Inﬂuence of Slag and Refractory Materials on Inclusions during the Ladle Reﬁning of Low Carbon Aluminum Killed Steel

: The evolution of inclusions in low carbon Al killed steel during ladle reﬁning of was studied based on industrial experiments, in which high basicity slag was used. The results showed that inclusions experienced the changes from Al 2 O 3 → MgO-Al 2 O 3 → CaO-MgO-Al 2 O 3 → CaO-Al 2 O 3 . Without calcium treatment, MgO-Al 2 O 3 inclusion in steel were largely transformed into CaO-MgO-Al 2 O 3 or CaO-Al 2 O 3 . With the aim to decrease MgO-Al 2 O 3 inclusions and to clarify the effects of reﬁning slag and refractory materials on inclusions, laboratory experiments were performed with lower basicity reﬁning slag (lower basicity slag theoretically helps reduce spinel-type inclusions) in MgO and Al 2 O 3 crucibles. The results indicated that, the dissolved Al in liquid steel would react with MgO and CaO in slag or in refractory at 1600 ◦ C. Hence, [Mg] and [Ca] would be supplied into bulk steel. Due to the large contact area between MgO-based refractory and steel, as well as the higher activity of MgO in the refractory, Mg can be more easily reduced, which accounts for the easy modiﬁcation of Al 2 O 3 into MgO-Al 2 O 3 . By contrast, because of the limited supply of [Ca] to steel, modiﬁcation of MgO-Al 2 O 3 into CaO-MgO-Al 2 O 3 or CaO-Al 2 O 3 was incomplete. With the use of Al 2 O 3 -based refractory and reeﬁng slag basicity of about 2.45, MgO-Al 2 O 3 inclusions were obviously decreased.


Introduction
Low carbon aluminum killed steel (LCAK) with good toughness, ductility, stamping, surface and aging resistance is mainly used to manufacture automobiles, household appliances, etc. [1][2][3][4]. To ensure the properties, size, quantity and shape of inclusions in steel should be strictly controlled, lower contents of O tot (total oxygen) and N are necessary [5][6][7].
During continuous casting, LCAK usually has poor castability due to the easy occurrence of clogging problems, which are often initiated by solid inclusions such as Al 2 O 3 and MgO-Al 2 O 3 in steel. To improve castability of liquid steel, calcium treatment is often adopted during ladle furnace (LF) refining by modifying such solids into semi-liquid or liquid states [8][9][10][11][12][13][14][15][16]. However, the yield ratio of Ca is usually unstable because of its high vaporization pressure at high temperatures [17][18][19][20]. Moreover, the application of calcium treatment means higher costs and longer production times. Furthermore, the addition of Ca into liquid steel often causes environmental problems such as smoke in production sites.
Hence, improving the castability of LCAK without the use of calcium treatment is very significant for its industrial production. However, the challenge is: how to modify the solid inclusions into liquid states without calcium treatment.
Holappa [17] et al. found that contents of O tot and [S] in steel significantly influence modifications of inclusions in calcium treatment. When [S] content in steel is constant, the lower the content of O tot in molten steel, the narrower the "window" of [Ca] content required for the formation of liquid inclusion. When the content of O tot is constant, with the 2 of 10 rise of [S] content in steel, the narrower the "window" of [Ca] required for the formation of liquid inclusions. Pretorius [18] et al. found that calcium treatment could improve the castability of steel, but it would be hard to fully liquify MgO-Al 2 O 3 at a low melting point. During calcium treatment, calcium would preferentially react with magnesium oxide in the inclusions, but the free magnesium produced after the reaction would produce new MgO-Al 2 O 3 inclusions with aluminum oxide in molten steel. During casting, the re-oxidation of steel causes large formations of solid calcium-aluminates [19,20]. Park proposed that refining the slag of CaO-A1 2 O 3 -MgO system with a mass ratio of CaO/SiO 2 over 6.0 helps to modify inclusions in stainless steel from solid to liquid states [21].
Most studies on targeting liquid inclusions in Al steel mainly focused on calcium treatment. Additionally, CaO-A1 2 O 3 -MgO system refining slag is seldom used in LCAK. Hence, enhancing castability of LCAK without the application of calcium treatment is needed for industrial production. Considering that, the present study was carried out.

Industrial Test
LCAK steel (SPHC) was produced by "BOF → LF refining → CC (the section sizes of the slab were about 1400 mm × 240 mm)". Operations in the refining process included: (1) The carried slag from the tapping was removed from the ladle. MgO-C refractory was used at the slag line of the ladle, while MgO-Al 2 O 3 refractory bricks (MgO content was about 10%) were chosen as the lining for other parts of the ladle. (2) Aluminum, ferromanganese alloy and lime were added into liquid steel for deoxidation, alloying and slag-making during BOF tapping. (3) Liquid steel was heated in LF refining, and aluminum and lime were also added for further deoxidation and slag-making. During LF refining, the basicity of ladle slag w (CaO)/w (SiO 2 ) was ≥8, while the mass ratio of w (CaO) to w (Al 2 O 3 ) was about 1.0-2.0. (4) After desulphurization in LF refining, FeMn alloy was added to adjust [Mn] in liquid steel, and the melt was softly blown for 6 min by argon gas with a flow rate of 600-800 L/min before LF departure. (5) Basicity of tundish flux was ≥2.0 and mass ratio of w(CaO) to w(Al 2 O 3 ) was about 1.0-1.5.
During the test, steel and slag were sampled at the stage of LF arrival, before slagmaking, after slag-making, before soft blowing, LF departure and also sampled in casting tundish (about 100 tons of liquid steel was poured). The samples obtained were labelled as LF1, LF2, LF3, LF4, LF5 and CC1, respectively, as shown in Figure 1. Species were cut from the samples for the analysis of total oxygen (O tot ) and inclusion inspection.

Laboratory Study
In order to clarify the effects of refining slag and refractory on inclusions, labor experiments were carried out. During the experiment, 500 g steel was put into the cru in a vacuum induction furnace. After vacuum extraction, purified argon gas was pu into the furnace. The furnace was then heated to 1600 °C and steel was melted. Alum

Laboratory Study
In order to clarify the effects of refining slag and refractory on inclusions, laboratory experiments were carried out. During the experiment, 500 g steel was put into the crucible Metals 2023, 13, 866 3 of 10 in a vacuum induction furnace. After vacuum extraction, purified argon gas was pumped into the furnace. The furnace was then heated to 1600 • C and steel was melted. Aluminum and 50 g slag were added into the molten steel. When the melt has been held for about 25 min, the liquid steel was tapped into a prepared ingot set in the vacuum chamber of the furnace. In order to clarify the influence of refractory on inclusions, two types of crucibles, viz. MgO (w(MgO) > 97.5%) and corundum (w(Al 2 O 3 ) > 99%), were used in the experiments.

Analysis Method
Contents of O tot and N in steel samples were analyzed. Inclusions ≥1µm were inspected by the automatic SEM-EDS machine of ASPEX explorer, and the scanning area was about 25 mm 2 for each sample. According to the EDS results, Ca, Si, Al, Mn, Mg, S and O inclusions were converted into the mass percentage of the corresponding oxides and sulfides. In this paper, the attention was mainly on oxide inclusions.

Composition of Industrial Samples
Chemical compositions of slag samples were shown in Table 1. As shown in Table 1, mass ratios of CaO/SiO 2 (C/S: usually named basicity) in refining slag were in the range of 13.21-29.65, and mass ratios of CaO/Al 2 O 3 (C/A) in slag were between 1.09 and 1.92. The basicity of the tundish flux was about 2.83, and the ratio of C/A was about 1.24. Compositions of liquid steel at different stages were shown in Table 2. As can be seen, contents of Mg and Ca in steel increased with the rise of refining time, indicating a similar tendency reported by Deng et al. [21]. Mg contents were in the range of 4-7 ppm. Ca contents in steel were in the range of 3-7 ppm during LF refining, which was attributed to the reduction of CaO in slag by slag-steel chemical reactions in the refining.  Figure 2 shows the changes of cleanliness in liquid steel. It can be seen from Figure 2 that the O tot and N contents in the steel were 0.0034% and 0.0023% at LF arrival, respectively. With the proceeding of refining, O tot decreased continuously while N contents increased. The lowest O tot content was 0.00163%, and the N content was 0.0029% before soft blowing. At LF departure, O tot and N in liquid steel were 0.0020% and 0.0033%, respectively, with an Metals 2023, 13, 866 4 of 10 increase of 0.0004% and 0.00042%, respectively. Compared with that before soft blowing, the pick-ups of O tot and N indicated the occurrence of re-oxidation of molten steel during soft blowing. In tundish, O tot and N were 0.0021% and 0.0033%, respectively. Afterwards, contents of O tot and N increased by 0.00015% and 0.00005%, respectively. Despite that, O tot contents in steel were controlled in the range of 0.0016-0.0022% in continuous casting.

Changes of O tot and N Contents in Steel
tively. With the proceeding of refining, Otot decreased continuously while N conte creased. The lowest Otot content was 0.00163%, and the N content was 0.0029% befo blowing. At LF departure, Otot and N in liquid steel were 0.0020% and 0.0033%, r tively, with an increase of 0.0004% and 0.00042%, respectively. Compared with that soft blowing, the pick-ups of Otot and N indicated the occurrence of re-oxidation of m steel during soft blowing. In tundish, Otot and N were 0.0021% and 0.0033%, respec Afterwards, contents of Otot and N increased by 0.00015% and 0.00005%, respective spite that, Otot contents in steel were controlled in the range of 0.0016%-0.0022% in c uous casting.

Types of Inclusions
Observed oxide inclusions can be categorized into four types: Al2O3 (with ≥95%), MgO-Al2O3 (with CaO <5%), CaO-MgO-Al2O3 (with CaO ≥5%, MgO ≥5%) and Al2O3 (with MgO <5%). The number fractions of various inclusions at different tim shown in Figures 3 and 4. As can be seen, 92% of the inclusions were Al2O3 at LF a which were either in blocky or cluster shapes. Afterwards, the proportion of Al2 creased significantly to about 53%, and the fraction of MgO-Al2O3 inclusions inc rapidly to 34%, which were mainly in irregular spheres. While the fractions of CaO Al2O3 and CaO-Al2O3 were below 8%, both of which were mostly in spherical sha shown in Figure 4c

Compositions of Inclusions
Compositions of inclusions at different stages were projected into Al2O3-MgO-CaO ternary phase diagram to show the evolution route of inclusions in Figure 5. Figure 5a

Compositions of Inclusions
Compositions of inclusions at different stages were projected into Al 2 O 3 -MgO-CaO ternary phase diagram to show the evolution route of inclusions in Figure 5. Figure 5a Figure 6 shows variations in the quantity and size of inclusions. It can be seen that at LF arrival, the number density of inclusions was about 13/mm 2 , with an average size of about 2.8 µm. Afterwards, the number density of inclusions reached a maximum of 17/mm 2 , which could be related to more intensive argon gas bottom bubbling of ladle to promote desulfurization. Then, the number density of inclusions decreased sharply to 8/mm 2 . However, the average size of inclusions slightly increased to 3.4 µm. At LF departure, number density of inclusions was 11/mm 2 , about 2/mm 2 higher than that before soft blowing. Additionally, the average size of inclusions was 2.9 µm, which was a little lower than that before soft blowing, indicating that soft blowing was favorable to the floatation of large inclusions. In the tundish, the number density of inclusions was about 10/mm 2 , which was slightly lower than that of LF departure. While the average size of inclusions was 2.6 µm, about 0.3 µm smaller than that at LF departure. stage of LF departure, as shown in Figure 5e. In casting tundish (Figure 5f), inclusions were mainly solid MgO-Al2O3 and CaO-Al2O3 inclusions, with sizes smaller than 10 µm.  Figure 6 shows variations in the quantity and size of inclusions. It can be seen that at LF arrival, the number density of inclusions was about 13 /mm 2 , with an average size of about 2.8 µm. Afterwards, the number density of inclusions reached a maximum of 17 /mm 2 , which could be related to more intensive argon gas bottom bubbling of ladle to promote desulfurization. Then, the number density of inclusions decreased sharply to 8 /mm 2 . However, the average size of inclusions slightly increased to 3.4 µm. At LF departure, number density of inclusions was 11 /mm 2 , about 2/mm 2 higher than that before soft blowing. Additionally, the average size of inclusions was 2.9 µm, which was a little lower than that before soft blowing, indicating that soft blowing was favorable to the floatation of large inclusions. In the tundish, the number density of inclusions was about 10 /mm 2 , which was slightly lower than that of LF departure. While the average size of inclusions was 2.6 µm, about 0.3 µm smaller than that at LF departure.

Formation of MgO-Al2O3 Inclusions
According to previous studies, during the basic slag refining of Al-deoxidize reaction (1)
The laboratory experiments conducted in this study clarified the influences of refining slab and choices of crucible materials (refractory) on the change of Al 2 O 3 into MgO-Al 2 O 3 .
The laboratory experiments conducted in this study clarified the influences of ing slab and choices of crucible materials (refractory) on the change of Al2O3 into Al2O3. Figure 7 shows the compositions of inclusions in steel, without refining sla with basicity of slag about 2.17 and 3.34, respectively, in which MgO crucibles were The inclusions were mainly MgO-Al2O3 or Al2O3-MgO-CaO inclusions with a amount of CaO. Particularly, MgO-Al2O3 existed in steel even without refining slag MgO crucible was used.  Figure 8 shows the compositions of inclusions in steel, without refining slag and with the basicity of slag of about 4.51 and 8.24, respectively, when Al 2 O 3 crucible steel was used. As can be seen, MgO-Al 2 O 3 inclusions were well decreased, and many inclusions were composed of CaO-MgO-Al 2 O 3 with MgO contents about or within 10% when the slag contained about 5% MgO and with a basicity of about 4.5. As is known, such complex inclusions usually have lower melting points with an outer surface layer of CaO-Al 2 O 3 , which would be more desirable to enhance the castability of liquid steel. By contrast, with a slag basicity increased to 8.24, although some liquid CaO-Al2O3 were formed, most inclusions were pure MgO-Al 2 O 3 inclusions or Al 2 O 3 -MgO-based ones (with CaO <5%), which were bad to the castability of molten steel.
The results clearly showed that transformation of inclusions closely related to refractory (the choice of crucible) and slag basicity. Because of larger contact area between ladle refractory and liquid steel than that between slag and liquid steel, and higher MgO activity (about 1) in the MgO-based refractory, the reaction between MgO in refractory and [Al] would be faster. It is known that MgO-Al 2 O 3 inclusions can be easily formed when MgO-crucible are used. sions usually have lower melting points with an outer surface layer of CaO-Al2O3, would be more desirable to enhance the castability of liquid steel. By contrast, with basicity increased to 8.24, although some liquid CaO-Al2O3 were formed, most incl were pure MgO-Al2O3 inclusions or Al2O3-MgO-based ones (with CaO <5%), which bad to the castability of molten steel. The results clearly showed that transformation of inclusions closely related to tory (the choice of crucible) and slag basicity. Because of larger contact area between refractory and liquid steel than that between slag and liquid steel, and higher MgO ity (about 1) in the MgO-based refractory, the reaction between MgO in refractor [Al] would be faster. It is known that MgO-Al2O3 inclusions can be easily formed MgO-crucible are used.

Formation Mechanism of CaO-MgO-Al2O3 or CaO-Al2O3 Inclusions
During the refining, although Ca-treatment was not used, reaction (4) could oc supply Ca into the liquid steel [27,[29][30][31][32]. Then, [Ca] would in turn react with MgO and Al2O3 in steel to modify them into CaO-MgO-Al2O3 or CaO-Al2O3, as express reactions (5) and (6)  [Ca] + n/3(Al2O3)inclusion = (CaO·(n − 1)/3Al2O3)inclusion + 2/3 [Al] With regard to the modification of MgO-Al2O3 into CaO-MgO-Al2O3 or CaOprevious studies showed that the transformation of MgO-Al2O3 would proceed fro surface to the inner center, in which the rate-controlled step was the diffusion of C Mg in the intermediate reaction layer [27]. In this study, variations of CaO, MgO and contents in inclusions were observed, their sizes shown in Figure 9. As can be seen, CaO contents and lower MgO contents were observed in small inclusions. By co lower CaO contents and higher MgO contents were observed in large inclusions. plied that the modification of larger MgO-Al2O3 inclusions needed a much longer which was justifiable and consistent with previous findings.

Formation Mechanism of CaO-MgO-Al 2 O 3 or CaO-Al 2 O 3 Inclusions
During the refining, although Ca-treatment was not used, reaction (4) could occur to supply Ca into the liquid steel [27,[29][30][31][32] (5) and (6) With regard to the modification of MgO-Al 2 O 3 into CaO-MgO-Al 2 O 3 or CaO-Al 2 O 3 , previous studies showed that the transformation of MgO-Al 2 O 3 would proceed from the surface to the inner center, in which the rate-controlled step was the diffusion of Ca and Mg in the intermediate reaction layer [27]. In this study, variations of CaO, MgO and Al 2 O 3 contents in inclusions were observed, their sizes shown in Figure 9. As can be seen, higher CaO contents and lower MgO contents were observed in small inclusions. By contrast, lower CaO contents and higher MgO contents were observed in large inclusions. It implied that the modification of larger MgO-Al 2 O 3 inclusions needed a much longer time, which was justifiable and consistent with previous findings.

Conclusions
This study attempted to improve the castability of LCAK by modifying solid inclusions into semi-liquid or liquid state during LF refining. It was found that evolutions of inclusions from alumina to MgO-Al2O3 and then to CaO-MgO-Al2O3 were hard to avoid.

Conclusions
This study attempted to improve the castability of LCAK by modifying solid inclusions into semi-liquid or liquid state during LF refining. It was found that evolutions of inclusions from alumina to MgO-Al 2 O 3 and then to CaO-MgO-Al 2 O 3 were hard to avoid. Evolution mechanisms of inclusions were clarified, and a possible way to reduced MgO-Al 2 O 3 -type inclusions was proposed based on laboratory experiments, in which different slags and refractories (crucibles) were used. The obtained results can be briefly concluded as follows: