ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Kinetic Analysis of Compositional Changes in Inclusions during Ladle Refining
Akifumi HaradaNobuhiro MaruokaHiroyuki ShibataMasafumi ZezeNorifumi AsaharaFuxiang HuangShin-ya Kitamura
Author information
JOURNALS OPEN ACCESS FULL-TEXT HTML

2014 Volume 54 Issue 11 Pages 2569-2577

Details
Abstract

It is well known that the composition of inclusions is determined by alloying elements and by reaction with slag. For example, MgO·Al2O3 spinel-type inclusions form, even though Mg is not added, due to the supply of Mg through the reaction between slag and metal. To clarify the mechanism of compositional changes in inclusions, the authors have developed a kinetic model to simulate the reactions during the ladle refining process. In this study, experiments were conducted using an induction furnace, and the compositional changes in molten steel, slag, and inclusions were investigated. The inclusions were analyzed by P-SEM, which incorporates an automatic analysis system. By the application of the developed simulation model to these experiments, the validity of the model was evaluated. The inclusion composition gradually changed from Al2O3 to MgO·Al2O3 after the addition of Al, and the inclusions originating from slag were also observed at all times. The compositional change of the deoxidation product by the model calculation corresponded well to the observed variation in the composition of inclusions, and the calculated composition of inclusions originating from slag also agreed with the experimental results. The rate of compositional change increased with increasing Ar gas flow rate, and this tendency was captured well by the model. Therefore, the validity of the developed model is considered to be confirmed.

1. Introduction

It is well known that the composition of inclusions is determined by alloying elements and by reaction with slag. For example, MgO·Al2O3 spinel-type inclusions form, even though Mg is not added, due to the supply of Mg through the reaction between slag and metal.1,2,3) To clarify the mechanism of compositional change in inclusions, the authors have developed a kinetic model to simulate the reactions during the ladle refining process.4,5) In this model, five phases were considered: molten steel, slag, refractory, deoxidation product, and inclusions originating from slag. For the reactions between molten steel/slag and molten steel/inclusions originating from slag, a coupled reaction model6) was applied. The empirical equation of the MgO dissolution rate from refractory into slag7) was applied to the reaction between slag and refractory. The formation of deoxidation product was calculated based on the free energy change of formation. The formation of inclusions originating from slag due to the entrapment of top slag, agglomeration of deoxidation product with inclusions originating from slag, and floatation of deoxidation product and inclusions originating from slag were considered. By the developed model, the mass and compositional changes of molten steel, slag, and inclusions originating from slag and various deoxidation products are able to be calculated. The results of calculations exhibited good agreement with the operational results reported by Graham et al.8) However, this agreement is not sufficient to confirm the validity of the model because the operational conditions of that study were limited.

In this study, the experiments that simulate the ladle refining process were performed using an induction furnace, and the compositional changes in the molten steel, slag, and inclusions were investigated. The developed model was applied to these experimental conditions, and the validity of the model was evaluated.

2. Experimental Methods

The experiment was conducted in a high-frequency induction furnace shown in Fig. 1. Stamped MgO powder was used as the refractory material. Porous brick was set in the bottom of the furnace to stir the molten steel using Ar. A total of 70 kg electrolytic iron, metallic chrome, and granular carbon were placed in the furnace. The composition of the metal is listed in Table 1. To maintain an Ar atmosphere in the furnace, two sets of sealing boxes with sampling holes made of steel were placed on the top of the furnace, and Ar was injected from the side wall of the sealing boxes. During the experiment, the top hole of the outer box was covered by ceramic wool, except for the purpose of adding materials, temperature measurements, or sampling. After the steel was melted, 3.5 kg of an oxide mixture was added on the steel surface. The mixture was composed of the reagents of SiO2 and Al2O3 and the powder of metallurgical lime in an appropriate mixing ratio. The composition of the initial slag is listed in Table 2. MgO was not added to the initial slag because it was predicted to dissolve from the refractory. After the added slag was melted and the temperature reached around 1923 K, Al foil wrapped around a steel rod was immersed, and this time was set as the start of the experiments. The target value of Al was 0.2 mass%. The molten steel was sampled using a dip-up-type sampler, and the molten slag that adhered to the steel rod was sampled. Temperature measurements were achieved by the immersion of consumable thermocouples. The sampling and temperature measurements were conducted within adequate time intervals. The experimental conditions are listed in Table 3. The Ar gas flow rate for stirring was varied from 5 to 20 Nl/min. The composition of the metal was analyzed by optical emission spectroscopy. The Mg and Ca contents in the metal were analyzed by inductively coupled plasma atomic emission spectrometry (ICP–AES). The slag composition was analyzed by fluorescent X-ray analysis. The inclusions in the metal sample were analyzed using field emission scanning electron microscopy (FE-SEM) with energy dispersive X-ray spectrometry (EDS). However, because each inclusion tends to have a different composition, analysis of the compositional distribution of inclusions by FE-SEM requires that many inclusions must be analyzed, which takes a very long time. Recently, a SEM instrument equipped with an automatic analysis system has been developed. This system can analyze sizes, compositions, and number of inclusions automatically, and it is easy to obtain the results of more than 1000 inclusions in a few hours. Thus, in this study, the inclusions in the metal samples were analyzed using this system (P–SEM:9,10) ASPEX corporation) in addition to the normal analysis using FE-SEM with EDS.

Fig. 1.

Schematic diagram of the experimental apparatus employed.

Table 1. Initial composition of the metal (mass%).
CSiCr
0.0150.319.5
Table 2. Initial composition of the slag (mass%).
CaOSiO2Al2O3MgO
555400
Table 3. Experimental conditions.
ExperimentDeoxidizerTemperature (K)Ar flow rate (NL/min)
1Al19235
220

3. Calculation Conditions

The kinetic model reported in previous studies4,5) was applied to the experiments conducted. In addition, empirical Eqs. (1) and (2) for the Mg dissolution rate from MgO refractory into molten steel obtained in a previous study11) were applied to the model. Equation (1) represents the dissolution rate of Mg from the refractory, and Eq. (2) represents the dimensionless correlation to calculate the mass transfer coefficient of Mg in molten metal:   

d[%Mg] dt  =  A  ρ m   k Mg W  ( [%Mg] *  -[ %Mg ] ) (1)
  
k Mg ( l/D ) =1.3R e 1/2 S c 1/3 , (2)
where A is the interracial area (m2), W is the mass of metal (kg), kMg is the mass transfer coefficient of Mg in the metal phase (m/s), ρm is the density of the metal (kg/m3), [%Mg] is the concentration of Mg in the metal, and [%Mg]* is the equilibrium concentration of Mg. In Eq. (2), Sc is the Schmidt number (v/D), Re is the Reynolds number (l·u′/v), u′ is the characteristic velocity (m/s), v is the dynamic viscosity (m2/s), D is the diffusion coefficient (m2/s), and l is the characteristic length (m).

To apply this equation to the model, A is calculated by the geometrical interracial area between molten steel and refractory, l is determined as the one-half power of A and u′ is assumed as the rising velocity of molten steel which was used to calculate the relative velocity of slag and refractory in the previous paper.4) In addition, [%Mg]* is calculated by the equilibrium relation with spinel which form on the refractory surface. For the physical properties, 7.86×10–7 (m2/s), 3.5×10–9 (m2/s) are used as v and D, respectively.

The calculation conditions are listed in Table 4. Most of the parameters in the kinetic model included in Eqs. (1) and (2) are able to be determined theoretically or empirically;4,5,11) however the mass ratio of the entrapment of slag in the molten steel to the total mass of slag (entrapment parameter, α), the mass ratio of the floatation of deoxidation products and inclusions originating from slag to the total mass of them (floatation parameter, β), the mass ratio of the agglomeration of deoxidation products with inclusions originating from slag to the total mass of deoxidation products (agglomeration parameter, γ), and the mass ratio of the bulk zone to the total mass for molten steel and slag phase (Vb/V) have to be determined. These parameters were determined to match the calculation results with the change in total oxygen content and the change in the average composition of inclusions.5) In this study, the parameters listed in Table 5 were used. The values of the parameters were different from those used in the previous paper5) and they had to be changed by the experimental condition. These parameters have great influence on the total inclusion content and the average inclusion composition.5) However, to clarify the ruling factors, more investigations would be necessary.

Table 4. Calculation conditions.
Weight of metal (kg)70
Weight of slag (kg)3.5
Cross-sectional area of furnace (cm2)434
Calculation time (s)1800
Calculation step (s)0.1
Addition time of Al (s)10
Addition amount of Al in Exp. 1 (g/kg)3.6
Addition amount of Al in Exp. 2 (g/kg)3.3
Table 5. Parameters used for the calculation of experiment 1 and 2.
ParametersExperiment 1Experiment 2
(1)Entrapment parameter (α)1.0 × 10–6 (mass%/s)1.0 × 10–5 (mass%/s)
(2)Floatation parameter (β)0.05 (mass%/s)0.1 (mass%/s)
(3)Agglomeration parameter (γ)1.0 × 10–5 (mass%/s)1.0 × 10–5 (mass%/s)
(4)Bulk zone in molten steel and slag phase (Vb/V)0.80.8

In general, by Al deoxidation, the deoxidation products of Al2O3 form large-sized clusters having a high floatation rate. Thus, in the calculations, the floatation ratio of each inclusion was changed based on Stokes’ law, as shown in Eq. (3):   

=  d 2 ( ρ Fe - ρ i ) g 18μ , (3)
where u is the floatation rate of inclusions (m/s), d is the diameter of inclusions (m), ρFe and ρi are the density of molten steel and inclusions, respectively (kg/m3), g is the gravity acceleration (m/s2), and μ is the viscosity coefficient of molten steel (kg/(m·s)). Because Stokes’ law is not valid for a bath stirred by gas bubbling, in this calculation, the ratio of the floatation rate of each inclusion to that of each single particle of Al2O3 calculated by Eq. (3) was used as a parameter to indicate the difference of the floatation parameter. The ratios between the flotation rates of each type of inclusion to that of single particles of Al2O3 were calculated, and the same values were used as the ratios of floatation parameters of each inclusion to that of a single particle of Al2O3. The floatation rate and the floatation parameter of the cluster, spinel, and inclusions originating from slag are listed in Table 6. The deoxidation product of Al2O3 was assumed to behave as a cluster for the first minute after the addition of Al, and to behave as a single particle after that.
Table 6. Floatation rate and floatation parameter of each inclusion type.
InclusionDensity (g/cm3)Diameter (μm)Flotation RateFloatation Parameter
Alumina4.010ufloatβ
Alumina cluster4.0100100 ufloat100 β
Spinel3.6101.1 ufloat1.1 β
Inclusion originating from slag3.0101.3 ufloat1.3 β

※ Density of molten steel = 7.0 (g/cm3)

The analyzed compositions of the molten steel and slag before the addition of Al were used as the initial compositions.

4. Results

The changes in the content of each element of molten steel and slag are shown in Figs. 2, 3, 4, 5. In both experiments, the Al content increased rapidly after deoxidation and decreased gradually by the reaction with slag. The Si content increased due to the reduction of SiO2 in the slag by Al in the molten steel. Both Mg and Ca contents were less than 0.5 mass ppm; however, the accuracy of the quantitative analysis of these components was not sufficiently high in the concentration range lower than 1 mass ppm. The MgO content in the slag increased as the refractory was dissolved into the slag. In each case, the initial MgO content was not zero because the refractory had already begun to dissolve before Al addition. The calculation results for the molten steel and slag were in a good agreement with the experiments. However, the calculation results for the Mg and Ca content were different from the experimental results. The accuracy of Mg and Ca content analysis should be improved in the future. Figure 6 shows the change in the oxygen content of the molten steel. In the experiments, the oxygen content decreased rapidly after deoxidation and reached about 25 ppm after 2.5 min. The calculated value of total oxygen was close to the experimental results.

Fig. 2.

Calculation results of the compositional change in molten steel compared with the experimental results for experiment 1.

Fig. 3.

Calculation results of the components in molten slag compared with the experimental results for experiment 1.

Fig. 4.

Calculation results of the compositional change in molten steel compared with the experimental results for experiment 2.

Fig. 5.

Calculation results of the components in molten slag compared with the experimental results for experiment 2.

Fig. 6.

Calculation results of the change in oxygen content in molten steel for experiments 1 (a) and 2 (b).

The typical inclusions observed by FE-SEM are shown in Figs. 7 and 8. Most inclusions were Al2O3 and clusters at 0.5 min after deoxidation. At 10 min, the content of inclusions that included around 20 mass% MgO increased, and some inclusions were considered to be MgO·Al2O3 spinel-type inclusions because the solid solution of MgO·Al2O3 spinel contained 20–30 mass% MgO at this experimental temperature.

Fig. 7.

Typical inclusions of experiment 1 observed by FE-SEM.

Fig. 8.

Typical inclusions of experiment 2 observed by FE-SEM.

The analyzed results by P-SEM are shown in Figs. 9 and 10, and are positioned within the CaO–Al2O3–MgO ternary phase diagram at 1923 K. In experiment 1, as shown in Fig. 9, most inclusions were the Al2O3 type at 0.5 min. After 5 min, the inclusions that included MgO increased gradually. Finally, Al2O3 and MgO·Al2O3 spinel-type inclusions coexisted. Inclusions located in a liquid phase were also observed in each sample. These inclusions would have originated from the slag entrapped in the molten steel by stirring. In experiment 2, as shown in Fig. 10, Al2O3-type inclusions were observed at 0.5 min. However, many inclusions changed to the spinel type at 5 min, and most inclusions were located in the compositional region of spinel after 10 min. The rate of change in the composition of inclusions in experiment 2 was faster than that in experiment 1 due to an increase in the Ar gas flow rate. Liquid phase inclusions were also observed in experiment 2.

Fig. 9.

Composition of inclusions at each time observed by P-SEM compared with the calculation results for experiment 1.

Fig. 10.

Composition of inclusions at each time observed by P-SEM compared with the calculation results for experiment 2.

In Figs. 9 and 10, the calculated changes in the average composition of the deoxidation products and inclusions originating from slag were plotted. In experiment 1 (Fig. 9), the calculated average composition of the deoxidation product was close to Al2O3 at 0.5 min. After 5 min, the average composition gradually moved to the spinel region. The calculated composition of inclusions originating from slag was located in the liquid phase during the experiment. Conversely, in experiment 2 (Fig. 10), at 0.5 min, the calculated composition of the average composition of the deoxidation products was also Al2O3; however, the composition moved to the spinel region at 5 min. The inclusions originating from slag were located in the liquid region during the entire time. These calculated results agreed well with those of the experiments.

Figure 11 shows the calculated changes in the average composition of total inclusions, including deoxidation products and inclusions originating from slag. The phenomenon whereby the rate of increase of the MgO content increased with a higher Ar gas flow rate was expressed well by the model calculations. Therefore, the validity of the developed model is considered to be confirmed.

Fig. 11.

Change in the average composition of total inclusions compared with the calculation results for experiments 1 (a) and 2 (b).

Figure 12 shows the calculated change in the amount of each type of inclusion with time. Al2O3, MgO·Al2O3 spinel, and inclusions originating from slag were formed in both experiments. Soon after the addition of Al, Al2O3 formed as a deoxidation product. Then, the content of Al2O3 decreased due to floatation into the top slag and agglomeration with inclusions originating from slag. MgO·Al2O3 also formed as a deoxidation product at the early stage, and its content increased gradually. The formation of the spinel was caused by the increase in the Mg content of the molten steel. Inclusions originating from slag increased from the early stage due to the entrapment of the slag. Figure 13 shows the calculated change in the composition of inclusions originating from slag. At the early stage, the Al2O3 content increased slightly as the inclusions originating from slag agglomerated with the alumna. After the middle period of the treatment, CaO, Al2O3, and MgO contents became almost constant. Figure 14 shows the calculated change in the mass ratio of each inclusion type to the total. In both experiments, the ratio of alumina was very high shortly after deoxidation. Then, the ratio of MgO·Al2O3 and inclusions originating from slag increased gradually. Comparing the calculation results of experiment 1 with those of experiment 2, the rate of change of inclusions from alumina to MgO·Al2O3 and inclusions originating from slag in experiment 2 was faster than that in experiment 1.

Fig. 12.

Change in the amount of each inclusion type with time by the model calculations for experiments 1 (a) and 2 (b).

Fig. 13.

Change in the composition of inclusions originating from slag with time by the model calculations for experiments 1 (a) and 2 (b).

Fig. 14.

Change in the ratio of inclusions originating from slag and deoxidation product with time by the model calculations for experiments 1 (a) and 2 (b).

5. Discussion

The experimental results of the compositional change in the inclusions after Al deoxidation were described well by the developed model. In this section, the effects of each reaction on the Mg content in the molten steel and the formation of spinel-type inclusions are discussed. The following reactions that affected the Mg content and spinel formation are considered: (1) the reaction between the molten steel and top slag, (2) the reaction between the top slag and refractory, and (3) the reaction between the molten steel and refractory. The calculation conditions are listed in Table 7. The calculation condition1 corresponds to the model calculation shown in the previous session. On the contrary, in the calculation condition 2 and 3, the reaction between slag and refractory and the reaction between molten steel and refractory was not considered, respectively. By the comparison of these calculation results, the effect of the reaction between the slag and refractory can be determined from the results of conditions 1 and 2. In addition, the effect of the reaction between the molten steel and refractory can be determined from the results of conditions 1 and 3. Moreover, the effect of the reaction between the molten steel and slag can be clarified from the differences between the calculation results of conditions 1, 2, and 3. Figure 15 shows the effect of each reaction on the Mg content and spinel formation in experiment 1. For the Mg content, the effects of the reactions between steel/slag and slag/refractory were large. For the spinel formation, the effect of the reaction between slag/refractory was the largest, and its contribution ratio was more than 70%. The reaction between steel/refractory had a minor effect on the Mg content and spinel formation. Figure 16 shows the effect of each reaction on the increase in Mg content and spinel formation in experiment 2. The effect of the reaction between molten steel/slag on the Mg content and spinel formation was the largest, and the contribution ratio was almost 99%. The reaction between molten steel/refractory and slag/refractory did not affect the Mg content and spinel formation. The difference of the contribution ratio of slag/refractory among the results of experiments 1 and 2 is caused by the difference of MgO content in the slag at 0 min. Compared with experiment 1, because the MgO content was very high in experiment 2, MgO did not dissolve much from the refractory during experiment 2.

Table 7. Calculation conditions to discuss the effect of each reaction on the Mg content and spinel formation.
Calculation conditions123
(1) Molten steel/SlagConsideredConsideredConsidered
(2) Slag/RefractoryConsideredNot ConsideredConsidered
(3) Molten steel/RefractoryConsideredConsideredNot Considered
Fig. 15.

Effect of each reaction on the Mg content (a) and spinel formation (b) for experiment 1.

Fig. 16.

Effect of each reaction on the Mg content (a) and spinel formation (b) for experiment 2.

From the above discussion, the mechanism for the change in the Mg content and spinel formation in this experimental system can be considered as follows. First, the MgO-type refractory dissolved into the slag phase, and the MgO content in the slag increased if the MgO content before deoxidation was low. Then, the MgO was reduced by Al in the molten steel, and the Mg content in the molten steel increased. Finally, spinel-type inclusions were formed as the deoxidation product.

6. Conclusions

Experiments that simulate the ladle refining process were conducted using an induction furnace, and the compositional changes in the molten steel, slag, and inclusions were investigated. The inclusions in the steel samples were analyzed by P-SEM, which included an automatic analysis system. By the application of the developed simulation model to these experiments, the validity of the model was evaluated. To improve the model, the empirical equation for the reaction between molten steel and refractory was applied. In addition, the ratio of the floatation parameter of each inclusion was varied based on Stokes’ law.

The inclusion composition gradually changed from Al2O3 to MgO·Al2O3 after the addition of Al, and the inclusions originating from slag were also observed at all times. The compositional change of the deoxidation product by the model calculation corresponded to the observed variation of inclusion composition, and the calculated composition of inclusions originating from slag also agreed with the experimental results. The rate of the compositional change increased with increasing Ar gas flow rate, and this tendency was captured by the model well. Therefore, the validity of the developed model is considered to be confirmed.

Then, the effects of reactions between molten steel/slag, slag/refractory, and molten steel/refractory on spinel formation were discussed using the developed model. The results showed that the effect of the reaction between molten steel/slag was the largest, and the effect of the reaction between molten steel/refractory was negligible.

Acknowledgments

The authors appreciate the financial support of the Research Grant of the 19th Committee of Steelmaking, the Japan Society for the Promotion of Science, and the ISIJ Research Promotion Grant of Iron and Steel Institute of Japan.

References
 
© 2014 by The Iron and Steel Institute of Japan
feedback
Top