Effect of Tundish Flux on Compositional Changes in Non-metallic Inclusions in Stainless Steel Melts

Abstract

The effect of the tundish flux on the evolution of non-metallic inclusions in Si-killed 304 (18%Cr-8%Ni) stainless steel has been investigated at 1773 K. The interfacial reaction between molten steel and the CaO–Al₂O₃–MgO flux causes the aluminum pick-up from the liquid slag into the steel melt, resulting in a decrease in the oxygen content in the steel. The aluminum originating from the slag modifies the pre-existing Mn-silicate inclusions into alumina-rich inclusions in the steel. Because the oxygen content in the steel decreases as it reacts with the CaO–Al₂O₃–MgO flux, the degree of supersaturation for alumina formation is too low to precipitate new-born alumina particles in the steel. By analyzing the population density function (PDF) results for inclusions, it can be observed that the growth of spinel-type inclusions occurs by the diffusion of aluminum and magnesium in the steel. On the other hand, the composition of the steel, as well as the evolution of inclusions, is negligibly changed when the CaO–SiO₂–MgO flux is added to the molten steel. Furthermore, the computational simulation for predicting the evolution of inclusions in molten steel during a continuous casting tundish process was carried out based on a refractory-slag-metal-inclusion (ReSMI) multiphase reaction model.

1. Introduction

It has been recognized that the formation of fine non-metallic inclusions (NMI) has a beneficial effect on forming equiaxed grains during the casting of ferritic stainless steel because the inclusions act as heterogeneous nucleation sites of delta ferrite.^1,2,3,4) On the other hand, it is well known that NMI with a high melting point and strength, such as alumina and MgAl₂O₄ spinel, cause nozzle clogging in a submersed entry nozzle and surface defects in steel products. Many researchers have investigated the evolution of NMI in stainless steels during various steelmaking processes.^{5,6,7,8,9,10,11,12,13,14,15,16,17)}

It is necessary to investigate the mechanism for the formation of inclusions in a continuous casting tundish process to improve the cleanliness of the steel products because the tundish is the last metallurgical reactor used for removing NMI before casting molten steel. Because the exposure of molten steel to air in the tundish is less significant after the tundish fluxes cover the steel at the stable teeming stage, many researchers have focused on the reoxidation behavior of the molten steel during the interfacial reaction between the tundish fluxes and steel.^18,19,20) It has been reported that the thermodynamic equilibrium between Si-killed steel and the CaO–Al₂O₃–SiO₂–MgO slags increases the Al₂O₃ content in the composition of inclusions.^21,22) However, the mechanism for the formation of alumina and spinel inclusions in Si-killed 304 (18%Cr-8%Ni) stainless steel by the slag/metal reaction in the tundish process has yet to be fully understood.

The particle size distribution of inclusions has been studied to understand the nucleation, growth, and removal mechanism of NMI after deoxidation in the steel.^{23,24,25,26,27,28)} Zhang and Lee²³⁾ demonstrated that oxygen diffusion and Ostwald ripening essentially determine the growth of alumina particles at the initial stage of deoxidation, while collision mechanisms caused by Brownian coagulation, turbulent coagulation, Stokes coagulation, and gradient coagulation significantly affect the growth of the particles with increasing time after nucleation. Because the population density function (PDF) eliminates the arbitrariness originating from users, PDF of inclusions has been applied to further understand the growth mechanism of NMI in the molten steel after deoxidation or reoxidation in laboratory-scale experiments and plant-scale steelmaking processes.^{24,25,26,27,28)}

Various kinetic models have been widely developed to understand the chemical reactions between metal and slag including primary steelmaking, secondary refining, and a continuous casting tundish and mold.^{29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44)} In particular, the refractory-slag-metal-inclusion (ReSMI) multiphase reaction model, which was developed from an effective equilibrium reaction zone (EERZ) model,^31,37,41) is easily applicable to various metallurgical processes for predicting not only the compositional changes of the metal and steel but also the evolution of inclusions in the steel with reaction time; this model is useful because it applies the effective reaction volumes in the metal and slag phases and the dissolution of MgO from the refractory to the slag phase.^39,40,43) Recently, the reoxidation behavior of molten steel and the evolution of NMI in the tundish during a continuous casting process have been studied by using the ReSMI multiphase reaction model.⁴⁴⁾

In the present study, the mechanism of the formation of NMI in 304 stainless steel with tundish fluxes based on Ca-silicate and Ca-aluminate systems has been investigated in conjunction with the application of a computational simulation for an industrial-scale tundish process based on the ReSMI multiphase reaction model.

2. Experimental Procedure

2.1. Experimental Details

Reagent grade SiO₂ and Al₂O₃ were used in the present experiments. CaO and MgO were obtained via calcination of reagent grade CaCO₃ and MgCO₃ at 1273 K for 10 h. The powders were weighed and mixed to obtain the synthetic slag listed in Table 1. The mixtures were melted in a graphite crucible at 1873 K for 1 h and quenched on a copper plate cooled by water. The grinded synthetic slag was heated up to 1273 K in a box furnace to remove dissolved carbon in the slag.

Table 1. Initial composition of tundish fluxes used in the present study (wt%).

Type	CaO	SiO2	Al2O3	MgO	C/S or C/A
CS-flux	47.5	47.5	–	5.0	1.0
CA-flux	53.5	–	41.5	5.0	1.3

A schematic diagram of the present experimental apparatus is shown in Fig. 1. 600 g of 304 stainless steel (18%Cr-8%Ni-1%Mn-0.4%Si) was contained in a magnesia crucible (99.9%, 60 mm [outer diameter, OD] × 50 mm [inner diameter, ID] × 120 mm [height, H]). The crucible was set in a high-frequency induction furnace inside a quartz reaction chamber. After the reaction chamber was placed under vacuum using a mechanical rotary pump, an Ar-3%H₂ gas mixture was injected into the chamber. Then, the steel in the magnesia crucible was heated up to 1773 K using a B-type thermocouple and a proportional integral differential controller.

Fig. 1.

A schematic diagram for the experimental apparatus. (Online version in color.)

The metal sample was obtained using a quartz sampler (OD: 6 mm, ID: 4 mm) before adding synthetic slag on the molten steel at 1773 K. 50 g of synthetic slag was added on the molten steel as soon as the metal was sampled through a quartz tube (14 mm [OD] × 12 mm [ID] × 500 mm [H]). After the addition of the synthetic slag, metal samples were obtained at 5, 10, 15, 30 min by the quartz suction sampler. The total oxygen amounts in the steel samples were measured by a combustion analyzer (TC-300, LECO Co.). The amounts of Si, Al, Mn, Ni, and Cr in the steel samples were analyzed using inductively coupled plasma-atomic emission spectroscopy (ICP-AES, ACROS II, Spectro).

2.2. Characterization of Inclusions using Automatic Feature Analysis (AFA) System and Electrolytic Extraction (EE) Method

The automatic feature analysis (AFA) technique has been widely utilized to quantitatively characterize inclusions in polished steel samples. The cross section of the steel sample (diameter: 4 mm) was polished for AFA. The AFA technique in steel has been carried out using field-emission scanning electron microscopy (FE-SEM, JSM-6980LV, JEOL) coupled with an energy-dispersive spectrometer (EDS) equipped with INCA Feature (Oxford Instruments) to measure the composition and size of inclusions. The accelerating voltage was 20 kV, the magnification was ×500, the pixel size was 0.5 μm, and the inclusions were detected with a 1 μm limit diameter.

The electrolytic extraction (EE) method with a 10% AA (10% acetyleacetone – 1% tetramethylammonium chloride – methanol) solution was carried out to separate inclusions from the steel matrix.^45,46,47) For the extraction, a current of 500 mA was applied for 3 h. The solution containing inclusions was filtered through a polycarbonate membrane film filter with pore size of 0.1 um. Then, the inclusions on the filter were observed using FE-SEM (MIRA 3, TESCAN) with EDS at an operating voltage of 15 kV after coating the filter with Pt at room temperature.

2.3. Inclusion Population Density Function (PDF)

Inclusion data obtained from the AFA technique is limited in stereological analysis because data is acquired from a two-dimensional (2D) section. Many researchers have adopted the population density function (PDF) to investigate the size distribution of inclusions in the steel samples because the PDF eliminates any arbitrariness caused by the user.^{23,24,25,26,27,28)} The PDF, with a length⁻⁴ unit, is defined by the following equation:   

$$\text{PDF}=\frac{n_v\left(L_{XY}\right)}{\left(L_Y-L_X\right)}$$(1)

where, n_v(L_XY) is the frequency of inclusions in a given size bin per volume and (L_Y−L_X) is the bin width. The CSD correction program (ver. 1.5), which was developed to convert 2D data into 3D data, was utilized to transform the AFA inclusion data into a PDF size distribution. Raw data for the area, length, and width of each inclusion are given as the inputs to the program. The overall shape of the particles should be estimated using the short dimension, intermediate dimension, long dimension, and the roundness degree of the particles. The program calculates the PDF size distribution based on the input data.

3. Results and Discussion

3.1. Compositional Changes in 304 Stainless Steel Melt

The compositional changes in the stainless steel melt as a function of reaction time with the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes are shown in Fig. 2. The contents of chromium, nickel, manganese, and silicon in the steel are negligibly changed with reaction time. Otherwise, the aluminum content in the steel continuously increases up to around 150 ppm when the CaO–Al₂O₃–MgO flux is added to the molten steel. Aluminum pick-up occurs because the dissolved silicon in the molten steel reduces Al₂O₃ in the CaO–Al₂O₃–MgO slag by a chemical reaction between the molten steel and slag at 1773 K:   

$$2{\left({\text{Al}}_2{\text{O}}_3\right)}_{\text{slag}}+3\left[\text{Si}\right]=3{\left({\text{SiO}}_2\right)}_{\text{slag}}+4\left[\text{Al}\right]$$(2)

Fig. 2.

Compositional changes of (a) [Cr], [Ni], [Mn]; (b) [Si]; and (c) [Al] in the steel with the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes as a function of reaction time. (Online version in color.)

Roy et al.³⁴⁾ presented experimental and simulated results for Al pick-up in the Al-killed steel ([Al] = 0.03 to 0.05 wt%) based on the chemical reaction between the silicon (0.8 to 1.9 wt%) in steel and Al₂O₃ in the CaO–Al₂O₃–SiO₂–MgO slags (C/A = 1.0 to 1.4, SiO₂ = 5.3 to 10.1 wt%, MgO = 6.8 to 11.1 wt%), resulting in an increase in the SiO₂ content in the slag. Because the aluminum content is under several ppm in the Si-killed stainless steel melt before slag addition, aluminum could be transferred from the slag by a chemical reaction with the CaO–Al₂O₃–MgO slag in the present study.

The total oxygen content in the steel melt is shown in Fig. 3. The initial oxygen content in the Si-killed stainless steel was around 50 ppm before the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes were added to the molten steel. The initial oxygen content in the steel is maintained with little deviation when the CaO–SiO₂–MgO flux is added to the molten steel. On the other hand, when the CaO–Al₂O₃–MgO flux is added, the oxygen content in the steel continuously decreases to around 20 ppm at 10 min, and then it becomes almost constant until 30 min.

Fig. 3.

Compositional changes of T.[O] in the steel with the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes as a function of reaction time. (Online version in color.)

The stability diagram, which was calculated using FactSage^TM7.3 with FToxid and FTmisc databases, is shown in Fig. 4. The solid lines of the stability diagram indicate the calculated soluble oxygen in the steel melts. Because the addition of the CaO–SiO₂–MgO flux has less of an effect on the composition of the steel, the data for the addition of the CaO–SiO₂–MgO flux is not indicated in Fig. 4. The calculated soluble oxygen in the steel is 80 to 100 ppm when it is assumed that the aluminum content in the steel is around 1 to 2 ppm before the CaO–Al₂O₃–MgO flux addition. As the aluminum content in the steel increases with reaction time, the soluble oxygen content continuously decreases to several ppm.

Fig. 4.

A stability predominance diagram for the Fe-18%Cr-8%Ni-1%Mn-Si-Al-O system at 1773 K (Calculated from FactSage^TM7.3).

Tanaka et al.^48,49) investigated a mechanism for the dynamic interfacial reaction between Al-killed steel and the CaO–SiO₂–Al₂O₃(–CaF₂) slags: 1) Silica decomposition and oxygen adsorption at the interface, 2) Diffusion of silicon from the interface to bulk steel with oxygen remaining in the interface layer, 3) Oxygen desorption from the interface to the bulk steel and reaction of oxygen with aluminum forming alumina at the interface, 4) Silica mass transfer from the bulk slag to the interface and reaction between oxygen and aluminum, and 5) Blockage of silica mass transfer due to the accumulation of the formed alumina at the interface and a decrease in free oxygen at the interface due to oxygen desorption and oxygen reaction with aluminum.

The dynamic interfacial phenomena in the present study would be opposite of the results of Tanaka et al. when the CaO–Al₂O₃–MgO flux is added to the Si-killed stainless steel melt. Aluminum would be accumulated at the steel sub-interface layer because alumina at the slag sub-interface layer is reduced by silicon in the steel, as shown in Eq. (2). The released aluminum would react with the oxygen at the interface layer, resulting in a decrease in the oxygen content at the interface. Desorption of aluminum from the interface to the bulk steel occurs simultaneously with the transfer of oxygen from the bulk steel to the interface. Therefore, the oxygen content in the steel decreases as the aluminum content increases with reaction time after the CaO–Al₂O₃–MgO flux is added to the steel. The relevant mechanism of the dynamic behavior at the slag-metal interface is schematically shown in Fig. 5 following Tanaka et al.’s conceptualization.^48,49)

Fig. 5.

Microscopic view of the mechanism of the dynamic behavior at the slag-metal interface under the chemical reactions.

It is of interest that the aluminum content continuously increases with reaction time while the oxygen content is almost constant after 10 min. Because the increase in the aluminum content in the bulk steel enhances the driving force for the formation of alumina inclusions (Fig. 4), the behavior of oxygen after 10 min is strongly related to the formation of non-metallic inclusions.

3.2. Evolution of Non-metallic Inclusions with Slag-Metal Reaction

The compositions of non-metallic inclusions reacted with the CaO–Al₂O₃–MgO flux are shown in Fig. 6 with reaction time. Because the composition of inclusions reacted with the CaO–SiO₂–MgO flux only change slightly from the initial composition, the case with the CaO–SiO₂–MgO flux is not illustrated in Fig. 6. Before adding the CaO–Al₂O₃–MgO flux to the steel, the composition of inclusions is mainly located in the liquid and SiO₂-rich Mn-silicate region in the MnO–SiO₂–Al₂O₃ ternary phase diagram. Because alumina inclusions are spontaneously precipitated in the steel as the aluminum content increases with reaction time (Fig. 4), the composition of inclusions moves from the MnO–SiO₂ binary region to the alumina-rich region in the phase diagram at 10 min. After 10 min, the MgO content in the inclusions continuously increases with reaction time; hence, spinel inclusions are predominant in the steel. The formation of spinel inclusions has been well established in Al-killed steel. The alumina inclusions react with magnesium that originated from the slag phase, as follows:^10,20,44)   

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

  

$${\left({\text{Al}}_2{\text{O}}_3\right)}_{\text{inclusion}}+\left[\text{Mg}\right]+\left[\text{O}\right]={\text{MgAl}}_2{\text{O}}_4\left(\text{s}\right)$$(4)

Fig. 6.

Compositional changes in non-metallic inclusions in the steel reacted with the CaO–Al₂O₃–MgO flux as the reaction time increases.

The stability diagram for Fe-18%Cr-8%Ni-1%Mn-Si-Al-O system with 1 ppm Mg at 1773 K is shown in Fig. 7. The points at 10, 15 and 30 min are located in the spinel phase region with low oxygen solubility value compared with Fig. 4, which indicates that even 1 ppm of magnesium pick-up from slag phase can enhance the formation of spinel inclusions in the steel melts.¹⁰⁾

Fig. 7.

A stability predominance diagram for the Fe-18%Cr-8%Ni-1%Mn-Si-Al-O (1 ppm Mg) system at 1773 K (Calculated from FactSage^TM7.3).

The average sizes of inclusions in the steel reacted with the CaO–SiO₂–MgO (C/S = 1.0) and the CaO–Al₂O₃–MgO (C/A = 1.3) fluxes are shown in Fig. 8 as a function of reaction time. The average size of inclusions (under 1.5 μm) is negligibly changed with reaction time when the CaO–SiO₂–MgO flux is added to the steel because the driving force for the growth of MnO–SiO₂ inclusion is very low. On the other hand, the average size of inclusions increases from 1.5 μm to around 3 μm with increasing reaction time when the CaO–Al₂O₃–MgO flux is added to the steel.

Fig. 8.

Average size of non-metallic inclusions in the steel reacted with the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes as the reaction time increases. (Online version in color.)

The calculated PDF of the inclusions is plotted over each inclusion size in Fig. 9 with different flux compositions and reaction times on a logarithmic scale. The PDF of inclusions for the CaO–SiO₂–MgO flux has a tendency to linearly decrease, irrespective of the reaction time, showing a fractal distribution (even though there is some scattering). The PDF of inclusions for the CaO–Al₂O₃–MgO flux at 0 and 10 min has a similar tendency with the PDF of inclusions for the CaO–SiO₂–MgO flux, but then the PDF of inclusions shows a lognormal distribution with reaction times beyond 10 min. When the Al-killed steel is deoxidized or reoxidized during steelmaking processes, the PDF of inclusions has a lognormal shape because the growth of inclusions is followed by the diffusion of solutes and the coarsening of inclusions.^25,26,27) A fractal distribution in the PDF of inclusions indicates that the growth of inclusions is determined by the steady state breakup and coalescence of inclusions after the solute and inclusions are in equilibrium.^25,26,27) Therefore, the lognormal shape in the PDF after 10 min implies the aluminum and magnesium pick-up affect the growth of inclusions in the steel.

Fig. 9.

Population density function (PDF) of non-metallic inclusions in the steel reacted with (a) the CaO–SiO₂–MgO and (b) the CaO–Al₂O₃–MgO fluxes as the reaction time increases. (Online version in color.)

The morphologies of inclusions reacted with the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes are shown in Fig. 10 with different reaction times after electrolytic extraction of the samples. The peaks of C and Pt in EDS data were originated from the polycarbonate membrane and Pt coating. Before the fluxes are added to the steel, spherical MnO–SiO₂ inclusions are observed. When the CaO–SiO₂–MgO flux is added to the molten steel, neither the composition nor the morphology of MnO–SiO₂ inclusions are appreciably changed with reaction time. The MnO–SiO₂ is mainly located at the liquid (or SiO₂-saturated) region (Figs. 4 and 6) so the MnO–SiO₂ inclusions maintained a spherical morphology to reduce their surface energy in the molten steel. When the CaO–Al₂O₃–MgO flux is added to the molten steel, spherical MnO–SiO₂–Al₂O₃ inclusions are observed at 5 min because the aluminum, which was transferred from the slag by Eq. (2), causes the reduction of MnO and SiO₂ in the MnO–SiO₂ inclusions as given in Eqs. (5) and (6).   

$$3{\left(\text{MnO}\right)}_{\text{inclusion}}+2\left[\text{Al}\right]=3\left[\text{Mn}\right]+{\left({\text{Al}}_2{\text{O}}_3\right)}_{\text{inclusion}}$$(5)

  

$$3{\left({\text{SiO}}_2\right)}_{\text{inclusion}}+4\left[\text{Al}\right]=3\left[\text{Si}\right]+2{\left({\text{Al}}_2{\text{O}}_3\right)}_{\text{inclusion}}$$(6)

Finally, the polyhedral spinel inclusions are observed after 15 min from Eqs. (3) and (4).

Fig. 10.

Observed morphology of non-metallic inclusions in the steel reacted with the CaO–SiO₂–MgO and the CaO–Al₂O₃–MgO fluxes after electrolytic extraction method. (Online version in color.)

The formation of alumina inclusions is strongly dependent on the supersaturation degree of alumina, S_Al2O3, which is defined by following equations:⁵⁰⁾   

$$S_{A{l}_2{O}_3}=\left(a_{Al}^2\cdot a_O^3\right)/{\left(a_{Al}^2\cdot a_O^3\right)}_{eq}=\left(a_{Al}^2\cdot a_O^3\right)/\left(1/K_{A{l}_2{O}_3}\right)$$(7)

  

$$log{a}_{Al}=\sum e_{Al}^j\left[\%j\right]+log\left[\%Al\right]$$(8)

  

$$log{a}_O=\sum e_O^j\left[\%j\right]+log\left[\%O\right]$$(9)

where, a_Al, a_O, and $e_i^j$ are the activity of aluminum, the activity of oxygen, and the interaction coefficient (i = Al and O; j = Al, O, Si, Mn, Ni, and Cr) in a liquid steel with respect to 1 wt% standard state, respectively. The interaction coefficients used for the calculation are listed in Table 2.^51,52,53) The calculated supersaturation degree is shown in Fig. 11. As the aluminum content increases with reaction time, the calculated S_Al2O3 in the steel increases slightly after 10 min. Suito and Ohta⁵⁴⁾ reported that the critical value of the supersaturation degree per one mole of oxygen, S_O, for the precipitation of alumina is approx. 15 from experimental results and 529 from theoretical calculations. Because the soluble oxygen is consumed by an interfacial reaction between the slag and metal in the present study, the calculated S_Al2O3 is too low to form new alumina particles in the steel, even though the aluminum content increases with reaction time.

Table 2. Interaction coefficients, $e_i^j$, in the present study.^51,52,53)

i\j	Al	O	Si	Mn	Ni	Cr
Al	0.045	−1.98	0.056	–	−0.0173	0.012
O	−1.17	−0.23	−0.066	−0.021	0.0027	−0.063

Fig. 11.

Degree of supersaturation for precipitating alumina inclusions in the steel reacted with the CaO–Al₂O₃–MgO flux as the reaction time increases.

Dekkers et al.⁵⁵⁾ demonstrated that the morphology of non-metallic inclusions is determined by the relative growth rate, the relative surface energy of the facets, or the presence of impurities on the facets. The degree of supersaturation is an important factor to understand the morphology of inclusions because the degree of supersaturation has a proportional influence on the growth rate.^55,56) An increase in the degree of supersaturation causes a change in the inclusion morphology from polyhedral toward dendritic and spherical via different mechanisms.^55,56) Therefore, the polyhedral morphology of inclusions is mainly observed due to the low degree of supersaturation in the steel after 15 min (Fig. 10).

The schematic diagram showing the mechanism of the formation of spinel inclusions is illustrated in Fig. 12 when the CaO–Al₂O₃–MgO flux is added to the Si-killed stainless steel melt. Because the aluminum, which was transferred from the slag, reduces MnO and SiO₂ in the pre-existing MnO–SiO₂ inclusions as given in Eqs. (5) and (6), alumina-rich inclusions are formed. As the aluminum content increases in the steel, magnesium is also transferred from the slag. The magnesium transforms the alumina-rich inclusions into spinel-type inclusions by Eq. (4). Finally, spinel-type inclusions are grown by the diffusion of aluminum, magnesium, and oxygen, leading to the polyhedral morphology.

Fig. 12.

A schematic diagram showing the evolution mechanism of non-metallic inclusions in the steel reacted with the CaO–Al₂O₃–MgO flux. (Online version in color.)

As described above, CaO–Al₂O₃ based flux causes the aluminum and magnesium pick-up into the steel which modifies the initial MnO–SiO₂ type inclusions into alumina-rich and spinel-type inclusions, while CaO–SiO₂ based flux maintains the initial silicate inclusions. Therefore, it is recommended to use CaO–SiO₂ based flux to suppress the formation of harmful inclusions in the tundish when producing Si-killed stainless steel with high cleanliness.

3.3. Prediction of Inclusion Evolution in Molten 304 Stainless Steel in Continuous Casting Tundish

The molten steel poured from the ladle flows for a few minutes through the tundish before solidifying in the continuous casting mold. The molten steel is reacted with the tundish flux while it remains in the tundish, which results in the formation of reoxidative alumina inclusions.^20,44) In the present study, the computational simulation used to predict the inclusion evolution in the molten in tundish is carried out based on the ReSMI multiphase reaction model, modified from the EERZ model.^{37,38,39,40,41,42,43,44)} The ReSMI multiphase reaction model has been successfully utilized in kinetic calculations using FactSage^TM (ver. 7.3) macro simulation with FTmisc and FToxid databases.

To simplify the model and calculation, the simulations were performed using the following assumptions: 1) the molten steel is reacted with the slag at steady state in the tundish without external interventions, 2) the dimensions of the molten steel and slag per unit time are constant during the calculation, 3) the molten steel per unit time moves to the next step while the slag per unit time remains after the reaction, and 4) the mixing of steel to steel and slag to slag is negligible during the simulation in the tundish (Fig. 13).

Fig. 13.

A schematic diagram showing the reactions in the tundish during the continuous casting process.

It has been reported that the exogenous slag-type inclusions, which usually have an average composition of 40 wt% CaO, 34 wt% SiO₂, 14 wt% Al₂O₃, and 12 wt% MgO (CSAM system), mainly remain in the stainless steel melts after ladle treatment stage.^9,17) It is assumed that 50 ppm of CSAM inclusions are contained in the initial melt poured into tundish. After each steel melt per unit time is simulated for the duration time in the tundish, the simulated results are output; these correspond to steel data at the outlet of the tundish process. The detailed conditions for the present simulation are listed in Table 3.

Table 3. Simulation conditions for predicting the evolution of inclusions in the tundish during continuous casting.

Parameters	Value (unit)
Weight of the tundish flux per unit time	40 (kg)
Weight of metal per unit time	1.84 (ton)
Total weight of tundish flux in the tundish	200 (kg)
Area of slag-metal interface per unit time	0.72 (m2)
Density of molten steel	6.96 (kg/m3)
Density of molten slag	2.50 (kg/m3)
Temperature in the tundish	1773 (K)
Unit time	1 (min)
Duration of molten steel in the tundish	5 (min)
Casting time	60 (min)

The calculation results for predicting the evolution of inclusions in molten 304 stainless steel in the tundish with different flux compositions are shown in Fig. 14 as a function of casting time. Spinel and melilite-based inclusions are formed in the molten steel, and the Al₂O₃ content in the total inclusions also increases from 14 wt%, which refers to the Al₂O₃ content in the original slag-type CSAM inclusion system, to 43 wt% due to the slag-metal reaction in tundish at the initial casting time. Then, the percentages of spinel and melilite inclusions and Al₂O₃ content in the total inclusions continuously decrease with casting time. This is because the Al₂O₃ content in the CaO–Al₂O₃–MgO flux continuously decreases as the slag/metal reaction proceeds without additional input of flux and the molten steel reacting with the flux is renewed in tundish as the casting continues. On the other hand, the inclusions in molten 304 stainless steel maintain their initial composition (solid lines in Fig. 14(b)) during the casting process when the CaO–SiO₂–MgO flux is used for the simulation in the tundish.

Fig. 14.

The simulation results for the evolution of non-metallic inclusions during the steel reaction with (a) the CaO–Al₂O₃–MgO and (b) the CaO–SiO₂–MgO fluxes in the tundish as a function of casting time.

Because the tundish flux is continuously (or on-and-off) added to molten steel in the real continuous casting process, the present simulation implies that the CaO–Al₂O₃-based flux causes the formation of alumina- and/or spinel-type inclusions in the tundish before the steel reaches the casting mold. Therefore, it is recommended to use the CaO–SiO₂-based flux to suppress the formation of harmful inclusions in the tundish when producing Si-killed stainless steel with high cleanliness.

4. Conclusions

Using high temperature experiments, the effects of the CaO–Al₂O₃–MgO and the CaO–SiO₂–MgO fluxes on the evolution mechanism of non-metallic inclusions in Si-killed 304 stainless steel melt in a continuous casting tundish were investigated. Furthermore, the computational simulation based on the ReSMI multiphase reaction model was carried out to predict the evolution of non-metallic inclusions in the tundish during a continuous casting process. The results of this study can be summarized as follows.

(1) The aluminum content in the steel continuously increases with reaction time when the CaO–Al₂O₃–MgO flux is added to the Si-killed 304 stainless steel because silicon in the steel reduces Al₂O₃ from the slag. The accumulated aluminum, caused by a chemical reaction between the steel and the CaO–Al₂O₃–MgO flux at the steel/slag interface, decreases the soluble oxygen level at the interface so that the oxygen content in the bulk steel decreases with reaction time. On the other hand, the elemental contents in the steel are negligibly changed when the CaO–SiO₂–MgO flux is added to the Si-killed 304 stainless steel melt.

(2) The aluminum originating from the interfacial reaction between the steel and the CaO–Al₂O₃–MgO flux modifies the pre-exiting MnO–SiO₂ inclusion into the alumina-rich inclusions by reducing MnO and SiO₂ in the MnO–SiO₂ inclusions. It is difficult to precipitate new-born particles in the steel, even though the aluminum content continuously increases, because the decrease in the oxygen content caused by the interfacial reaction between the steel and the CaO–Al₂O₃–MgO flux decreases the level of supersaturation for the precipitation of alumina in the steel. The polyhedral morphology of spinel-type inclusions is determined by diffusion of aluminum, magnesium, and oxygen due to the relatively slow growth rate of inclusions. With the CaO–SiO₂–MgO flux, the composition and morphology of inclusions are negligibly changed in the Si-killed 304 stainless steel.

(3) For the computational simulation, it is noteworthy that the CaO–Al₂O₃-based flux considerably modifies Mn-silicate inclusions into spinel- and melilite-type inclusions, even at the initial stage of the continuous casting process, while the CaO–SiO₂-based flux has a smaller influence on the modification of inclusions in the Si-killed 304 stainless steel.

Acknowledgements

This work was partly supported by Korea Evaluation Institute of Industrial Technology (KEIT, with Grant number 20009956), funded by the Ministry of Trade, Industry and Energy (MOTIE), Korea.

References

• 1)   J. H.  Park: Calphad, 35 (2011), 455.
• 2)   J. S.  Park,  C.  Lee and  J. H.  Park: Metall. Mater. Trans. B, 43 (2012), 1550.
• 3)   M. H.  Lee and  J. H.  Park: Metall. Mater. Trans. B, 49 (2018), 877.
• 4)   M. H.  Lee,  R.  Kim and  J. H.  Park: Sci. Rep., 9 (2019), 6369.
• 5)   J. W.  Kim,  S. K.  Kim,  D. S.  Kim,  Y. D.  Lee and  P. K.  Yang: ISIJ Int., 36 (1996), S140.
• 6)   C.  Mapelli and  P.  Nolli: ISIJ Int., 43 (2003), 1191.
• 7)   J. H.  Park and  D. S.  Kim: Metall. Mater. Trans. B, 36 (2005), 495.
• 8)   J. H.  Park,  D. S.  Kim and  S. B.  Lee: Metall. Mater. Trans. B, 36 (2005), 67.
• 9)   J. H.  Park: Mater. Sci. Eng. A, 472 (2008), 43.
• 10)   J. H.  Park and  H.  Todoroki: ISIJ Int., 50 (2010), 1333.
• 11)   Y.  Bi,  A. V.  Karasev and  P. G.  Jönsson: ISIJ Int., 53 (2013), 2099.
• 12)   P.  Yan,  S.  Huang,  L.  Pandelaers,  J.  Van Dyck,  M.  Guo and  B.  Blanpain: Metall. Mater. Trans. B, 44 (2013), 1105.
• 13)   S.  Li,  L.  Zhang,  Y.  Ren,  W.  Fang,  W.  Yang,  S.  Shao,  J.  Yang and  W.  Mao: ISIJ Int., 56 (2016), 584.
• 14)   D.  Kruger and  A.  Garbers-Craig: Metall. Mater. Trans. B, 48 (2017), 1514.
• 15)   J.  Li,  G.  Cheng,  Q.  Ruan,  J.  Pan and  X.  Chen: Metall. Mater. Trans. B, 49 (2018), 2357.
• 16)   H.  Du,  A.  Karasev,  O.  Sundqvist and  P. G.  Jönsson: Metals, 9 (2019), 74.
• 17)   W. Y.  Kim,  K. S.  Kim and  S. Y.  Kim: Metall. Mater. Trans. B, 52 (2021), 652.
• 18)   H.  Sun and  K.  Mori: ISIJ Int., 36 (1996), S34.
• 19)   D. C.  Park,  I. H.  Jung,  P. C. H.  Rhee and  H. G.  Lee: ISIJ Int., 44 (2004), 1669.
• 20)   T. S.  Kim,  Y.  Chung,  L.  Holappa and  J. H.  Park: Metall. Mater. Trans. B, 48 (2017), 1736.
• 21)   T.  Nishi and  K.  Shinme: Tetsu-to-Hagané, 84 (1998), 97 (in Japanese).
• 22)   J. S.  Park and  J. H.  Park: Metall. Mater. Trans. B, 45 (2014), 953.
• 23)   J.  Zhang and  H. G.  Lee: ISIJ Int., 44 (2004), 1629.
• 24)   M. A.  Van Ende,  M.  Guo,  E.  Zinngrebe,  R.  Dekkers,  J.  Proost,  B.  Blanpain and  P.  Wollants: Ironmaking Steelmaking, 36 (2009), 201.
• 25)   E.  Zinngrebe,  C.  Van Hoek,  H.  Visser,  A.  Westendorp and  I. H.  Jung: ISIJ Int., 52 (2012), 52.
• 26)   M. A.  Van Ende,  M.  Guo,  E.  Zinngrebe,  B.  Blanpain and  I. H.  Jung: ISIJ Int., 53 (2013), 1974.
• 27)   M. D.  Seo,  J. W.  Cho,  K. C.  Kim and  S. H.  Kim: ISIJ Int., 54 (2014), 475.
• 28)   S. P. T.  Piva and  P. C.  Pistorius: Metall. Mater. Trans. B, 52 (2021), 6.
• 29)   Y. S.  Hsieh,  Y.  Watanabe,  S.  Asai and  I.  Muchi: Tetsu-to-Hagané, 69 (1983), 596 (in Japanese).
• 30)   R.  Ding,  B.  Blanpain,  P. T.  Jones and  P.  Wollants: Metall. Mater. Trans. B, 31 (2000), 197.
• 31)   M. A.  van Ende,  Y. M.  Kim,  M. K.  Cho,  J.  Choi and  I. H.  Jung: Metall. Mater. Trans. B, 42 (2011), 477.
• 32)   A.  Harada,  N.  Maruoka,  H.  Shibata and  S.  Kitamura: ISIJ Int., 53 (2013), 2110.
• 33)   A.  Harada,  N.  Maruoka,  H.  Shibata and  S.  Kitamura: ISIJ Int., 53 (2013), 2118.
• 34)   D.  Roy,  P. C.  Pistorius and  R. J.  Fruehan: Metall. Mater. Trans. B, 44 (2013), 1095.
• 35)   D.  Roy,  P. C.  Pistorius and  R. J.  Fruehan: Metall. Mater. Trans. B, 44 (2013), 1086.
• 36)   A.  Harada,  N.  Maruoka,  H.  Shibata,  M.  Zeze,  N.  Asahara,  F.  Huang and  S.  Kitamura: ISIJ Int., 54 (2014), 2569.
• 37)   M. A.  Van Ende and  I. H.  Jung: ISIJ Int., 54 (2014), 489.
• 38)   Y.  Ren,  L.  Zhang,  H.  Ling,  Y.  Wang,  D.  Pan,  Q.  Ren and  X.  Wang: Metall. Mater. Trans. B, 48 (2017), 1433.
• 39)   J. H.  Shin,  Y.  Chung and  J. H.  Park: Metall. Mater. Trans. B, 48 (2017), 46.
• 40)   J. H.  Shin and  J. H.  Park: Metall. Mater. Trans. B, 48 (2017), 2820.
• 41)   M. A.  Van Ende and  I. H.  Jung: Metall. Mater. Trans. B, 48 (2017), 28.
• 42)   M. S.  Kim and  Y. B.  Kang: Calphad, 61 (2018), 105.
• 43)   J. H.  Shin and  J. H.  Park: ISIJ Int., 58 (2018), 88.
• 44)   T. S.  Kim,  L.  Holappa and  J. H.  Park: Ironmaking Steelmaking, 47 (2020), 84.
• 45)   J. S.  Park and  J. H.  Park: Steel Res. Int., 85 (2014), 1303.
• 46)   S. K.  Kwon,  Y. M.  Kong and  J. H.  Park: Met. Mater. Int., 20 (2014), 959.
• 47)   N.  Choi,  K. R.  Lim,  Y. S.  Na,  U.  Glatzel and  J. H.  Park: J. Alloy. Compd., 763 (2018), 546.
• 48)   T.  Tanaka,  H.  Goto,  M.  Nakamoto,  M.  Suzuki,  M.  Hanao,  M.  Zeze,  H.  Yamamura and  T.  Yoshikawa: ISIJ Int., 56 (2016), 944.
• 49)   P.  Ni,  T.  Tanaka,  M.  Suzuki,  M.  Nakamoto and  P. G.  Jönsson: ISIJ Int., 59 (2019), 737.
• 50)   G.  Li and  H.  Suito: ISIJ Int., 37 (1997), 762.
• 51)  The Japan Society for the Promotion of Science, The 19th Committee on Steelmaking: Steelmaking Data Sourcebook, Revised ed., Gordon and Breach Science, New York, (1988), 95.
• 52)   M.  Kishi,  R.  Inoue and  H.  Suito: ISIJ Int., 34 (1994), 859.
• 53)   G.  Li,  R.  Inoue and  H.  Suito: Steel Res., 67 (1996), 528.
• 54)   H.  Suito and  H.  Ohta: ISIJ Int., 46 (2006), 33.
• 55)   R.  Dekkers,  B.  Blanpain and  P.  Wollants: Metall. Mater. Trans. B, 34 (2003), 161.
• 56)   I.  Sunagawa: Morphology of Crystals, Terra Scientific Publishing Company, Tokyo, (1987), 509.