Modiﬁcation Mechanism of Spinel Inclusions in Medium Manganese Steel with Rare Earth Treatment

: In aluminum deoxidized medium manganese steel, spinel inclusions are easily to form during reﬁning, and such inclusions will deteriorate the toughness of the medium manganese steel. Rare earth inclusions have a smaller hardness, and their thermal expansion coe ﬃ cients are similar to that of steel. They can avoid large stress concentrations around inclusions during the heat treatment of steel, which is beneﬁcial for improving the toughness of steel. Therefore, rare earth Ce is usually used to modify spinel inclusions in steel. In order to clarify the modiﬁcation mechanism of spinel inclusions in medium manganese steel with Ce treatment, high-temperature simulation experiments were carried out. Samples were taken step by step during the experimental steel smelting process, and the inclusions in the samples were analyzed by SEM-EDS. Finally, the experimental results were discussed and analyzed in combination with thermodynamic calculations. The results show that after Ce treatment, the amount of inclusions decrease, the inclusion size is basically less than 5 µ m, and the spinel inclusions are transformed into rare earth inclusions. After Ce addition, Mn and Mg in the spinel inclusions are ﬁrst replaced by Ce, and the spinel structure is destroyed to form CeAlO 3 . When the O content in the steel is low, S in the steel will replace the O in the inclusion, and CeAlO 3 and spinel inclusions will be transformed into Ce 2 O 2 S. By measuring the total oxygen content of the steel, the total Ce content required for complete modiﬁcation of spinel inclusions can be obtained. Finally, the critical conditions for the formation and transformation of inclusions in the Fe-Mn-Al-Mg-Ce-O-S system at 1873K were obtained according to thermodynamic calculations.


Introduction
Medium manganese steel is a typical representative of third generation automotive steel with high strength and high plasticity. It can be applied to many body structure parts including automobile A-pillars, B-pillars, anticollision beams, and sill reinforcement. Industrial experimental studies have shown that (Mn,Mg)O·Al 2 O 3 -and MgO·Al 2 O 3 -type spinel inclusions will form during refining of aluminum deoxidized medium manganese steel [1]. Spinel inclusions are generally angular and have high hardness, which is the main cause of stamping and cracking defects in automotive steel [2,3]. Therefore, it is necessary to modify the spinel inclusions to improve the toughness of medium manganese steel. At present, low-melting CaO-MgO-Al 2 O 3 inclusions obtained by calcium treatment are generally expected to avoid the damage of hard inclusions [4,5]. However, large-size inclusions cannot be completely modified due to kinetic conditions, which means that the spinel phase core of the complex inclusions will still exist and be exposed after rolling [6].
The rare earth element Ce has a good modification effect on the spinel [7][8][9], which can modify the hard oxide inclusions into rare earth inclusions with a soft hardness [10]. Huang et al. [7,9] Metals 2019, 9, 804 2 of 13 and Wang et al. [8] respectively studied the modification behavior of Ce on inclusions in drill steel and high-speed railway steel, and the transformation mechanism of spinel to Ce 2 O 3 and Ce 2 O 2 S was described. However, there are relatively few reports on rare earth modified spinel inclusions. The modification mechanism of Ce on the spinel inclusions is not comprehensive, and the effect of aluminum content on Ce treatment has not been reported. Therefore, in this paper, the modification behaviors of spinel inclusions after Ce treatment in medium manganese steel with different aluminum contents were studied. Meanwhile, the Ce content required for complete modification of spinel inclusions was discussed in combination with thermodynamic calculations, which are intended to guide the appropriate addition of rare earth in the actual production process.

Experimental Procedure
High-temperature simulation experiments were carried out by using a MoSi 2 furnace (TianJin Muffle Technology Co. Ltd, Tianjin, China). According to the general composition of medium manganese steel, the target Al content was set to three levels: low aluminum (0.02%), numbered R1; medium aluminum (0.2%), numbered R2; and high aluminum (2.0%), numbered R3. The experimental steps are as follows. Approximately 400 g of pure iron and an MnFe alloy was placed in a corundum crucible and then placed in the constant temperature zone of the furnace. The master irons (pure iron and MnFe alloy) were heated to 1873 K under argon atmosphere with a flow rate of 0.4 L min −1 . Subsequently, the oxygen activity in the molten steel was measured by an electrolyte oxygen probe, and it was defined as the initial oxygen activity (No. 1 a [O] ). Based on the target Al content, a certain amount of Al wire was added, and NiMg alloy was added 10 min later. No. 2 a [O] was measured by electrolyte oxygen probe after another 10 min from the addition of NiMg alloy, and then sample 0 was taken by a quartz tube and quenched in an ice bath. Finally, CeFe alloy was added into the molten steel to modify inclusions. Thereafter, the molten steel was respectively held for 1, 10, and 30 min at 1873 K (1600 • C) and then successively sampled as above (sample 1-sample 3). After taking sample 2, the oxygen activity was measured by an electrolyte oxygen probe, which was defined as the final oxygen activity (No. 3 a [O] ). The experimental process is provided in Figure 1. Table 1 shows the compositions of raw materials, including pure iron, MnFe alloy, Al wire, NiMg alloy, and CeFe alloy. Wang et al. [8] respectively studied the modification behavior of Ce on inclusions in drill steel and high-speed railway steel, and the transformation mechanism of spinel to Ce2O3 and Ce2O2S was described. However, there are relatively few reports on rare earth modified spinel inclusions. The modification mechanism of Ce on the spinel inclusions is not comprehensive, and the effect of aluminum content on Ce treatment has not been reported. Therefore, in this paper, the modification behaviors of spinel inclusions after Ce treatment in medium manganese steel with different aluminum contents were studied. Meanwhile, the Ce content required for complete modification of spinel inclusions was discussed in combination with thermodynamic calculations, which are intended to guide the appropriate addition of rare earth in the actual production process.

Experimental Procedure
High-temperature simulation experiments were carried out by using a MoSi2 furnace (TianJin Muffle Technology Co. Ltd, Tianjin, China). According to the general composition of medium manganese steel, the target Al content was set to three levels: low aluminum (0.02%), numbered R1; medium aluminum (0.2%), numbered R2; and high aluminum (2.0%), numbered R3. The experimental steps are as follows. Approximately 400 g of pure iron and an MnFe alloy was placed in a corundum crucible and then placed in the constant temperature zone of the furnace. The master irons (pure iron and MnFe alloy) were heated to 1873 K under argon atmosphere with a flow rate of 0.4 L min −1 . Subsequently, the oxygen activity in the molten steel was measured by an electrolyte oxygen probe, and it was defined as the initial oxygen activity (No. 1 a [O]). Based on the target Al content, a certain amount of Al wire was added, and NiMg alloy was added 10 min later. No. 2 a [O] was measured by electrolyte oxygen probe after another 10 min from the addition of NiMg alloy, and then sample 0 was taken by a quartz tube and quenched in an ice bath. Finally, CeFe alloy was added into the molten steel to modify inclusions. Thereafter, the molten steel was respectively held for 1, 10, and 30 min at 1873 K (1600 °C) and then successively sampled as above (sample 1-sample 3). After taking sample 2, the oxygen activity was measured by an electrolyte oxygen probe, which was defined as the final oxygen activity (No. 3 a[O]). The experimental process is provided in Figure 1. Table 1 shows the compositions of raw materials, including pure iron, MnFe alloy, Al wire, NiMg alloy, and CeFe alloy.   ----CeFe alloy  79  ------19  --2 To observe inclusions of steel, the steel samples were polished by SiC paper and diamond suspensions. The chemical compositions and morphology of inclusions were analyzed through a scanning electron microscope (SEM, Phenom Prox, Eindhoven, The Netherlands) and an energy dispersive spectroscope (EDS, Phenom Prox, Eindhoven, The Netherlands). The contents of  To observe inclusions of steel, the steel samples were polished by SiC paper and diamond suspensions. The chemical compositions and morphology of inclusions were analyzed through a scanning electron microscope (SEM, Phenom Prox, Eindhoven, The Netherlands) and an energy dispersive spectroscope (EDS, Phenom Prox, Eindhoven, The Netherlands). The contents of aluminum, manganese, magnesium, and cerium in the steel were determined by using an inductively coupled plasma optical emission spectrometer (ICP-OES, PerkinElmer, Waltham, MA, USA). The content of sulfur was measured by a carbon and sulfur analyzer. The total oxygen content was measured by an oxygen and nitrogen analyzer. The contents of each element in the three steels (numbered R1, R2, and R3) are listed in Table 2.

Inclusion Transformation
The types of typical inclusions in the experimental steel are shown in Figure 2; there are five types in common. The type 1 inclusions are Al 2 O 3 , as shown in Figure   aluminum, manganese, magnesium, and cerium in the steel were determined by using an inductively coupled plasma optical emission spectrometer (ICP-OES, PerkinElmer, Waltham, MA, USA). The content of sulfur was measured by a carbon and sulfur analyzer. The total oxygen content was measured by an oxygen and nitrogen analyzer. The contents of each element in the three steels (numbered R1, R2, and R3) are listed in Table 2.

Inclusion Transformation
The types of typical inclusions in the experimental steel are shown in Figure 2; there are five types in common. The type 1 inclusions are Al2O3, as shown in Figure 2a. Almost no obvious Mn element was detected in this type of inclusion. Both the type 2 and type 3 are spinel inclusions, namely (Mn,Mg)O·Al2O3 and MgO·Al2O3 spinel, respectively. The type 4 and type 5 inclusions are rare earth inclusions, and they appear bright white under SEM backscatter conditions. The type 4 inclusions are CeAlO3, which are mostly block-shaped and have a small size of about 2~5 μm. The type 5 inclusions are Ce2O2S, which are mostly spherical and have a size of about 3~8 μm. After adding Ce, in addition to the above inclusions with uniform composition, there was a small amount of complex inclusions in the steel that underwent modification, as shown in Figure 3. In Figure 3a, the inclusion core is Mg-Al-O, and the outer layer is Ce-Al-O. In Figure 3b, the inclusion core is Mg-Al-O, and the outer layer is Ce-O-S. In Figure 3c, the inclusion is divided into three layers. The core is Mg-Al-O, the middle layer is Ce-Al-O, and the outermost layer is Ce-O-S. In Figure 3d, the inclusion core is Ce-O-S, and the outer layer is Ce-Al-O. After adding Ce, in addition to the above inclusions with uniform composition, there was a small amount of complex inclusions in the steel that underwent modification, as shown in Figure 3. In Figure 3a, the inclusion core is Mg-Al-O, and the outer layer is Ce-Al-O. In Figure 3b, the inclusion core is Mg-Al-O, and the outer layer is Ce-O-S. In Figure 3c, the inclusion is divided into three layers. The core is Mg-Al-O, the middle layer is Ce-Al-O, and the outermost layer is Ce-O-S. In Figure 3d, the inclusion core is Ce-O-S, and the outer layer is Ce-Al-O. The types of uniform inclusions and complex inclusions for each sampling period are listed in Table 3. It can be seen that the inclusions in the three experimental steels were mainly spinel after aluminum deoxidation and magnesium treatment, and there was also a small amount of Al2O3 inclusions. The R1 experimental steel mainly contained (Mn,Mg)O·Al2O3 inclusions. The spinel inclusions in the R2 and R3 experimental steels were mainly MgO·Al2O3.

Samples in R2 Steel
Note: √√ represents the main type of inclusions, and √ represents a small number of inclusions. The types of uniform inclusions and complex inclusions for each sampling period are listed in Table 3. It can be seen that the inclusions in the three experimental steels were mainly spinel after aluminum deoxidation and magnesium treatment, and there was also a small amount of Al 2 O 3 inclusions. The R1 experimental steel mainly contained (Mn,Mg)O·Al 2 O 3 inclusions. The spinel inclusions in the R2 and R3 experimental steels were mainly MgO·Al 2 O 3 .

The Number Density and Size Change of Inclusions
In order to investigate the influence of Ce treatment on the number and size of inclusions in steel, Image pro plus software (Version 6.0, Media Cybernetics, Shanghai, China) was used to count the number density (number of inclusions per square millimeter) and size (equivalent diameter) of inclusions for each sampling period in the three experimental steels. Each sample counted 150 fields of view at 2000 magnification, for a total area of 1.5 mm 2 . The number density and size change of inclusions during the experiment are shown in Figure 4.
It can be seen from Figure 4 that the variation law of the size and number density of inclusions in the three experimental steels is basically identical. After Ce treatment for 1 min, the proportion of inclusions with a size 0.5-2.0 µm (see Figure 4a) and the number density of inclusions (see Figure 4b) increased significantly. With the prolongation of treatment time, the number density of inclusions showed a decreasing trend, and the size of inclusions showed a slight increasing trend. However, inclusions with a size less than 5 µm were still dominant after 30 min.  It can be seen from Figure 4 that the variation law of the size and number density of inclusions in the three experimental steels is basically identical. After Ce treatment for 1 min, the proportion of inclusions with a size 0.5-2.0 μm (see Figure 4a) and the number density of inclusions (see Figure 4b) increased significantly. With the prolongation of treatment time, the number density of inclusions showed a decreasing trend, and the size of inclusions showed a slight increasing trend. However, inclusions with a size less than 5 μm were still dominant after 30 min.

Thermodynamics Analysis
After Ce treatment, the oxygen activity in the R1 experimental steel decreased (see Table 2), and the inclusions in the steel were mainly CeAlO3 at 30 min after Ce treatment (see Table 3). The oxygen activity in R2 and R3 steels before Ce treatment had dropped to a very low level due to the high Al content. Therefore, the oxygen activity would not be further decreased after Ce addition (see Table  2). The inclusions present in the steels were mainly Ce2O2S at 30 min after Ce treatment, and only a small amount of CeAlO3 inclusions were found in the R2 experimental steel (see Table 3). It indicates that Ce treatment can modify spinel inclusions into rare earth inclusions. The deoxidation reaction between Ce and O will preferentially proceed after the addition of Ce to steel, and Ce will react with S when the O content is at a very low level. The experimental phenomenon is consistent with previous research results [11,12].
To assess the evolution mechanism of inclusions in the steel, an inclusion stability diagram of Fe-10%Mn-0.001%Mg-0.015%S-0.015%Ce-Al-O system at 1873 K was calculated by FactSage TM

Thermodynamics Analysis
After Ce treatment, the oxygen activity in the R1 experimental steel decreased (see Table 2), and the inclusions in the steel were mainly CeAlO 3 at 30 min after Ce treatment (see Table 3). The oxygen activity in R2 and R3 steels before Ce treatment had dropped to a very low level due to the high Al content. Therefore, the oxygen activity would not be further decreased after Ce addition (see Table 2). The inclusions present in the steels were mainly Ce 2 O 2 S at 30 min after Ce treatment, and only a small amount of CeAlO 3 inclusions were found in the R2 experimental steel (see Table 3). It indicates that Ce treatment can modify spinel inclusions into rare earth inclusions. The deoxidation reaction between Ce and O will preferentially proceed after the addition of Ce to steel, and Ce will react with S when the O content is at a very low level. The experimental phenomenon is consistent with previous research results [11,12].
To assess the evolution mechanism of inclusions in the steel, an inclusion stability diagram of Fe-10%Mn-0.001%Mg-0.015%S-0.015%Ce-Al-O system at 1873 K was calculated by FactSage TM program (Version 7.1, developed by Thermfact/CRCT (Montreal, Canada) and GTT-Technologies (Achen, Germany)), as shown in Figure 5. The compositions of the three experimental steels are also marked in it. In the calculation, in addition to FactPS, FToxid, and FSstel databases, a privately created database MYDT containing thermodynamic data of Ce 2 O 2 S [13] was also selected. As can be seen from the figure, the stable inclusion region shows a change of CeAlO 3 → CeAlO 3 + Ce 2 O 2 S → Ce 2 O 2 S as the oxygen content decreases within a fairly large range of Al content in the steel. The composition of R1 experimental steel was located in the CeAlO 3 stable region, and the stable inclusions in the R2 experimental steel were CeAlO 3 + Ce 2 O 2 S, which was consistent with the experimental results (see Table 3). The thermodynamic calculation results of R3 experimental steel showed that there were no inclusions, which was inconsistent with the experimental results. The main reason is that thermodynamic calculations by Factsage have large deviations from the actual situation when the Al content is higher. This phenomenon is also proved by others [14], and a large amount of data indicate that the actual equilibrium oxygen content in steel is lower than the calculated value under high aluminum conditions in steel. Therefore, the activity relationship will be used to express the critical condition of inclusion transformation as follows.
program (Version 7.1, developed by Thermfact/CRCT (Montreal, Canada) and GTT-Technologies (Achen, Germany)), as shown in Figure 5. The compositions of the three experimental steels are also marked in it. In the calculation, in addition to FactPS, FToxid, and FSstel databases, a privately created database MYDT containing thermodynamic data of Ce2O2S [13] was also selected. As can be seen from the figure, the stable inclusion region shows a change of CeAlO3 → CeAlO3 + Ce2O2S → Ce2O2S as the oxygen content decreases within a fairly large range of Al content in the steel. The composition of R1 experimental steel was located in the CeAlO3 stable region, and the stable inclusions in the R2 experimental steel were CeAlO3 + Ce2O2S, which was consistent with the experimental results (see Table 3). The thermodynamic calculation results of R3 experimental steel showed that there were no inclusions, which was inconsistent with the experimental results. The main reason is that thermodynamic calculations by Factsage have large deviations from the actual situation when the Al content is higher. This phenomenon is also proved by others [14], and a large amount of data indicate that the actual equilibrium oxygen content in steel is lower than the calculated value under high aluminum conditions in steel. Therefore, the activity relationship will be used to express the critical condition of inclusion transformation as follows. According to the experimental results, the chemical reactions involved in the Fe-Mn-Al-Mg-Ce-O-S system and related standard Gibbs free energy [15][16][17][18][19][20] are shown in Table 4. The deoxidation equilibrium relationship of the molten steel includes Al-O, Al-Mg-O, Al-Ce-O, and Ce-S-O, and the formation conditions of the corresponding inclusions at 1873 K are as follows.
The formation condition of Al2O3 inclusions is that oxygen activity in steel is greater than that limited by the Al-O equilibrium, as shown in Equation (1) According to the experimental results, the chemical reactions involved in the Fe-Mn-Al-Mg-Ce-O-S system and related standard Gibbs free energy [15][16][17][18][19][20] are shown in Table 4. Table 4. Possible chemical reactions in steel and its standard Gibbs free energy.

No. Reaction Equations in Molten Steel
∆G θ (J/mol) The deoxidation equilibrium relationship of the molten steel includes Al-O, Al-Mg-O, Al-Ce-O, and Ce-S-O, and the formation conditions of the corresponding inclusions at 1873 K are as follows.
The formation condition of Al 2 O 3 inclusions is that oxygen activity in steel is greater than that limited by the Al-O equilibrium, as shown in Equation (1) a O > a −2/3 Al ·K −1/3 where K represents the equilibrium constant of the corresponding reaction, and a represents the activity of the corresponding component (the same as below). The formation condition of MgO·Al 2 O 3 inclusions is that the actual oxygen activity in steel is greater than that limited by the Mg-Al-O equilibrium, as shown in Equation (2) a O > a −1/2 Al ·a −1/4 Mg ·K −1/4 The formation conditions of CeAlO 3 inclusions is that the actual oxygen activity in steel is greater than that limited by Ce-Al-O equilibrium, as shown in Equation (3) a O > a −1/3 Al ·a −1/3 Ce ·K −1/3 The formation conditions of Ce 2 O 2 S inclusions is that the actual oxygen activity in steel is greater than that limited by Ce-O-S equilibrium, as shown in Equation (4) a O > a −1 The transformation reaction between the inclusions can be obtained by associating the reactions in Table 2, as follows.
Conditions for MgO·Al 2 O 3 transformation into CeAlO 3 : Let ∆G < 0, at 1873 K: Conditions for MgO·Al 2 O 3 transformation into Ce 2 O 2 S: Let ∆G < 0, at 1873 K: According to the above analysis, the inclusion formation conditions can be obtained as follows.
(1) The formation condition of MgO·Al 2 O 3 in steel at 1873 K after magnesium treatment needs to be considered from two different formation mechanisms, as above mentioned. One is that Al 2 O 3 can transform into MgO·Al 2 O 3 (as shown in Equation (6)), which will provide a limit on magnesium activity. The other one is that MgO·Al 2 O 3 can directly form from steel (as shown in Equation (2)), which requires the oxygen activity to be higher than the equilibrium oxygen activity limited by Mg-Al-O, as follows. (2) There are also two considerations for the formation condition of single CeAlO 3 in steel at 1873 K. One is that MgO·Al 2 O 3 can transform into stable CeAlO 3 (as shown in Equation (8)), which means CeAlO 3 cannot spontaneously transform into Ce 2 O 2 S (as shown in Equation (10)); this will provide both the upper and lower limits of Ce activity. The other consideration is that CeAlO 3 can directly form from steel (as shown in Equation (3)), but Ce 2 O 2 S should not spontaneously precipitate (as shown in Equation (4)). This will also limit the upper and lower limits of oxygen activity, which means the oxygen activity should be higher than the equilibrium oxygen activity limited by Ce-Al-O but lower than the equilibrium oxygen activity limited by Ce-O-S, as follows.
(3) There are also two considerations for the formation condition of CeAlO 3 and Ce 2 O 2 S in steel at 1873 K. One is that MgO·Al 2 O 3 can transform into stable CeAlO 3 (as shown in Equation (8)), and CeAlO 3 can transform into Ce 2 O 2 S (as shown in Equation (10)); this will provide both the lower limits of Ce activity. The other consideration is that CeAlO 3 and Ce 2 O 2 S can directly form from steel (as shown in Equations (3) and (4)). This will also limit the lower limits of oxygen activity, which means the oxygen activity should be higher than the equilibrium oxygen activity limited by Ce-Al-O and Ce-O-S, as follows.
(4) There are also two considerations for the formation condition of single Ce 2 O 2 S in steel at 1873 K. One is that MgO·Al 2 O 3 can transform into stable Ce 2 O 2 S (as shown in Equation (10)), which means CeAlO 3 cannot spontaneously transform into CeAlO 3 (as shown in Equation (8)); this will provide both the upper and lower limits of Ce activity. The other consideration is that Ce 2 O 2 S can directly form from steel (as shown in Equation (4)), but CeAlO 3 should not spontaneously precipitate (as shown in Equation (3)). This will also limit the upper and lower limits of oxygen activity, which means the oxygen activity should be higher than the equilibrium oxygen activity limited by Ce-O-S but lower than the equilibrium oxygen activity limited by Ce-Al-O, as follows.
In summary, the modification mechanism of Ce to spinel inclusions can be shown in Figure 6. Ce can modify the spinel inclusions into CeAlO 3 or Ce 2 O 2 S. There are three main modification paths: 1 Spinel can directly transform into CeAlO 3 by Ce, namely spinel → Ce-Al-O + Mg-Al-O → CeAlO 3 , which can be proved by the complex inclusion shown in Figure 3a. 2 After Ce treatment, spinel can be preferentially transformed to CeAlO 3 . After a while, the resulting CeAlO 3 product layer will be transformed to Ce 2 O 2 S. The first-step reaction was considered to be difficult to perform completely because we observed some inclusions with a three-layer structure in the experiment, as shown in Huang et al. [7,9] and Wang et al. [8] reported that the sequence of Ce-modified spinel is spinel → CeAlO3 → Ce2O3 → Ce2O2S or spinel → Ce2O3 → Ce2O2S. However, no Ce2O3 inclusions were found in the present experimental steel. After comparing the experimental conditions, it was found that the contents of Al and S in the steel were different. Therefore, the stability diagram in the Fe-10%Mn-0.001%Mg-0.015%Ce-0.005%O-Al-S system at 1873 K was calculated using FactSage 7.1 (databases FactPS, FToxid, FSstel, MYDT), as shown in Figure 7. As can be seen from the red line in Figure 7, as the S content in the steel increased, the precipitation condition for the Ce2O3 required a lower Al content. The S content in our experimental steels was as high as 0.015%. Therefore, the formation condition of Ce2O3 was not reached when the Al content was higher than 0.02%.

Ce Content Required for Complete Modification of Spinel Inclusions
There are complex interactions between different elements, such as Ce, Al, O, and S, in the steel during the modification process. Under present experimental conditions, spinel is preferentially transformed to CeAlO3. In order to obtain the minimum Ce content required for complete modification of spinel inclusions, the effect of aluminum and oxygen content on the related Ce content in steel was calculated by FactSage. The content of each element in the steel discussed in this section is total content, and the calculation results are shown in Figure 8. Huang et al. [7,9] and Wang et al. [8] reported that the sequence of Ce-modified spinel is spinel → CeAlO 3 → Ce 2 O 3 → Ce 2 O 2 S or spinel → Ce 2 O 3 → Ce 2 O 2 S. However, no Ce 2 O 3 inclusions were found in the present experimental steel. After comparing the experimental conditions, it was found that the contents of Al and S in the steel were different. Therefore, the stability diagram in the Fe-10%Mn-0.001%Mg-0.015%Ce-0.005%O-Al-S system at 1873 K was calculated using FactSage 7.1 (databases FactPS, FToxid, FSstel, MYDT), as shown in Figure 7. As can be seen from the red line in Figure 7, as the S content in the steel increased, the precipitation condition for the Ce 2 O 3 required a lower Al content. The S content in our experimental steels was as high as 0.015%. Therefore, the formation condition of Ce 2 O 3 was not reached when the Al content was higher than 0.02%. Huang et al. [7,9] and Wang et al. [8] reported that the sequence of Ce-modified spinel is spinel → CeAlO3 → Ce2O3 → Ce2O2S or spinel → Ce2O3 → Ce2O2S. However, no Ce2O3 inclusions were found in the present experimental steel. After comparing the experimental conditions, it was found that the contents of Al and S in the steel were different. Therefore, the stability diagram in the Fe-10%Mn-0.001%Mg-0.015%Ce-0.005%O-Al-S system at 1873 K was calculated using FactSage 7.1 (databases FactPS, FToxid, FSstel, MYDT), as shown in Figure 7. As can be seen from the red line in Figure 7, as the S content in the steel increased, the precipitation condition for the Ce2O3 required a lower Al content. The S content in our experimental steels was as high as 0.015%. Therefore, the formation condition of Ce2O3 was not reached when the Al content was higher than 0.02%.

Ce Content Required for Complete Modification of Spinel Inclusions
There are complex interactions between different elements, such as Ce, Al, O, and S, in the steel during the modification process. Under present experimental conditions, spinel is preferentially transformed to CeAlO3. In order to obtain the minimum Ce content required for complete modification of spinel inclusions, the effect of aluminum and oxygen content on the related Ce content in steel was calculated by FactSage. The content of each element in the steel discussed in this section is total content, and the calculation results are shown in Figure 8.

Ce Content Required for Complete Modification of Spinel Inclusions
There are complex interactions between different elements, such as Ce, Al, O, and S, in the steel during the modification process. Under present experimental conditions, spinel is preferentially transformed to CeAlO 3 . In order to obtain the minimum Ce content required for complete modification of spinel inclusions, the effect of aluminum and oxygen content on the related Ce content in steel was calculated by FactSage. The content of each element in the steel discussed in this section is total content, and the calculation results are shown in   Figure 8a shows the effect of oxygen content on inclusions in steel, where w Al = 0.02% and w S = 0.015%. It can be seen from the red line in the figure that the minimum Ce content required for modification of spinel inclusions is linear with the oxygen content in the steel. When w Ce = 2.99·w O , the inclusions in the steel can be controlled as CeAlO 3 . Wang et al. [8] also obtained a similar conclusion: that increasing the O content would increase the Ce content required for complete modification of spinel inclusions.
Theoretically, the inclusions can be controlled to be CeAlO 3 when the Ce content in the steel is three times that of the oxygen content. Therefore, by measuring the oxygen content in the steel, a suitable rare earth Ce addition amount required for complete modification of spinel inclusions can be obtained. Figure 8b shows the effect of aluminum content on types of stable inclusions in steel, where w O = 0.0045% and w S = 0.015%. It can be seen from the red line in the figure that as the aluminum content in the steel increases, the minimum Ce content required for modification spinel inclusions does not change much, which indicates that the aluminum content has little effect on the minimum Ce content required for modification of spinel inclusions.

Conclusions
In present paper, the transformation process of inclusions in medium manganese steel was studied by high-temperature simulation experiments. The modification mechanism of inclusions in steel under different aluminum contents was discussed by thermodynamic calculations. Under the present experimental conditions, the following conclusions were obtained.
(1) Ce treatment can modify spinel inclusions into rare earth inclusions. After Ce treatment, the amount of inclusions in the steel will decrease, and the size of inclusions is basically less than 5 µm.
(2) The modification mechanism of spinel inclusions by Ce is as follows: After Ce is added to steel, Mn and Mg in the spinel inclusions are first replaced by Ce to form CeAlO 3 , which will cause the destruction of the spinel structure. When the O content in the steel is at a lower level, S in the steel will take part in the replacement reaction to replace the O in the inclusion, then the CeAlO 3 and spinel inclusions will be finally transformed into Ce 2 O 2 S.