s}\right)=\text{CaO}\cdot {\text{Al}}_2{\text{O}}_3\left(\text{s}\right)\\ \mathrm{\Delta }{G}_{\text{f}-\text{CA}}^0=-17\mathrm{\hspace{0.17em} }910-17.38T\end{array}$$(15)

  

$$\begin{array}{l}3\text{CaO}\left(\text{s}\right)+{\text{Al}}_2{\text{O}}_3\left(\text{s}\right)=3\text{CaO}\cdot {\text{Al}}_2{\text{O}}_3\left(\text{s}\right)\\ \mathrm{\Delta }{G}_{\text{f}-\text{C}3\text{A}}^0=-11\mathrm{\hspace{0.17em} }790-28.27T\end{array}$$(16)

The activity of CaO and Al₂O₃ in a CaO–Al₂O₃–CaS system saturated by CaS was measured by Fujisawa et al.⁷⁵⁾ They studied the equilibrium between molten Fe–Al–S and CaO–Al₂O₃–CaS slags in a CaS crucible. By application of the Gibbs-Duhem equation, the activities of CaO and Al₂O₃ in slag were calculated, as shown in Fig. 15.

Fig. 15.

Activities of CaO and Al₂O₃ in CaO–Al₂O₃–CaS_sat. system at 1873 K.⁷⁵⁾

The formed CaO–Al₂O₃-type inclusions usually contain a significant percentage of MgO, which helps to decrease the melting point of the pure CaO–Al₂O₃ product. Ohta et al.⁷⁶⁾ investigated the activity in CaO–MgO–Al₂O₃ slags. The activity of CaO, MgO, and Al₂O₃ in a liquid region at 1823 K and 1873 K were determined.

The stability diagram of CaO–Al₂O₃ and Al₂O₃ for Fe-11mass%Cr steel was calculated using reaction (17) and is shown in Fig. 16. In this calculation, the equilibrium relation of reaction (1) measured by Itoh et al.²⁹⁾ was used.   

$$3\left[\text{Ca}\right]+{\left({\text{Al}}_2{\text{O}}_3\right)}_{\text{inclusion}}=3{\left(\text{CaO}\right)}_{\text{inclusion}}+2\left[\text{Al}\right]$$(17)

Fig. 16.

Stability phase diagram of Al₂O₃ and CaO–Al₂O₃ at 1873 K.

The equilibrium constant of reaction (17) was calculated by combining reactions (1) and (11). The activity of Al₂O₃ was unity, and that of CaO in CaO–Al₂O₃-type inclusions was assumed as 0.2. This value was obtained by Fig. 15 assuming that the composition of the inclusions was 40%CaO-60%Al₂O₃ without CaS.

The interaction coefficients needed for the calculation are listed in Table 6.^29,40,68) To control the formation of CaO–Al₂O₃-type inclusions, it is essential to control the steel composition into or out of the CaO–Al₂O₃-stable region. However, the calculated critical content of Ca for the formation of CaO–Al₂O₃-type inclusions varied greatly based on different values of the equilibrium constant of reaction (11).

Table 6. Interaction coefficients used to calculate Fig. 16.

$e_{\text{i}}^{\text{j}}$	j i	Ca	Al	O	Cr
Ca		−0.002[68]	−0.072	−1293	0
Al		−0.047	0.0429	−1.98	0.0122
O		−515	−1.17	−0.174	−0.052
${\gamma }_{\text{Al}}^{\text{O}}={39.82}^{\left[29\right]}\text{;}\mathrm{\hspace{0.17em} } {\gamma }_{\text{Al}}^{\text{Al},\text{O}}=-{0.02836}^{\left[29\right]};$ ${\gamma }_{\text{Ca}}^{\text{O}}=2\mathrm{\hspace{0.17em} }240;\mathrm{\hspace{0.17em} } {\gamma }_{\text{Ca}}^{\text{Ca},\text{O}}=1\mathrm{\hspace{0.17em} }801$ ${\gamma }_{\text{O}}^{\text{Ca}}=357;\mathrm{\hspace{0.17em} } {\gamma }_{\text{O}}^{\text{Ca},\text{O}}=1\mathrm{\hspace{0.17em} } 788;\mathrm{\hspace{0.17em} } {\gamma }_{\text{O}}^{\text{Cr}}=-0.000576;$

In addition, in considering the transfer of MA to CaO–Al₂O₃, the equilibrium relation of reaction (18) is important. The stability diagram of MA and CaO–Al₂O₃ for Fe-11mass%Cr steel was calculated and is shown in Fig. 17.⁷⁷⁾ The interaction coefficients listed in Tables 3 and 6 were used for calculation, and in addition, $e_{Ca}^{Mg}=0$ and $e_{Mg}^{Ca}=0$ are also used.⁷⁷⁾   

$${\left(\text{MgO}\right)}_{\text{inclusion}}+\left[\text{Ca}\right]=\left[\text{Mg}\right]+{\left(\text{CaO}\right)}_{\text{inclusion}}$$(18)

Fig. 17.

Stability diagram of MgO–CaO system at 1873 K (a_MgO = 1, a_CaO = 0.2). (Online version in color.)

In this calculation, the activity of MgO was unity and the activity of Al₂O₃ is 0.034.⁴⁴⁾ In addition, the activity of CaO was assumed to be 0.2 when considering a CaO–Al₂O₃ binary system with a 0.55 molar fraction of CaO.⁷⁵⁾ As a significant variation of the equilibrium constant of reactions (2) and (11), the calculated lines are widely different depending on the combination. It is difficult to evaluate the accuracy of the equilibrium relation, although the experimental results were mostly located under the stable conditions of CaO–Al₂O₃. The result of the calculation when the activity of MgO decreased to 0.1 and thus that of Al₂O₃ increased to 0.43⁴⁴⁾ is shown in Fig. 18. Comparing to Fig. 17, all the calculated lines move to upper side where Ca content is high. In this case, even though the concentrations of Ca and Mg is in the CaO–Al₂O₃ stable region in Fig. 17, CaO–Al₂O₃ becomes no longer stable when the activity of MgO is decreased. This indicates that for the transformation from a spinel to CaO–Al₂O₃, the MgO concentration of the spinel has to increase to saturation. This calculation agrees well with the result of the inclusion of the CaO–Al₂O₃ system observed after the formation of the MgO-saturated spinel and MgO-rich oxide.

Fig. 18.

Stability diagram of MgO–CaO system at 1873 K (a_MgO = 0.1, a_CaO = 0.2). (Online version in color.)

3.3. Formation Mechanism of CaO–Al₂O₃-type Inclusion

As mentioned previously, a CaO–Al₂O₃-type inclusion was formed spontaneously during secondary refining without the intentional addition of Ca. Many possible formation mechanisms have been proposed by researchers. The main formation mechanisms are summarized as Ca in added alloy, slag entrapment, and inclusion transformation.

Mizuno et al.⁷⁸⁾ studied the effects of Al and Ca in ferrosilicon alloys on the inclusion composition in stainless steel. They found that the impurity of Ca in the added alloy promoted the formation of CaO–Al₂O₃-type inclusions in steel. Compared with the addition of pure Ca, the addition of Ca as the impurity of the alloying material has the advantage of avoiding the formation of the local enrichment of Ca. Therefore, the formation of CaO and CaS inclusions can be suppressed. However, the content of Ca is not specified as the composition of the alloy; therefore, it is difficult to add Ca from the impurity of the alloy in industrial production.

Many researchers considered that entrapped slag droplets are the main source of CaO–Al₂O₃-type inclusions in steel during secondary refining.^{57,79,80,81,82)} The secondary refining slag is generally composed of CaO–Al₂O₃–MgO–SiO₂. At the slag-steel interface, some slag droplets are separated from the slag bulk by stirring and are entrapped in the steel. The entrapped slag droplets in the steel further react with the Al in the steel to form CaO–Al₂O₃-type inclusions. Harada et al.⁵⁷⁾ conducted a kinetic analysis of compositional changes in inclusions by the reaction with slag using a laboratory-scale furnace under Ar bubbling conditions. They found that after the reaction of slag and steel, Al₂O₃ inclusions, observed just after the addition of Al, changed to MA spinel. On the other hand, CaO–Al₂O₃ inclusions containing MgO were observed during the entire period of the experiment. Their composition agreed well with the calculated composition of the inclusions originating from slag. They concluded that the origin of the CaO–Al₂O₃-type inclusions was not Al₂O₃ but entrapped slag droplets in which unstable oxides (i.e., SiO₂, FeO, and MnO) were reduced by the Al in steel.

In addition, many experiments using slag tracers have been conducted to confirm this phenomenon.^{27,79,80,81,82,83,84)} A slag tracer should meet the requirements that the cation of the tracer does not dissolve in Fe at steelmaking temperatures, and stable in the slag phase without large changes in the physical properties of the slag. BaO and SrO are typically used as slag tracers. Ohta et al.⁷⁹⁾ studied the source of CaO-containing inclusions during the ladle refining of ultraclean bearing steel by using a slag tracer (SrO). The slag tracer was observed in the detected inclusions. In addition, they found that the tracer content in the inclusions was lower than that in the slag, which indicated the coalescence of the entrapped top slag and the existing Al₂O₃ inclusions, as shown in Fig. 19. Kawakami et al.⁸²⁾ studied the generation mechanism of nonmetallic inclusions in high-cleanliness steel by using BaO as a tracer. In their experiment, the slag tracer was observed in nearly all inclusions. They considered that the dissolved Ca and Ba, which were supplied by the reduction of CaO and BaO in the slag with the Al in steel, reacted with the existing inclusions and formed CaO–Al₂O₃-type inclusions.

Fig. 19.

Relationship between CaO and SrO content in inclusions after ladle furnace refining.⁷⁹⁾

From the above analysis, although the slag tracer was detected in the inclusions, an explanation for the origin of the CaO-containing inclusions is difficult. The employed slag tracers (BaO and SrO) could also be reduced and supply Ba or Sr into the steel. In addition, the information about the solubility of Ba and Sr in steel at steelmaking temperatures is limited. Therefore, experiments with a slag tracer are not a suitable technique to study the origin of CaO-containing inclusions.

The mechanism of the inclusion transformation was proposed by many researchers as the main formation mechanism for CaO–Al₂O₃-type inclusions.^{15,16,17,18,19,20,85)} By the reaction of slag and steel, Ca is reduced from the slag by reaction (19). The dissolved Ca in the steel further reacts with the existing Al₂O₃ or spinel inclusions to form CaO–Al₂O₃-type inclusions by reaction (17).   

$$2\left[\text{Al}\right]+3{\left(\text{CaO}\right)}_{\text{slag}}=3\left[\text{Ca}\right]+{\left({\text{Al}}_2{\text{O}}_3\right)}_{\text{slag}}$$(19)

If the system of slag/steel/inclusion reaches equilibrium, then the activity of CaO in the slag and that in the inclusions should be equal. This indicates that the CaO-containing slag has the potential to increase the CaO content in the existing inclusions. Todoroki¹⁵⁾ studied the formation mechanism of inclusions in Al-killed stainless steel that reacted with the top slag of a CaO–Al₂O₃–MgO–F system. They found that alumina or spinel inclusions were formed immediately after the addition of Al. MgO and CaO in the slag phase were reduced by Al to generate soluble Mg and Ca in the steel. CaO–Al₂O₃-type inclusions, which had the most stable oxide according to their thermodynamic calculations, formed at the last stage of the experiment, when the Ca content increased by at least 1 mass ppm. The same phenomenon, where the Al₂O₃ inclusions in steel were modified by CaO-containing slag, was confirmed by other researchers for various Al-killed steels.^{16,17,18,19,20)} Yoshioka et al.¹⁹⁾ found that inclusions changed from the primary deoxidation product of Al₂O₃ to MA spinel and CaO–Al₂O₃-type inclusions in the Al-killed steel with a 0.08 mass% content of Al during LF refining. However, in another study⁸⁴⁾ where the Al content was 1 mass%, the observed inclusions in steel were composed only of Al₂O₃ and spinel. CaO–Al₂O₃-type inclusions were not observed during the LF refining.

Recently, Mg and Ca dissolution behavior from MgO and CaO-saturated CaO–MgO–Al₂O₃ slag and its effect on the composition changes of inclusions during the refining process were studied by Liu et al.⁷⁷⁾ They showed that both MgO and CaO in slag were reduced by the Al in steel, and Mg and Ca were dissolved into the steel. When the Al concentration in the steel was 0.25%, the concentration of dissolved Mg was 30 ppm or higher, whereas that of Ca was only 0.3 ppm. The initial Al₂O₃ inclusions transformed into MA spinel and finally changed into MgO inclusions. However, CaO–Al₂O₃-type inclusions were not observed, as shown in Fig. 20. When the Al concentration was 0.75%, the concentration of dissolved Ca increased to 0.9 ppm, and CaO–Al₂O₃-type inclusions with MgO were observed via the route of Al₂O₃, MA, and MgO.

Fig. 20.

Changes in inclusion composition for steel containing 0.25% of Al.⁷⁷⁾

This result indicated that CaO–Al₂O₃-type inclusions never formed by the reaction with steel and slag for the normal low-Al-containing steel. Kumar et al.⁸⁶⁾ also concluded by the equilibrium and kinetic discussion that the Ca transfer from slag is generally small and only affects inclusions in a significant manner for highly deoxidized steels, long ladle contact times, and low total oxygen concentrations.

3.4 Countermeasures for Control of CaO–Al₂O₃-type Inclusions

With the development of steelmaking technology, the requirement for inclusion control has become increasingly strict, and CaO–Al₂O₃-type inclusions need to be removed to further increase the cleanliness of the steel product. According to the published literature, research on countermeasures for the control of CaO–Al₂O₃-type inclusions is limited. Ohta et al.⁷⁹⁾ investigated the behavior of CaO-containing inclusions during the ladle refining of ultraclean bearing steel. In their study, they found that the main origin of the CaO in the inclusions was slag that was entrapped during gas stirring. The quality of the ultrafine bearing steel was improved by applying countermeasures to suppress the entrapment of slag. Watanabe et al.⁸⁷⁾ published recent developments to improve the cleanliness of bearing steel. In the secondary steelmaking process, the flow rate of stirring gas in a ladle furnace was optimized to minimize oxide inclusions originating from slag. In the continuous casting process, nozzle sand is discharged to the outside of the tundish to prevent contamination. In their study, using the combination of techniques introduced above, the fatigue life of bearing steel was extended to 3.9 times the original value. The basic idea for controlling CaO–Al₂O₃-type inclusion formation is to control the Ca source for its formation. The effects of other sources such as Ca in alloy as an impurity⁸⁶⁾ are also important.

4. Conclusions

Many researchers have reported on inclusions transformed in the routine of Al₂O₃ → MA spinel → CaO–Al₂O₃ during secondary refining in Al-killed steel. The deoxidizer of Al is intentionally added to the steel, and Al₂O₃ inclusions are formed as a deoxidation product. However, MA spinel and CaO–Al₂O₃-type inclusions have been observed even without the intentional addition of Mg and Ca. Therefore, it is necessary to clarify the source of Mg and Ca for the formation of MA spinel and CaO–Al₂O₃-type inclusions in order to control the compositions of the inclusions. With regard to this phenomenon, a good review was published by Park and Todoroki on 2010.⁶⁾ Several studies have since been conducted. This paper summarized the research activities on the composition changes of inclusions during secondary refining from the viewpoints of thermodynamics and kinetics.

The composition changes of inclusions from Al₂O₃ to MA spinel have been observed by many researchers, and the supplement of Mg was achieved by the reaction between MgO in slag and Al in steel. In addition, the supplement of Mg from the MgO-based refractory was investigated. The equilibrium constant of the deoxidation reaction by Mg was widely different by the researchers. Therefore, the critical contents of Mg and Al from which Al₂O₃ changes to MA and MA changes to MgO were difficult determine. The rate-controlling step to change from Al₂O₃ to MA was a diffusion in the formed MA layer on Al₂O₃, but the time to change the Al₂O₃ inclusion to MA was considered to be very short. The dissolution rate of Mg from slag was controlled by the mass transfer in slag and metal, and an analysis using a coupled reaction model was conducted.

On the other hand, the composition change of MA to CaO–Al₂O₃-type inclusions was indefinite. Even though many researchers observed the composition change from MA to a CaO–Al₂O₃ system, the dissolved Ca content was very low as the Al content in normal grades of steel is less than 0.1%. The equilibrium constant of the deoxidation reaction by Ca was quite widely different by the researchers. Therefore, the stable oxide was impossible to determine by the equilibrium relation of Ca, Mg, and Al with oxygen.

Acknowledgment

We would like to thank Prof. Bart Blanpain for giving one of the authors (SK) an opportunity to stay at KU Leuven in writing this paper.

References

• 1)   Y.  Murakami: Tetsu-to-Hagané, 75 (1989), 1267 (in Japanese).
• 2)   Y.  Murakami,  T.  Toriyama,  Y.  Koyasu and  S.  Nishida: Tetsu-to-Hagané, 79 (1993), 678 (in Japanese).
• 3)   J.  Takamura and  S.  Mizoguchi: Proc. 6th Int. Iron and Steel Congr., ISIJ, Tokyo, (1990), 2331.
• 4)   T.  Sawai,  M.  Wakoh,  Y.  Ueshima and  S.  Mizoguchi: ISIJ Int., 32 (1992), 169.
• 5)   S.  Inada and  H.  Todoroki: Textbook of the 182nd and 183rd Nishiyama Memorial Seminar, ISIJ, Tokyo, (2004), 227 (in Japanese).
• 6)   J. H.  Park and  H.  Todoroki: ISIJ Int., 50 (2010), 1333.
• 7)  Slag Atlas, 2nd ed., ed. by Verein Deutscher Eisenhüttenleute, Verlag Stahleisen, Düsseldorf, (1995), 522.
• 8)   P.  Auerkari: VTT Manufacturing Technology, VTT Tiedotteita – Meddelanden – Research Notes 1792, Espoo, (1996), 14, https://www.vtt.fi/inf/pdf/tiedotteet/1996/T1792.pdf, (accessed 2019-10-10).
• 9)   H.  Yasunaka,  R.  Yamanaka,  T.  Inoue and  T.  Saito: Tetsu-to-Hagané, 81 (1995), 529 (in Japanese).
• 10)   I.  Ganesh: Int. Mater. Rev., 58 (2013), 63.
• 11)   N.  Shinozaki,  N.  Echida,  K.  Mukai,  Y.  Takahashi and  Y.  Tanaka: Tetsu-to-Hagané, 80 (1994), 748 (in Japanese).
• 12)   T.  Hori,  M.  Hishiki,  H.  Kawabe and  M.  Uemura: J. Jpn. Soc. Precis. Eng., 49 (1983), 1345.
• 13)   S.  Yano,  K.  Nakanishi,  H.  Ooi,  K.  Fujimoto,  Y.  Mihara and  S.  Iwaoka: Trans. Iron Steel Inst. Jpn., 16 (1976), 258.
• 14)   S.  Kitamura: Proc. 7th Int. Symp. on Molten Slags, Fluxes and Salts, The South African Institute of Mining and Metallurgy, Johannesburg, (2004), 769.
• 15)   H.  Todoroki and  K.  Mizuno: Iron Steelmaker, 30 (2003), 60.
• 16)   M.  Jiang,  X.  Wang,  B.  Chen and  W.  Wang: ISIJ Int., 50 (2010), 95.
• 17)   Z.  Deng and  M.  Zhu: ISIJ Int., 53 (2013), 450.
• 18)   J.  Guo,  S.  Cheng and  Z.  Cheng: ISIJ Int., 53 (2013), 2142.
• 19)   T.  Yoshioka,  K.  Nakahata,  T.  Kawamura and  Y.  Ohba: ISIJ Int., 56 (2016), 1973.
• 20)   J. H.  Shin and  J. H.  Park: Metall. Mater. Trans. B, 48 (2017), 2820.
• 21)   E.  Sunami,  S.  Nozaki,  Y.  Tamaou and  T.  Miura: Tetsu-to-Hagané, 68 (1982), S248 (in Japanese).
• 22)   J.-H.  Park: Metall. Mater. Trans. B, 38 (2007), 657.
• 23)   Y.  Ehara,  S.  Yokoyama and  M.  Kawakami: Tetsu-to-Hagané, 93 (2007), 208 (in Japanese).
• 24)   Y.  Ehara,  S.  Yokoyama and  M.  Kawakami: Tetsu-to-Hagané, 93 (2007), 475 (in Japanese).
• 25)   J.-H.  Son,  I. H.  Jung,  S.-M.  Jung,  H.  Gaye and  H.-G.  Lee: ISIJ Int., 48 (2008), 1542.
• 26)   K.  Beskow and  D.  Sichen: Ironmaking Steelmaking, 31 (2004), 393.
• 27)   M.-H.  Song,  M.  Nzotta and  D.  Sichen: Steel Res. Int., 80 (2009), 753.
• 28)  Japan Society for the Promotion of Science and The 19th Committee on Steelmaking: Recommended Equilibrium Values for Steelmaking Reactions, Nikkan Kogyo Shimbun, Tokyo, (1968), 55 (in Japanese).
• 29)   H.  Itoh,  M.  Hino and  S.  Ban-ya: Tetsu-to-Hagané, 83 (1997), 773 (in Japanese).
• 30)   A. P.  Gorobetz: Metall. Koksokhim., 69 (1980), 34.
• 31)   I. S.  Kulikov: Izv. Akad. Nauk SSSR Met., 6 (1985), 9.
• 32)   M.  Nadif and  C.  Gatellier: Rev. Metall., 83 (1986), 377.
• 33)   R.  Inoue and  H.  Suito: Metall. Mater. Trans. B, 25 (1994), 235.
• 34)   H.  Ohta and  H.  Suito: Metall. Mater. Trans. B, 28 (1997), 1131.
• 35)   H.  Itoh,  M.  Hino and  S.  Ban-ya: Tetsu-to-Hagané, 83 (1997), 623 (in Japanese).
• 36)   Q.  Han,  D.  Zhou and  C.  Xiang: Steel Res., 68 (1997), 9.
• 37)   J. D.  Seo and  S. H.  Kim: Steel Res., 71 (2000), 101.
• 38)   W. G.  Seo,  W. H.  Han,  J. S.  Kim and  J. J.  Pak: ISIJ Int., 43 (2003), 201.
• 39)   J.  Gran and  S.  Du: Metall. Mater. Trans. B, 42 (2011), 921.
• 40)  Japan Society for the Promotion of Science and The 19th Committee on Steelmaking: Thermodynamic Data for Steelmaking, ed. by M. Hino and K. Ito, Tohoku University Press, Sendai, (2010), 45.
• 41)   H.  Itoh,  M.  Hino and  S.  Ban-ya: Tetsu-to-Hagané, 84 (1998), 85.
• 42)   S. K.  Jo,  B.  Song and  S. H.  Kim: Metall. Mater. Trans. B, 33 (2002), 703.
• 43)   M.  Kishi,  R.  Inoue and  H.  Suito: ISIJ Int., 34 (1994), 859.
• 44)   K.  Fujii,  T.  Nagasaka and  M.  Hino: ISIJ Int., 40 (2000), 1059.
• 45)   T.  Nishi and  K.  Shinme: Tetsu-to-Hagané, 84 (1998), 837 (in Japanese).
• 46)   G.  Okuyama,  K.  Yamaguchi,  S.  Takeuchi and  K. I.  Sorimachi: ISIJ Int., 40 (2000), 121.
• 47)   H.  Todoroki and  K.  Mizuno: ISIJ Int., 44 (2004), 1350.
• 48)   M.  Jiang,  X.  Wang,  B.  Chen and  W.  Wang: ISIJ Int., 48 (2008), 885.
• 49)   A.  Harada,  G.  Miyano,  N.  Maruoka,  H.  Shibata and  S.  Kitamura: ISIJ Int., 54 (2014), 2230.
• 50)   C.  Liu,  F.  Huang and  X.  Wang: Metall. Mater. Trans. B, 47 (2016), 999.
• 51)   V.  Brabie: ISIJ Int., 36 (1996), Suppl., S109.
• 52)   S.  Jansson,  V.  Brabie and  P.  Jönsson: Ironmaking Steelmaking, 33 (2006), 389.
• 53)   C.  Liu,  F.  Huang,  J.  Suo and  X.  Wang: Metall. Mater. Trans. B, 47 (2016), 989.
• 54)   C.  Liu,  X.  Gao,  S.-j.  Kim,  S.  Ueda and  S.  Kitamura: ISIJ Int., 58 (2018), 488.
• 55)   C.  Liu,  M.  Yagi,  X.  Gao,  S.-j.  Kim,  F.  Huang,  S.  Ueda and  S.  Kitamura: Metall. Mater. Trans. B, 49 (2018), 2298.
• 56)   X.  Wang,  X.  Li,  Q.  Li,  F.  Huang,  H.  Li and  J.  Yang: Steel Res. Int., 85 (2014), 155.
• 57)   A.  Harada,  N.  Maruoka,  H.  Shibata,  M.  Zeze,  N.  Asahara,  F.  Huang and  S.  Kitamura: ISIJ Int., 54 (2014), 2569.
• 58)   A.  Galindo,  G. A.  Irons,  S.  Sun and  K.  Coley: Proc. 6th Int. Congr., on the Science and Technology of Steelmaking (ICS2015), CSM, Beijing, (2015), 719.
• 59)   C.  Liu,  M.  Yagi,  X.  Gao,  S.-j.  Kim,  F.  Huang,  S.  Ueda and  S.  Kitamura: Metall. Mater. Trans. B, 49 (2018), 113.
• 60)   K. J.  Graham and  G. A.  Irons: Iron Steel Technol., 6 (2009), 164.
• 61)   J.  Mendez,  A.  Gomez,  C.  Capurro,  R.  Donayo and  C.  Cicutti: Proc. 8th Int. Conf. on Clean Steel, OMBKE, Budapest, Hungary, (2012), 1, CD-ROM.
• 62)   C.  Cicutti,  C.  Capurro and  G.  Cerrutti: Proc. 9th Int. Conf. on Clean Steel, OMBKE, Budapest, Hungary, (2015), 1, CD-ROM.
• 63)   G.  Chen,  Y.  Guo and  S.  He: Metall. Res. Technol., 113 (2016), 204.
• 64)   L.  Zhang and  B.  Thomas: ISIJ Int., 43 (2003), 271.
• 65)   S.  Kobayashi,  Y.  Omori and  K.  Sanbongi: Tetsu-to-Hagané, 56 (1970), 998 (in Japanese).
• 66)   T.  Ototani,  Y.  Kataura and  T.  Degawa: Tetsu-to-Hagané, 61 (1975), 2167 (in Japanese).
• 67)   S.  Gustafsson and  P.  Mellberg: Scand. J. Metall., 9 (1980), 111.
• 68)   Q.  Han,  X.  Zhang,  D.  Chen and  P.  Wang: Metall. Trans. B, 19 (1988), 617.
• 69)   T.  Wakasugi,  F.  Tsukihashi and  N.  Sano: Tetsu-to-Hagané, 75 (1989), 2018 (in Japanese).
• 70)   S.  Cho and  H.  Suito: ISIJ Int., 34 (1994), 265 (in Japanese).
• 71)   H.  Itoh,  M.  Hino and  S.  Ban-ya: Tetsu-to-Hagané, 83 (1997), 695.
• 72)   K.  Ogawa,  K.  Tomioka and  H.  Yamana: Progress in Research on Ultra Clean Steel: Fundamental and Applied Aspects, The Final Report of Ultra Clean Steel Research Group, ISIJ, Tokyo, (1999), 90 (in Japanese).
• 73)   I.  Seki,  K.  Nagata,  T.  Ashino and  J.  Tanabe: ISIJ Int., 51 (2011), 1369.
• 74)   K.  Nagata,  J.  Tanabe and  K. S.  Goto: Tetsu-to-Hagané, 75 (1989), 2023 (in Japanese).
• 75)   T.  Fujisawa,  C.  Yamauchi and  H.  Sakao: Tetsu-to-Hagané, 76 (1990), 368 (in Japanese).
• 76)   H.  Ohta and  H.  Suito: ISIJ Int., 36 (1996), 983.
• 77)   C.  Liu,  X.  Gao,  S.  Ueda and  S.  Kitamura: ISIJ Int., 59 (2019), 268.
• 78)   K.  Mizuno,  H.  Todoroki,  M.  Noda and  T.  Tohge: Iron Steelmaker, 28 (2001), 93.
• 79)   H.  Ohta,  S.  Kimura,  T.  Mimura and  H.  Matsumoto: Kobe Steel Eng. Rep., 61 (2011), No. 1, 98 (in Japanese).
• 80)   T.  Komai,  K.  Miyamura,  M.  Oosaki,  A.  Kusano,  S.  Yamada,  I.  Shimazu,  N.  Sato and  T.  Yamamoto: Tetsu-to-Hagané, 67 (1981), 1152 (in Japanese).
• 81)   H.  Nagayama: Tetsu-to-Hagané, 56 (1970), 1716 (in Japanese).
• 82)   K.  Kawakami,  T.  Taniguchi and  K.  Nakashima: Tetsu-to-Hagané, 93 (2007), 743 (in Japanese).
• 83)   F.  Fuhr,  G.  Torga,  F.  Medina and  C.  Cicutti: Ironmaking Steelmaking, 34 (2007), 463.
• 84)   M.  Hojo,  R.  Nakao,  T.  Umezaki,  H.  Kawai,  S.  Tanaka and  S.  Fukumoto: ISIJ Int., 36 (1996), Suppl., S128.
• 85)   T.  Yoshioka,  T.  Ideguchi,  A.  Karasev,  Y.  Ohba and  P.  Jönsson: Steel Res. Int., 89 (2018), 1700287. https://doi.org/10.1002/srin.201700287
• 86)   D.  Kumar,  S. P. T.  Piva and  P. C.  Pistorius: Proc. 10th Int. Conf. on Clean Steel, OMBKE, Budapest, Hungary, (2018), 1, USB.
• 87)   Y.  Watanabe,  S.  Nabeshima and  T.  Terabatake: JFE Giho, 38 (2016), 87 (in Japanese).