Factors to Determine Inclusion Compositions in Molten Steel during the Secondary Refining Process of Case-Hardening Steel

Abstract

A study was carried out to observe inclusions during the secondary refining process of case-hardening steel to understand the factors to determine inclusion compositions. During the LF refining, inclusions changed from the primary deoxidation product of Al₂O₃ to MgO·Al₂O₃ and the CaO–Al₂O₃ system. At the end of LF refining, the compositions were placed on the tie-line connecting the areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq in the CaO–Al₂O₃–MgO diagram. This change took place by the composition evolution targeting the thermodynamic equilibrium states. After the RH treatment, the inclusion compositions changed mainly to Al₂O₃ and the CaO–Al₂O₃ system because MgO·Al₂O₃ inclusions were removed, while the CaO–Al₂O₃ system inclusions, the most stable oxide, were remained. This behavior could be described by the interfacial properties between the oxides and the molten steel. Al₂O₃ inclusions were considered to be newly generated during the RH treatment. It was confirmed that three factors of (1) equilibrium state (2) removal and (3) generation of inclusions dominated to determine the inclusion compositions.

1. Introduction

Non-metallic inclusions affect not only product quality but also manufacturing process of steels. This is why demands of steels with much higher cleanliness are very high. It is generally known that inclusions deteriorate mechanical properties of the steels. For example, inclusions significantly deteriorate fatigue properties of the bearing steels and the case-hardening steels. Therefore, for commercial mass production of steels containing high contents of C and Cr, such as bearing steel and case-hardening steel, it is of primary importance to decrease oxygen content to minimize the inclusion amount and size.^1,2) Today, a novel technology to control oxygen content less than 5 ppm has been developed to improve steel cleanliness.^2,4,5)

In addition to the pursuit of high steel cleanliness, technologies to control inclusion composition have been greatly progressed to make inclusions harmless.³⁾ It has been clarified that the effects of inclusion species on product quality or manufacturing process strongly depend on the inclusion compositions. Al₂O₃ and MgO·Al₂O₃ inclusions tend to get agglomerated during the steelmaking process to finally make clusters in the slabs. For this view, these inclusions are especially detrimental to fatigue properties.^2,4,5)

So far, a number of investigations have been carried out to clarify the inclusion formation of Al-killed steels. It was reported that both MgO·Al₂O₃ and the CaO–Al₂O₃ system inclusions were detected in the refining stages of Al-killed steels.^{3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19)} According to the report, MgO·Al₂O₃ inclusions have the potential to cause nozzle clogging due to its high sinterability because they are stable with a solid state at the steelmaking temperatures.³⁾ Further, detachment of the sintered lump occasionally leads to surface defects of the final products.^3,9)

In addition to the above studies, a number of researchers have conducted work on the inclusion evolutions in detail including the variations from Al₂O₃ to the CaO–Al₂O₃ system,⁶⁾ from Al₂O₃ to MgO·Al₂O₃ or partially to MgO,^7,8,9,10) and from MgO·Al₂O₃ to the CaO–Al₂O₃ system or to the CaO–MgO–Al₂O₃ system.^{11,12,13,14,15,16,17,18)} It should be pointed out that most of researches stated above were done in laboratory scales. Thus, information about inclusion formation in practice is still lacking to understand inclusion evolution occurring in the practical refining stages. In fact, one has to consider various factors such as additives, gas blow, agitation, reoxidation and slag entrainment in practice. This is why it is still strongly required to collect the plant data focusing on actual steelmaking processes.

In the actual steelmaking, especially in the secondary refining process, steels are deoxidized and desulphurized to ensure steel quality as high as possible. It is not too much to say that this process determines the quality of the final products. Deoxidizers such as aluminum have to be appropriately added in combination with the fluxes such as CaO to create slag. In fact, at this stage, attention should be paid to the evolution of the inclusion composition from the primary deoxidation products to adequately control inclusion amount and composition. It has been pointed out that it is not enough only to consider the evolution mentioned above, but both generation of newly formed inclusions^2,19) and removal from the molten steel^{20,21,22,23,24)} have to be also taken into account. However, few studies have been so far reported with considering all of the factors documented above.

From this viewpoint, a study was carried out aiming at clarifying the evolution mechanism of inclusions contained in case-hardening steel focusing on formation, variation and removal as well as generation on the way of the process. In practice, steel samples were taken from the ladle during the LF-RH process. As the inclusions consisted of CaO, MgO and Al₂O₃ oxides described later in detail, the inclusion compositions were analyzed by projecting on the CaO–MgO–Al₂O₃ diagram in order to understand the formations of various inclusions with multiple phases. Finally, the behavior of the inclusion evolution was elucidated with considering these factors.

2. Experimental

2.1. Practice

The experiment was carried out in practice of the case-hardening steel standardized as JIS SCM420. The procedure of our melt shop was as follows; the process began with melting at a 150 t EAF followed by the secondary refining process in a LF and a RH and finally the molten steel was cast by means of a CC. The ladle was lined by MgO–C bricks. The refining conditions are shown in Table 1.¹⁹⁾ Table 2 shows the chemical composition of the present steel featuring 0.2%C, 1% Cr and 0.15% Mo. The LF practice mainly aims at deoxidation and precise control of composition and temperature. Aluminum flake was used as a deoxidizer, which was added to the molten steel. To promote deoxidation and desulphurization, the slag with relatively high basicity was applied for refining, which had the composition doubly saturated with both CaO and MgO. In the RH practice, degassing was the main purpose to decrease N and H contents along with further deoxidation and separation of inclusions.

Table 1. Process overview.

Process	Time	Function	Remarks	Stirring gas
EAF	4.2×103 seconds/heat	1) Scrap melting	1) Oxidation Refining	N2
		2) Steelmaking	2) Only carbon deoxidation
LF	3.0×103 seconds/heat	1) Complete slag-off	1) Al deoxidation	N2 and/or Ar
		2) Reduction refining	2) Slag-making CaO,MgO sat. CaO–MgO–Al2O3–SiO2–F system slag
		3) Raising temperature	3) Composition adjustment
			4) Temperature adjustment
RH	1.2×103 seconds/heat	Degassing	1) Degassing (H,N)	Ar
			2) Deoxidation
			3) Flotation separation of inclusions
CC	4.2×103 seconds/heat	Shrouded atmosphere casting	Floatation separation	–

Table 2. Chemical composition of SCM420 (mass%).

C	Si	Mn	Cr	Mo
0.20	0.26	0.83	1.02	0.15

2.2. Samples and Evaluation Method

Steel samples were taken during the LF refining and RH treatment processes. Liquid steel samples were taken at 900, 1800 and 2700 seconds during the LF practice, which were indicated as “LF initial”, “LF middle” and “LF end”, respectively. In the RH process, a sample was taken at the end of RH indicated as “RH end”. The composition of inclusions on the cross section of each steel sample were analyzed by using automated SEM/EDS inclusion analyzer,²⁵⁾ which can automatically detect inclusions by searching for the particles by identifying the difference of the BSED signal intensity. The scanning area was determined as 100 mm². All the inclusions larger than 1 μm in diameter were analyzed. The number of inclusions analyzed depended on the sample; approximately 1500 data were taken in the three samples of the LF and approximately 800 data were taken in the RH end. Smaller amount of inclusions in the RH end was owing to the decrease in total O content.

As the inclusions consisted of CaO, MgO and Al₂O₃ oxides described later, their compositions were projected on the CaO–MgO–Al₂O₃ diagram.²⁶⁾ Composition was compensated when inclusion contained S; the corresponding content was allotted firstly to Mn and successively to Ca accounting for the formation of MnS and CaS. After this procedure, the rest of Ca was allotted to CaO.¹⁹⁾ SEM equipped with EDS was used to observe the morphology and the element distribution of inclusions. Spark discharge atomic emission spectrometric analysis was applied for the measurement of Al content. The concentrations of Ca and Mg were measured by ICP mass spectrometry method. Total oxygen content was measured by the inert gas fusion-infrared absorptiometry.

3. Results

3.1. Variation of Minor Elements in the Steel

It is generally known that the behavior of the deoxidizing elements such as Al, Mg and Ca affects the inclusion compositions, morphology and size. Figure 1 shows the variations of Al, Mg and Ca contents during the secondary refining. The increase of Al content from LF end to RH end was due to the addition of Al because Al was consumed during the RH process. This consumption might be attributed to the reaction between metal and slag, oxygen supply from the atmosphere and desulfurization.^2,19,27)

Fig. 1.

Variation of Al, Ca and Mg content during secondary refining process.

According to the previous studies,^7,9,19) Ca and Mg were supplied through reduction of CaO and MgO contained in the slag by the Al addition. The concentrations of Ca and Mg were 5.9 and 2.5 ppm, respectively, at the LF initial. After the middle of LF, Ca increased to 6.5–7.9 ppm while Mg slightly decreased to 0.9–1.6 ppm. Finally, Ca and Mg contents decreased to less than 2 ppm, and less than 1 ppm, respectively during the RH treatment. The reason of Ca and Mg decrease was considered to its vaporization.^28,29) This behavior is quite important in understanding the evolution of inclusions. Total O content was measured as 25 ppm during the LF treatment and it decreased to less than 10 ppm after the RH treatment.

3.2. Variation of Inclusion Compositions

Figures 2(a) to 2(d) show the variation of inclusion compositions at each stage. At the step of LF initial (a), inclusion compositions were located at the vicinity of Al₂O₃, which was resulted from Al deoxidation.¹⁵⁾ Subsequently, inclusion compositions spread simultaneously to MgO·Al₂O₃ and the CaO–Al₂O₃ system at the step of LF middle (b). At the end of LF (c), inclusion compositions were positioned on the tie-line connecting the two areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq, but majority was at the MgO·Al₂O₃ side. Lastly, at the end of RH (d), the CaO–Al₂O₃ system inclusions were still remained in the molten steel while MgO·Al₂O₃ inclusions disappeared and Al₂O₃ inclusions were detected again.

Fig. 2.

Composition of inclusions at each stage (a) LF initial; (b) LF middle; (c) LF end; (d) RH end.

The element mappings of the typical inclusions are shown in Figs. 3(a), 3(b), 3(c) and 3(d), which indicate LF initial, LF middle, LF end and RH end, respectively. Obviously, Al₂O₃ inclusion formed at LF initial (a). At the middle of LF (b), three species of inclusions were observed that were Al₂O₃, MgO·Al₂O₃ and the CaO–Al₂O₃ system with the elements homogeneously distributed. At the end of LF (c), the inclusions of the CaO–MgO–Al₂O₃ system with MgO-rich phase core were observed along with MgO·Al₂O₃ and the CaO–Al₂O₃ system inclusions with the elements homogeneously distributed. At the RH end (d), Al₂O₃ inclusions were detected together with the homogeneous CaO–Al₂O₃ system inclusions. Accordingly, Fig. 3 evidently shows the variations described in Fig. 2.

Fig. 3.

Elemental mappings of typical oxide at each stage (a) Al₂O₃ at LF initial; (b1) Al₂O₃ at LF middle; (b2) CaO–Al₂O₃ at LF middle; (b3) MgO·Al₂O₃ at LF middle; (c1) Al₂O₃ at LF end; (c2) CaO–MgO–Al₂O₃ at LF end; (c3) MgO·Al₂O₃ at LF end; (d1) Al₂O₃ at RH end; (d2) CaO–Al₂O₃ at RH end.

Further observation was made to more clearly identify what oxide phases consisted of the inclusions whose compositions were plotted on the tie-line connecting the two areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq. The result is shown in Fig. 4 with the liquidus area at 1873 K calculated by FactSage. Some inclusions at the end of LF consisted of two separate phases of MgO·Al₂O₃ and the CaO–Al₂O₃ system while one was the single phase inclusion of the CaO–Al₂O₃ system. It is evident that the CaO–Al₂O₃ system inclusions were melt during refining because it is within the liquidus line. This observation has proved that solidus MgO·Al₂O₃ and the CaO–Al₂O₃ system are able to coexist. This is the result same as previously reported.^{11,12,13,14,15,16,17,18)}

Fig. 4.

Morphology and composition of inclusions at LF end.

4. Discussion

As shown above, inclusion compositions varied from the LF refining to RH treatment as the sequence of Al₂O₃→MgO·Al₂O₃ + CaO–Al₂O₃→Al₂O₃ + CaO–Al₂O₃. In the following section, the mechanism is discussed from the view point of equilibrium as well as interfacial properties between the inclusions and the molten steel.

4.1. Equilibrium

Thermodynamic calculations were carried out under the present experimental conditions. The used parameters include the measured contents of Al, Ca and Mg as well as the values shown in Table 2 and the first and second order interaction coefficients listed in Table 3.^{30,31,32,33,34,35)} It is noticed that the effect of Mo on the activity coefficient of the other components was not taken into account due to the lack of thermodynamic data.³⁾ The temperature substituted to the calculations was 1873 K considering the actual operation condition. Paying attention to the observed phases, thermodynamic calculations were carried out to output the stability diagrams considering the three systems of (1) MgO–Al₂O₃, (2) CaO–Al₂O₃ and (3) CaO–MgO–Al₂O₃.

Table 3. First and second order interaction coefficients used in the present study.

j i	C	Si	Mn	Cr	Al	Ca	Mg	O
Ca	−0.3431)	−0.096	−0.0156	0.014	−0.072	–	–	−78035)
Mg	−0.31	−0.096	–	0.022	−0.27	–	–	−43038)
Al	0.091	0.056	−0.004	0.0096	0.043	−0.047	–	−1.9836)
O	−0.45	−0.131	−0.021	−0.052	−1.1736)	−31035)	−28038)	−0.17

^*All data without notation are from Ref. 30

$r_{\mathrm{O}}^{\mathrm{M}\mathrm{g},\mathrm{A}\mathrm{l}}=-150,r_{\mathrm{O}}^{\mathrm{C}\mathrm{a},\mathrm{A}\mathrm{l}}=0,r_{\mathrm{O}}^{\mathrm{C}\mathrm{a},\mathrm{M}\mathrm{g}}=0,r_{\mathrm{M}\mathrm{g}}^{\mathrm{O},\mathrm{A}\mathrm{l}}=-120,r_{\mathrm{A}\mathrm{l}}^{\mathrm{O},\mathrm{M}\mathrm{g}}=-260$ ⋅⋅⋅(Ref. 32)

$r_{\mathrm{O}}^{\mathrm{C}\mathrm{a}}=-18\mathrm{\hspace{0.17em} }000,r_{\mathrm{C}\mathrm{a}}^{\mathrm{O}}=-650\mathrm{\hspace{0.17em} }000,r_{\mathrm{C}\mathrm{a}}^{\mathrm{C}\mathrm{a},\mathrm{O}}=-520\mathrm{\hspace{0.17em} }000,r_{\mathrm{O}}^{\mathrm{C}\mathrm{a},\mathrm{O}}=-90\mathrm{\hspace{0.17em} }000$ ⋅⋅⋅(Ref. 35)

$r_{\mathrm{O}}^{\mathrm{M}\mathrm{g}}=-350\mathrm{\hspace{0.17em} }000,r_{\mathrm{M}\mathrm{g}}^{\mathrm{O}}=-20\mathrm{\hspace{0.17em} }000,r_{\mathrm{M}\mathrm{g}}^{\mathrm{M}\mathrm{g},\mathrm{O}}=-61\mathrm{\hspace{0.17em} }000,r_{\mathrm{O}}^{\mathrm{M}\mathrm{g},\mathrm{O}}=-462\mathrm{\hspace{0.17em} }000$ ⋅⋅⋅(Ref. 38)

$r_{\mathrm{O}}^{\mathrm{A}\mathrm{l}}=40,r_{\mathrm{A}\mathrm{l}}^{\mathrm{O}}=-0.01,r_{\mathrm{A}\mathrm{l}}^{\mathrm{A}\mathrm{l},\mathrm{O}}=-0.0284,r_{\mathrm{O}}^{\mathrm{A}\mathrm{l},\mathrm{O}}=47.4$ ⋅⋅⋅(Ref. 36)

4.1.1. MgO–Al₂O₃ System

Equations (1) and (2) were used to determine the most stable phase in the MgO–Al₂O₃ system.³²⁾ The MgO/MgO·Al₂O₃ boundary was drawn by considering Eq. (1), with the activity data of MgO=1 and MgO·Al₂O₃=0.80.³³⁾ The MgO·Al₂O₃/Al₂O₃ boundary was drawn by considering Eq. (2), with the activity data of MgO·Al₂O₃=0.47 and Al₂O₃=1.³³⁾ The activities of components in molten steel were relative to a dilute solution of one mass% standard state. The activities of the oxides are relative to pure solids.   

$$\begin{array}{l}4\mathrm{M}\mathrm{g}\mathrm{O}(\mathrm{s})+2\underset{\_}{\mathrm{A}\mathrm{l}}=\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+3\underset{\_}{\mathrm{M}\mathrm{g}}\\ \mathrm{l}\mathrm{o}\mathrm{g}{K}_{\mathrm{M}\mathrm{g}\mathrm{O}/\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}=-23.856+32\mathrm{\hspace{0.17em} }280/T\end{array}$$(1) ³²⁾

  

$$\begin{array}{l}3\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+2\underset{\_}{\mathrm{A}\mathrm{l}}=4\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+3\underset{\_}{\mathrm{M}\mathrm{g}}\\ \mathrm{l}\mathrm{o}\mathrm{g}{K}_{\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3/\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}=-26.274+27\mathrm{\hspace{0.17em} }950/T\end{array}$$(2) ³²⁾

where K is equilibrium constant and T is temperature (K). The result is shown in Fig. 5 with the plots of the measured Al and Mg contents at each stage. Under the condition of this practice, it is clear that the most stable phase considering the MgO–Al₂O₃ system is MgO·Al₂O₃. The result is quite consistent with the observations since MgO·Al₂O₃ was one of the detected phases at the LF middle and end, proving that MgO·Al₂O₃ formation is thermodynamically stable accounting for Al and Mg contents.

Fig. 5.

Phase stability diagram of MgO/MgO·Al₂O₃/Al₂O₃.

4.1.2. CaO–Al₂O₃ System

Referring to the method proposed by Taguchi et al.,³⁴⁾ Eqs. (3)³⁵⁾ and (4)³⁶⁾ were used to determine the most stable phase in the CaO–Al₂O₃ system. Table 4 shows the activities of CaO and Al₂O₃ at each boundary calculated by FactSage, which were substituted to these equations.   

$$\mathrm{C}\mathrm{a}\mathrm{O}(\mathrm{s})=\underset{\_}{\mathrm{C}\mathrm{a}}+\underset{\_}{\mathrm{O}}\hspace{1em}\mathrm{l}\mathrm{o}\mathrm{g}{K}_{\mathrm{C}\mathrm{a}\mathrm{O}}=-3.29-7\mathrm{\hspace{0.17em} }220/T$$(3) ³⁵⁾

  

$$\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})=2\underset{\_}{\mathrm{A}\mathrm{l}}+3\underset{\_}{\mathrm{O}}\hspace{1em}\mathrm{l}\mathrm{o}\mathrm{g}{K}_{\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}=11.62-45\mathrm{\hspace{0.17em} }300/T$$(4) ³⁶⁾

Table 4. Activities of CaO and Al₂O₃ in various boundaries of CaO–Al₂O₃ system at 1873 K.

Boundary	a(CaO)	a(Al2O3)
Al2O3/CaO·6Al2O3	0.0049	1.0
CaO·6Al2O3/CaO·2Al2O3	0.010	0.88
CaO·2Al2O3/CaO·Al2O3	0.10	0.29
CaO·Al2O3/CaO–Al2O3 liq	0.17	0.18
CaO–Al2O3 liq/CaO	0.99	0.0089

The result is shown in Fig. 6 with the plots of the measured Al and Ca contents at each step. Under the condition of this practice, the most stable phase of the CaO–Al₂O₃ system is clearly CaO-Al₂O_{3 liq}, which agreed to the phases detected in the LF end and the RH end. Thus, the formation of the liquid CaO–Al₂O₃ inclusions is thermodynamically reasonable by considering Al and Ca contents.

Fig. 6.

Phase stability diagram of various CaO–Al₂O₃. (^*Al₂O₃ is abbreviated as A, CaO·6Al₂O₃ as CA6, CaO·2Al₂O₃ as CA2, CaO·Al₂O₃ as CA, liquid CaO–Al₂O₃ as C–A_liq and CaO as C.)

Considering phase rule, inclusion compositions should converge to a single phase. However, multiple phases were observed in this study. It might imply that non-equilibrium state could be established in the molten steel by which MgO·Al₂O₃ and the CaO-Al₂O₃ system inclusions might be able to coexist. It is probable that the distribution of Ca and Mg⁷⁾ is not completely uniform in the molten steel in practice. For example, just beneath the slag, Ca and Mg contents may be higher than those in the bulk steel. The state, where two species can coexist at a time, could be established during the transitional period of inclusion compositions.

4.1.3. CaO–MgO–Al₂O₃ System

As explained above, the coexistence of two different inclusion compositions may imply the temporary non-equilibrium state. To determine the most stable phase of the three species of (1) Al₂O₃, (2) MgO–Al₂O₃ system and (3) CaO–Al₂O₃ system, the stability diagram was calculated for the CaO–MgO–Al₂O₃ system referring to the method proposed by Deng.¹⁵⁾

CaO·Al₂O₃ was chosen as the phase of the CaO–Al₂O₃ system. The Al₂O₃/MgO·Al₂O₃ boundary was calculated by the Eq. (2), the Al₂O₃/CaO·Al₂O₃ boundary was by Eq. (6) which was derived by the combination of Eqs. (3), (4), and (5).³⁷⁾ The MgO·Al₂O₃/CaO·Al₂O₃ boundary was obtained by Eq. (9) which was derived by the combination of Eqs. (3), (5), (7),³⁸⁾ and (8).³³⁾ In the calculations, the activities of all the oxides were assumed as unity and Al content was taken as 0.035 mass%.   

$$\begin{array}{l}\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})=\mathrm{C}\mathrm{a}\mathrm{O}(\mathrm{s})+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})\hspace{1em}\\ \mathrm{\Delta }{G}_{\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}^0=17\mathrm{\hspace{0.17em} }910+17.38T\end{array}$$(5) ³⁷⁾

  

$$\begin{array}{l}3\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+2\underset{\_}{\mathrm{A}\mathrm{l}}=4\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+3\underset{\_}{\mathrm{C}\mathrm{a}}\\ \mathrm{\Delta }{G}_{\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3/\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}^0=-398\mathrm{\hspace{0.17em} }934+463.64T\end{array}$$(6)

  

$$\mathrm{M}\mathrm{g}\mathrm{O}(\mathrm{s})=\underset{\_}{\mathrm{M}\mathrm{g}}+\underset{\_}{\text{O}} \hspace{1em}\mathrm{l}\mathrm{o}\mathrm{g}{K}_{\mathrm{M}\mathrm{g}\text{O}}=-4.28-4\mathrm{\hspace{0.17em} }700/T$$(7) ³⁸⁾

  

$$\begin{array}{l}\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})=\mathrm{M}\mathrm{g}\mathrm{O}(\mathrm{s})+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})\\ \mathrm{\Delta }{G}_{\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}^0=20\mathrm{\hspace{0.17em} }682+11.57T\end{array}$$(8) ³³⁾

  

$$\begin{array}{l}\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+\underset{\_}{\mathrm{C}\mathrm{a}}=\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})+\underset{\_}{\mathrm{M}\mathrm{g}}\\ \mathrm{l}\mathrm{o}\mathrm{g}{K}_{\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3/\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3}=-0.689-2\mathrm{\hspace{0.17em} }360/T\end{array}$$(9)

Figure 7 presents the stability diagram of the CaO–MgO–Al₂O₃ system with the plots of the measured Mg and Ca contents in each process. This result indicates that the most thermodynamically stable phase is CaO·Al₂O₃ throughout the LF-RH processes. There are a number of reports concerning the thermodynamically stable phase of the CaO–MgO–Al₂O₃ system. Most of the researchers have pointed out that the most stable phase is the CaO–Al₂O₃ system when the steel contains Ca more than 1 ppm.^7,8,11) The steel samples in this research contained Ca of around 6 ppm and at least 2 ppm. This fact proves that the calculation result is fully consistent.

Fig. 7.

Phase stability diagram of MgO·Al₂O₃/CaO·Al₂O₃/Al₂O₃.

From the equilibrium point of view, the variation of the inclusion compositions in this research can be expressed as follows. At the initial stage of the LF refining, inclusion composition were located at the vicinity of Al₂O₃, which was the primary deoxidation product immediately after the steel was killed by Al. Al₂O₃ inclusions thereafter evolved toward both MgO·Al₂O₃ and the CaO–Al₂O₃ system under the non-equilibrium states at the period between the middle and the end of the LF refining. The CaO–Al₂O₃ system inclusions detected at the end of RH treatment was the most stable phase in the CaO–MgO–Al₂O₃ system.

However, the above thermodynamic calculations cannot describe the formation of Al₂O₃ inclusions observed at the end of RH. This may imply that those Al₂O₃ inclusions were generated in the manner different from the endogenous composition evolution. This reason will be discussed later in detail.

4.2. Factors to Determine Inclusion Composition

It is generally recognized that the factors of the variation of the inclusion compositions during secondary refining process includes the phenomena of generation,^2,19) removal^{20,21,22,23,24)} and composition evolution according to equilibrium state.^{6,7,8,9,10,11,12,13,14,15,16,17,18)} Discussion in the following sections focuses on how each phenomenon is related to the variation at each process based on the thermodynamic calculation results.

4.2.1. Contribution of Generation and Removal during LF Refining Process

Regarding the formation of inclusions during LF refining, a possible cause is the local oxygen supply from slag or atmosphere, which creates Al₂O₃ via FeO, MnO, or SiO₂ according to Mukai et al.³⁹⁾ Another cause could be the slag entrapment caused by the gas stirring. Despite these possibilities, the inclusions detected in the LF end did not locate at the vicinity of Al₂O₃. Furthermore, these did not locate at the CaO and MgO double-saturation position, corresponding to the present slag chemistry, at the LF initial. Thus, the inclusion compositions during the LF refining were not the ones affected by the local oxygen supply or the slag entrapment. As mentioned before, the activity of Al₂O₃ in the slag at the CaO and MgO double-saturation position was 0.005 calculated by FactSage, with which the dissolved oxygen content could be calculated⁹⁾ as 2 ppm by Eq. (4). Throughout the LF treatment, the dissolved oxygen content was considered to be almost constant as 2 ppm. Moreover, the total O content was mostly kept constant as explained earlier showing that the amount of the oxide inclusion was almost constant in the molten steel during the LF refining. This means that removal of inclusions did not happen so extensively.

According to the above descriptions, it is concluded that the generation and removal of inclusions during the LF refining process is not considerable. The decrease in Al content during the period of the LF refining seen in Fig. 1 is the proof of the generation of Al₂O₃ inclusions.^2,27) It is considered that the generated Al₂O₃ inclusions were mostly absorbed by the slag bulk at the slag/metal interface.

4.2.2. Composition Evolution during LF Refining Process

The thermodynamic calculations showed that MgO·Al₂O₃ and the CaO–Al₂O₃ system inclusions are more stable than Al₂O₃ resulting in the spontaneous changes of Al₂O₃→MgO·Al₂O₃ and Al₂O₃→CaO–Al₂O₃ occurring by the reactions between Ca, Mg dissolved in the molten steel and the inclusions. The composition evolution process of Al₂O₃ in molten steel is reported in several manners. Especially, Todoroki et al.,⁷⁾ Jiang et al.¹²⁾ and Deng et al.¹⁵⁾ reported that under the condition of the higher Mg activity rather than Ca, the evolution occurs in the sequence of Al₂O₃→MgO·Al₂O₃ in advance of Al₂O₃→CaO–Al₂O₃. Following this theory, the effects of the Ca and Mg activities on the composition evolution of Al₂O₃ was examined under the present conditions. The calculations to obtain the activity coefficients of the components were carried out by using the values shown in Table 3.

The variation of the activities of Ca and Mg are shown in Fig. 8. This result shows that the values of Mg activity are 1/2–1/5 times of those of Ca activity. Supposing that the above theory would be applicable, this condition would address the composition change of Al₂O₃→CaO–Al₂O₃ rather than Al₂O₃→MgO·Al₂O₃. As explained above, however, the both reactions of Al₂O₃→CaO–Al₂O₃ and Al₂O₃→MgO·Al₂O₃ took place at the same time in this practice. In other words, the evolution direction from Al₂O₃ cannot be explained by the difference of the activities of Mg and Ca. This tells us that each reaction independently takes place with the independent driving force.

Fig. 8.

Variation of Ca and Mg activity in molten steel during secondary refining process.

As mentioned above, Al₂O₃→CaO–Al₂O₃ can progress under the condition of Ca content of 1 ppm or higher, however, it is thought as for the composition evolution that the driving force much greater than the calculated value is required to cause in the actual refining process.¹⁶⁾ In the following, the condition for the inclusion change of Al₂O₃→CaO–Al₂O₃ is proposed by comparing the previous studies.^{7,8,12,15,16,17,18)} The Ca contents were compared to estimate the condition because it was very difficult to compare the activities of Ca because their experimental situations were very different from each other. Some were for carbon steels and tool steels, while some were for stainless steels. The data at the step of 1800 seconds after LF or of the LF middle stage were applied for comparison as shown in Table 5. It should be noted that not all the values listed in Table 5 were shown in the literatures. When the values were expressed as a certain range, the mean ones were taken. In the case of no values described, ones were read from the plots of their figures. As for Ref. 17), it was presumed that Ca content might be similar to ours by considering the slag compositions despite no descriptions regarding Ca content in the middle stage of LF refining.

Table 5. Comparison with the previous studies of molten steels deoxidized with Al at the middle of LF refining.

Slag composition		Steel composition		Inclusion evolution	Ref.
C/S	C/A	Al/mass%	Ca/mass ppm	A → C-A
18	2.0	0.046	6.5	observed	Present work
3.9	1.6	0.035	0.1	not observed	15
no silica	6.4	0.06	1	not observed	7
5.0	6.2	0.11	1	not observed	8
6.5	1.1	0.038	3	not observed	12
5.8	2.2	0.045	4	observed	16
7.8	2.7	0.062	6	observed	16
15.7	1.9	–	–	observed	17
5.5	1.3	0.052	9	observed	18
5.5	1.3	0.052	7	observed	18

^*CaO is abbreviated as C, SiO₂ as S, Al₂O₃ as A and CaO–Al₂O₃ as C–A.

According to this comparison, Ca content of 4 ppm seems to be the approximate threshold whether or not Al₂O₃→CaO–Al₂O₃ reaction proceeds in the Al-killed steels; Al₂O₃→CaO–Al₂O₃ reaction can occur with Ca contents higher than 4 ppm. As shown in Fig. 1, the present steel contained Ca contents higher than the threshold value of 4 ppm throughout the LF refining. Thus, the driving force was considered to be high enough for the two reactions to simultaneously take place from Al₂O₃ toward MgO·Al₂O₃ and the CaO–Al₂O₃ system.

As proven in Fig. 7, MgO·Al₂O₃ is less stable than the CaO–Al₂O₃ system leading to the composition evolution of MgO·Al₂O₃→CaO–Al₂O₃.^{11,12,13,14,15,16,17,18)} This happens along the tie-line connecting the areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq as seen in Fig. 2(c). Many reports^{11,12,13,14,15,16,17,18)} have indicated that it takes a long time for the composition evolution of MgO·Al₂O₃→CaO–Al₂O₃ system. Our result was quite similar to these reports in that the unreacted MgO·Al₂O₃ core surrounded by the CaO–Al₂O₃ system was observed as seen in Fig. 3(c2). Thus, the composition evolution mechanism of MgO·Al₂O₃→CaO–Al₂O₃ system was the same as previously reported.^{12,13,14,15,16,17,18)} According to the stability of the oxide species, Mg atoms diffuse toward outside while Ca atoms diffuse toward inside. Then, inclusion compositions would converge into the CaO–Al₂O₃ system in the LF refining step when the reaction time is longer enough.

As explained above, in the LF refining process, the dominant factors to determine the inclusion compositions are not generation and removal but equilibrium states addressed by the chemical potentials of the deoxidizing elements. The evolution degree and direction are affected by the content of each component in the molten steel.

4.2.3. RH Treatment

This stage is the most important for the steel cleanliness because this is the final refining stage before casting. As Fig. 2 indicates, the inclusion compositions changed from the plots, on the tie-line connecting the areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq, mainly to the vicinity of Al₂O₃ and the CaO–Al₂O₃ system. The effects of generation and removal on the inclusion compositions are considered to understand this change.

The decrease of total O content clearly proves the occurrence of removal of inclusions from the molten steel. Miki et al.²¹⁾ reported that the attachment of inclusions to Ar bubbles plays a significant role of inclusion removal during the RH treatment. Arai et al.²⁴⁾ reported that the particles with the large contact angle against the solvent easily attach to the bubbles and are more likely to be removed. Yin et al.⁴⁰⁾ and Wikström et al.⁴¹⁾ observed the inclusion behavior at the metal/gas interface reporting that attraction force was applied between the solid inclusions which are not wet with the molten steel. On the other hand, attraction force was not applied between the liquid inclusions which can intimately wet with the molten steel. This result indicates that solid inclusions tend to collide and aggregate with each other after attaching to a single bubble while liquid inclusions stay separated. Thus, the interfacial properties of inclusions strongly affect removal behavior during the RH treatment.

From this point of view, the contact angle and the interfacial energy of the oxides should be considered. Table 6^42,43,44,45) clearly shows that MgO·Al₂O₃, solid in the refining temperatures, does not wet while the CaO–Al₂O₃ system, liquid in the refining temperatures, does wet to the molten steel. As well, according to the calculation in 4.1.3, MgO·Al₂O₃ was in the transient stage to the CaO–Al₂O₃ system during the RH treatment. Therefore, it was considered that MgO·Al₂O₃ inclusions not only varied its composition toward the CaO–Al₂O₃ system but also they were separated during the RH treatment. Meanwhile, the CaO–Al₂O₃ system inclusions were remained due to the behavior of wetting well to the molten steel. This was how the variation of MgO·Al₂O₃ to CaO–Al₂O₃ was considered to proceed. For deoxidation, it is primarily important to decrease total O content by decreasing the inclusion population. It could be said that removal of the singular type MgO·Al₂O₃ inclusions from the molten steel was considered to contribute to the decrease of total O content.

Table 6. Interfacial energy and contact angle of various oxides.

Oxide	Interfacial Energy/mJ·m−2	Contact Angle/deg.
Al2O3 (solid)	2050 (at 1873 K)42)	140 (at 1873 K)42)
CaO–Al2O3 (liquid)	1600 (at 1853 K)43)	54–65 (at 1873 K)44)
MgO·Al2O3 (solid)	3700 (at 1823 K)45)	134 (at 1823 K)45)

Obviously in Fig. 4, some MgO·Al₂O₃ phases are coexisted with the CaO–Al₂O₃ system. This indicates that the interfacial properties of these multi-component inclusions are affected by the wetting CaO–Al₂O₃ system, which lowers the removability from the molten steel. This can evidence that some inclusions containing MgO were still left as can be seen in Fig. 2(d).

Al₂O₃ inclusions were also formed again until the end of the RH treatment. According to the thermodynamic consideration of section 4.1, Al₂O₃ inclusions are not stable at this stage under the present Al, Mg and Ca contents. It is reported that there is a certain probability of oxygen supply from slag and/or refractory to the molten steel during the RH treatment.²⁾ Furthermore, it is confirmed that this oxygen supply affects to lower the decreasing rate of total O content by generating Al₂O₃.²⁷⁾ Hence, detected Al₂O₃ inclusions are considered to be generated during the RH treatment. The removal of MgO·Al₂O₃ inclusions could overcome the generation of Al₂O₃ inclusions during RH treatment. This could contribute to the decrease in the total O content during this stage.

4.3. Evolution Mechanism of Inclusion Compositions

According to the experimental results and discussion, the variation mechanism of the inclusion compositions can be summarized as illustrated in Fig. 9. The variation can be classified as the following three steps:

Fig. 9.

Schematic diagram of variation of inclusion compositions during secondary refining process. (^*Al₂O₃ is abbreviated as A, MgO as M, MgO·Al₂O₃ as MA and CaO–Al₂O₃ system as C–A.)

1) At the initial stage of LF refining, the inclusion compositions start at the location of the vicinity of Al₂O₃, which are generated by Al deoxidation.

2) In the stage between the middle and the end of the LF refining, the inclusion composition of Al₂O₃ evolves toward the MgO–Al₂O₃ system and the CaO–Al₂O₃ system simultaneously. This can occur due to the sufficient driving force accounting for Al, Mg and Ca contents. The liquid CaO–Al₂O₃ system inclusions have been proved to be the most stable phase in the CaO–MgO–Al₂O₃ system. So, MgO·Al₂O₃ inclusions can evolve to the CaO–Al₂O₃ system inclusions. Owing to this composition evolution, the inclusion compositions are distributed on the tie-line connecting the areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq.

3) During the RH treatment, MgO·Al₂O₃ inclusions are likely to be removed due to their high interfacial energy against the molten steel. As for the multiphase inclusion with MgO·Al₂O₃ core surrounded by the CaO–Al₂O₃ system, MgO·Al₂O₃ core becomes smaller due to the thermodynamic instability. On the other hand, the CaO–Al₂O₃ system inclusions possess low interfacial energy so that they tend to be remained in the molten steel. In addition, Al₂O₃ inclusions are generated during the RH treatment due to the local oxygen supply. Thus, three phenomena including removal of MgO·Al₂O₃ inclusions, remaining the CaO–Al₂O₃ system inclusions and generation of Al₂O₃ determine the inclusion compositions after the RH treatment.

5. Conclusion

Experiments were undertaken to observe inclusions during the actual secondary refining process. The present study aimed at clarifying the factors to determine the inclusion compositions in practice. The following conclusions summarize this study.

(1) During the LF refining, inclusions changed from the primary deoxidation product of Al₂O₃ to MgO·Al₂O₃ and the CaO–Al₂O₃ system. The compositions were positioned on the tie-line connecting the areas of MgO·Al₂O₃ and CaO–Al₂O₃–MgO_liq. This change took place by the composition evolution targeting the thermodynamic equilibrium states.

(2) After the RH treatment, the inclusion compositions changed mainly to Al₂O₃ and the CaO–Al₂O₃ system. MgO·Al₂O₃ inclusions were removed while the CaO–Al₂O₃ system inclusions were remained, which was described by the interfacial properties between the oxides and the molten steel. Al₂O₃ inclusions were considered to be newly generated during the RH treatment.

(3) It was confirmed that three factors of the equilibrium states, removal and generation of inclusions dominated to determine the inclusion compositions in the molten steel during the LF-RH refining process.

References

• 1)   Y.  Kato: Sanyo Tech. Rep., 2 (1995), 15.
• 2)   K.  Kawakami: 182nd and 183rd Nishiyama Memorial seminar, ISIJ, Tokyo, (2004), 159.
• 3)   H.  Todoroki,  F.  Kirihara,  Y.  Kanbe and  Y.  Miyazaki: Tetsu-to-Hagané, 100 (2014), 539.
• 4)   H.  Nishimori,  T.  Masuda,  K.  Kawakami,  K.  Furumura and  K.  Hashizume: Bull. Jpn. Inst. Met., 32 (1993), 441.
• 5)   J.  Monnot,  B.  Heritier and  J. Y.  Cogne: ASTM STP987, Philadelphia, (1988), 149.
• 6)   G.  Ye,  P.  Jönsson and  T.  Lund: ISIJ Int., 36 (1996), S105.
• 7)   H.  Todoroki and  K.  Mizuno: ISS Trans., 30 (2003), 60.
• 8)   H.  Todoroki and  K.  Mizuno: ISIJ Iint., 44 (2004), 1350.
• 9)   J. H.  Park and  H.  Todoroki: ISIJ Int., 50 (2010), 1333.
• 10)   A.  Galindo,  G.  Irons,  S.  Sun and  K.  Coley: Proc. CTSSC-EMI Symp., CTSSC-EMI, Tokyo, (2015), 22.
• 11)   Y.  Kang,  F.  Li,  K.  Morita and  D.  Sichen: Steel Res., 77 (2006), 785.
• 12)   M.  Jiang,  X.  Wang,  B.  Chen and  W.  Wang: ISIJ Int., 50 (2010), 95.
• 13)   S.  Yang,  Q.  Wang,  L.  Zhang,  J.  Li and  K.  Peaslee: Metall. Mater. Trans. B, 43B (2012), 731.
• 14)   X.  Wang,  M.  Jiang,  B.  Chen and  W.  Wang: Adv. Steels, 6 (2011), 485.
• 15)   Z.  Deng and  M.  Zhu: ISIJ Int., 53 (2013), 450.
• 16)   W.  Yang,  L.  Zhang,  X.  Wang,  Y.  Ren,  X.  Liu and  Q.  Shan: ISIJ Int., 53 (2013), 1401.
• 17)   G.  Jing,  C.  Shu-Sen and  C.  Zi-Jian: ISIJ Int., 53 (2013), 2142.
• 18)   D.  Zhao,  H.  Li,  C.  Bao and  J.  Yang: ISIJ Int., 55 (2015), 2115.
• 19)   K.  Kawakami,  T.  Taniguchi and  K.  Nakashima: Tetsu-to-Hagané, 93 (2007), 743.
• 20)   S.  Takeuchi: 182nd and 183rd Nishiyama Memorial seminar, ISIJ, Tokyo, (2004), 72.
• 21)   Y.  Miki,  Y.  Shimada,  B. G.  Thomas and  A.  Denissov: Iron Steelmaker, 24 (1997), 31.
• 22)   T.  Murai,  H.  Matsuno,  E.  Sakurai and  H.  Kawashima: Tetsu-to-Hagané, 84 (1998), 13.
• 23)   H.  Arai,  S.  Nakamura,  A.  Okayama,  K.  Matsumoto and  S.  Taniguchi: Tetsu-to-Hagané, 92 (2006), 6.
• 24)   H.  Arai,  N.  Iida,  K.  Matsumoto,  S.  Taniguchi,  T.  Nakaoka,  K.  Kunii and  K.  Sakamoto: Tetsu-to-Hagané, 93 (2007), 1.
• 25)   M.  Harris,  O.  Adaba,  S.  Lekakh,  R.  O’Maller and  V.  Richards: Proc. AISTech, Warrendale, PA, (2015), 3315.
• 26)  Verlag Stahleisen GmbH, Ed.: Slag Atlas, 2nd ed., Verlag Stahleisen GmbH, Düsseldorf, (1995), 104.
• 27)   Y.  Higuchi,  Y.  Shirota,  T.  Obana and  H,  Ikenaga: CAMP-ISIJ, 4 (1991), 266.
• 28)   G. G.  Mikhailov and  E. P.  Baibulenko: Steel USSR, 11 (1981), 443.
• 29)   X.  Zhang, Q. Han and D. Chen: Metall. Trans. B, 22B (1991), 918.
• 30)  Steelmaking Data Sourcebook, 19th committee Steelmaking of JSPS, Tokyo, (1984).
• 31)  Thermodynamic Data for Steelmaking, eds. by M. Hino and K. Ito, Tohoku University Press, Sendai, (2010).
• 32)   H.  Itoh,  M.  Hino and  H.  Ban-ya: Tetsu-to-Hagané, 84 (1998), 85.
• 33)   K.  Fujii,  T.  Nagasaka and  M.  Hino: ISIJ Int., 40 (2000), 1059.
• 34)   K.  Taguchi,  H.  Ono-Nakazato,  T.  Usui,  K.  Marukawa,  K.  Katogi and  H.  Kosaka: ISIJ Int., 45 (2005), 1572.
• 35)   H.  Itoh,  M.  Hino and  H.  Ban-ya: Tetsu-to-Hagané, 83 (1997), 695.
• 36)   H.  Itoh,  M.  Hino and  H.  Ban-ya: Tetsu-to-Hagané, 83 (1997), 773.
• 37)   K.  Nagata,  J.  Tanabe and  K.  Goto: Tetsu-to-Hagané, 75 (1989), 2023.
• 38)   H.  Itoh,  M.  Hino and  H.  Ban-ya: Tetsu-to-Hagané, 83 (1997), 623.
• 39)   K.  Mukai,  H.  Sakao and  K.  Sano: J. Jpn. Inst. Met., 11 (1968), 1155.
• 40)   H.  Yin,  H.  Shibata,  T.  Emi and  M.  Suzuki: ISIJ int., 37 (1997), 946.
• 41)   J.  Wikström,  K.  Nakajima,  H.  Shibata,  A.  Tilliander and  P.  Jönsson: Ironmaking Steelmaking, 35 (2008), 589.
• 42)   K.  Nogi and  K.  Ogino: Can. Metall. Q., 22 (1983), 19.
• 43)   K.  Ogino,  S.  Hara,  T.  Miwa and  S.  Kimoto: Tetsu-to-Hagané, 65 (1979), 2012.
• 44)   A. W.  Cramb and  I.  Jimbo: Steel Res., 60 (1989), 157.
• 45)   N.  Zouvelou,  X.  Mantzouris and  P.  Nikolopoulos: Int. J. Adhes. Adhes., 27 (2007), 380.