Journal Information
Vol. 8. Issue 4.
Pages 3453-3458 (July - August 2019)
Share
Share
Download PDF
More article options
Visits
402
Vol. 8. Issue 4.
Pages 3453-3458 (July - August 2019)
Original Article
DOI: 10.1016/j.jmrt.2019.06.012
Open Access
Evaluating slag–metal deoxidation equilibrium in secondary steelmaking
Visits
402
Viviane Lima Freitas da Silva, Luciano da Silva Leitão, Erik de Aguiar Santos, Antônio Carlos Sant’Anna da Silva, André Luiz Vasconcellos da Costa e Silva
Corresponding author
andre@metal.eeimvr.uff.br

Corresponding author.
EEIMVR-UFF, Av. dos Trabalhadores, 420, 27255-125, Volta Redonda, RJ, Brazil.
This item has received
402
Visits

Under a Creative Commons license
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Figures (8)
Show moreShow less
Tables (2)
Table 1. Steelmaking processing information and sampling timeline.
Table 2. Slag and steel compositions used in the experiments.
Show moreShow less
Abstract

Steel–slag reactions are important in steelmaking, in special in secondary steelmaking. Currently, desulfurization, deoxidation and inclusion removal are mostly performed during ladle refining. The extent to which these reactions proceed is governed by the initial state of the material reaching the ladle, the additions made to the ladle and by undesired interactions such as reoxidation and steel-refractory reactions. In this work, the extent to which equilibrium is approached for the different deoxidation reactions is Al-killed and Si–Mn deoxidized steel is evaluated using thermodynamics. These results are compared to measurements performed in industrial ladle furnaces. The comparisons make possible evaluating the efficiency of the deoxidation process. The results indicate that reoxidation by air caused by agitation probably play a key role on the ability to approach equilibrium, reaching low oxygen content in steel. It is also concluded that relatively long treatments with controlled agitation are needed to achieve steel–slag equilibrium.

Keywords:
Ladle metallurgy
Non-metallic inclusions
Reoxidation
Computational
Thermodynamics
Steel
Full Text
1Introduction

Steel–slag reactions are of great importance in steelmaking. Many secondary steelmaking processes focus on these reactions. Currently, desulfurization, deoxidation and inclusion removal are mostly performed during refining in ladle furnace (LF) [1]. The extent to which these reactions proceed, and approach equilibrium depend on the initial state of the material reaching the ladle, the additions made to form slag and adjust the chemical composition of the steel and by undesired interactions such as reoxidation and steel-refractory reactions.

Under some LF processing conditions deoxidation equilibrium is achieved or closely approached. This is not a general rule, however. When equilibrium is attained, it is relatively easy to calculate the expected steel composition, in special if one has access to computational thermodynamics tools. If, however, equilibrium is not reached, the final state depends on many process variables and, in principle, a complex kinetic model would be needed. Process evaluation and improvement can only be achieved after correct understanding. Thus, in this work, a thermodynamic evaluation of the state of deoxidation reactions is performed in two industrial LFs. One producing Al-killed steel and the other Si–Mn deoxidized steel. This is done by evaluating the individual metal-oxide equilibria and comparing the results to the actual oxygen content in the steel. Results are discussed with the aim of improving the deoxidation practice and the LF process.

1.1Deoxidation thermodynamics and kinetics

The thermodynamics of deoxidation of steel has been well reviewed in many publications (e.g. Refs. [2,3]). Although there are still some doubts concerning thermodynamic data for Al [4], Ca and Mg [5], in special, the present knowledge is sufficiently good to perform predictions in the range of commercial steels. It is generally agreed that the deoxidations reactions involving Al and Si are not limited by nucleation and that the oxygen in solution in steel approaches equilibrium with the dissolved deoxidants very rapidly [6], at least when the pure oxide is formed. If, however, equilibrium must be attained with an oxide that is not pure, but dissolved in the slag, kinetics may play an important role (e.g. Refs. [7–10]).

While ladle stirring is critical for the kinetics of the metal-slag reactions, it is well established that control must be exercised to avoid reoxidation and slag entrainment [6,11]. Comparing the different calculated deoxidation equilibria to real values in the ladle is a known practice to evaluate the approach to equilibrium. For steel processed in LF under ideal conditions of agitation and with due care to prevent reoxidation, it has been observed that the final composition is very close to equilibrium (e.g. Refs. [10,12,13]). In this case, the equilibrium of all possible oxidation reactions should prevail and the single value of oxygen in solution should satisfy all the calculated equilibria, as indicated in Fig. 1. If processing time is short, agitation is inadequate (either too weak or too strong [6]) and if reoxidation is not properly controlled, this condition, however, is not fulfilled. In this case, the observed values of the chemical composition of the steel and of the slag will indicate how the process is evolving. This has been demonstrated, for instance, by Ref. [14], who tried to understand the reasons for the observed deviations. Due to the complexity of the problem, efficient kinetic models for the LF are still being developed (e.g. Refs. [7,15]).

Fig. 1.

When metal-slag equilibrium prevails in the ladle, the soluble oxygen content is such that it satisfies all metal-oxide equilibria [16]. Here μOsteel is the oxygen chemical potential in the steel, directly related to the oxygen content in solution in the steel.

(0.13MB).
2Experimental

Heats made in an oxygen converter (BOF) and in an electric arc furnace (EAF) were processed in ladle furnaces without vacuum protection. During LF processing slag and metal were sampled at defined points of the process. Temperature and oxygen dissolved in steel were measured using commercial, disposable electrolytic cell probes.

Experiments in BOF steel were performed in heats of a low-sulfur HSLA steel containing around 1% Mn and 0.1% C that was aluminum deoxidized and calcium treated. All sampling was done in the LF before the calcium treatment. Experiments performed in EAF steel were performed in heats of an engineering bar steel containing around 0.15% C. The steel was silicon-manganese deoxidized. Sampling was performed in the LF and in the tundish. Tables 1 and 2 give more details about the steels and slags used in these experiments the steelmaking processing, and sampling timeline.

Table 1.

Steelmaking processing information and sampling timeline.

Process  Heat size (t)  T range in ladle (oC)  Argon stirring Nm3/h t  Sampled heats  Sampling times, starting at arrival at LF (min) 
BOF-LF  220  1570–1590  0.11–0.48  17  0,15,25, release form LF (45–60 min) 
EAF-LF  98  1550–1600  0.18–0.42  14  0, release from LF (14–16 min) 
          Arrival at caster (Release + 15 min) 
Table 2.

Slag and steel compositions used in the experiments.

ProcessLadle slag composition (oxides)Steel composition
CaO  MgO  SiO2  Al2O3  FeO + MnO  Mn  Si  Al 
BOF-LF  35–57  8–12  7–21  8–27  1.7–27  ˜0.1  0.8–1.2  0.02–0.09  0.0–0.06 
EAF-LF  30–50  7–12  10–20  1–4  0.9–15.5  ˜0.15  0.4–1.2  0.09-0.19  – 

Computational thermodynamics calculations were performed using Thermo-calc [17] in conjunction with two thermodynamic databases: SLAG3 database [18], originally derived from the IRSID model, using the Kapoor–Frohberg–Gaye cell model [19,20] and TCOX7 [21], using the ionic liquid model developed at KTH [22,23]. The application of these databases to steelmaking is described and discussed elsewhere [12,20].

3Evaluating the deviation from metal-slag equilibrium

In order to evaluate the deviation from equilibrium, the following strategy was adopted. Using the composition of the slag and temperature measured in the LF the activities of the oxides in the slag were calculated using computational thermodynamics. In the case of aluminum deoxidation, the alumina activity was also calculated using Ohta and Suito’s empirical formulae [24,25]. The agreement of the activities calculated using the two databases was reasonable, as shown in Fig. 2. The agreement of the results of Ohta and Suito’s formulas with those calculated using computational thermodynamics was also reasonable (see Fig. 3).

Fig. 2.

Comparison of the calculated activities of oxides in slag in LF using two different databases and Thermo-calc. Reference states are pure oxides at the calculation temperature, and 1 atm.

(0.1MB).
Fig. 3.

Calculated values of oxygen concentration in equilibrium with different alumina activities. Pure alumina, alumina activity calculated according to Ohta and Suito’s formula and alumina activity calculated using computational thermodynamics. Dashed line shows measured values for each heat.

(0.09MB).

For each applicable oxide-metal reaction (Eqs. 1–4), equilibrium was calculated considering the metal activity determined from the steel chemical analysis (Al, Si, Mn and Fe) and the oxide activities, to determine the oxygen content in equilibrium with these conditions. If equilibrium had been achieved, all calculations should result in the same value for the oxygen content dissolved in the steel, which should agree with the experimental measurements. The difference between these values indicate the extent of deviation from metal–slag equilibrium.

This can also be expressed by recalling, from Fig. 1, that equilibrium must fulfill Eq. 5, since there is only one value for the dissolved oxygen content in the steel.

Thus, when the calculated values of oxygen in equilibrium with each metal and oxide are compared among each other and with the measured oxygen content, an estimate of the deviation from equilibrium is obtained as well as insights into the effectiveness of the deoxidation process.

3.1Aluminum killed steel

In the case of aluminum killed steels, the oxygen content that would be in equilibrium with the elements and the oxides (according to Reactions 1–4) were calculated for Al, Mn and Fe. As observed in previous work [26] in the same LF, the deviation between the measured oxygen and the Fe/FeO equilibrium was large. For this reason, the comparison between these calculations and the measured values is not presented in the following figures.

Fig. 3 presents the comparison of the measured oxygen content with the oxygen calculated according to the Al/Al2O3 equilibrium. This calculation was performed for three different values of alumina activity:

(a) Pure alumina, (aAl2O3=1) which is the condition prevailing right after the aluminum addition to the liquid metal,

(b) Alumina activity calculated based on the LF slag composition at the sampling moments, using Ohta and Suito’s formula and,

(c) Alumina activity based on the same slag composition and temperature, calculated using SLAG3 and Thermo-calc.

The results in Fig. 3 show that the oxygen content in the steel is lower than that which would be in equilibrium with pure alumina. This indicates that the dissolution of alumina and lowering of its activity is having some effect on the aluminum deoxidation. However, the calculated oxygen in equilibrium with conditions (b) and (c) above are lower than the measured oxygen content. The apparent equilibrium occurs with aAl2O3=0.60±0.17.

Fig. 4 presents the oxygen content in equilibrium with Mn/MnO. There is no equilibrium between Mn and MnO, the slag content of MnO being substantially higher than that which would be in equilibrium.

Fig. 4.

Calculated values of oxygen concentration in equilibrium with Mn and MnO. Oxide activity calculated using computational thermodynamics. Dashed line shows measured values for each heat.

(0.09MB).

Among the oxides considered in the calculations, MnO and FeO are part of the passing slag from the BOF taping operation. In order to follow the evolution of MnO and FeO, their values during the LF treatment were compared. The values of MnO and FeO along the treatment in the LF are presented in Fig. 5.

Fig. 5.

Evolution of FeO and MnO during the treatment in the LF for various Al-killed heats. Samples from the same heat are on a single line. Arrow indicates evolution with time. Dashed line indicates equilibrium FeO/MnO relation proposed by Turkdogan [3].

(0.11MB).

Most heats consistently approach lower values of FeO and MnO and the MnO/FeO equilibrium relationship. At least in one heat, the approach is interrupted during the process. The final contents, however, are still high for the level of deoxidation of the steel, as seen in Fig. 4.

The results in Figs. 3–5 indicate that aluminum is indeed the element controlling the lowering of the soluble oxygen in the process. Nonetheless, equilibrium is not attained. This could be caused either by reoxidation of the liquid metal by the atmosphere (“open eye” due to excessive agitation) or by the difficulty in lowering the FeO and MnO contents in the slag.

3.2Silicon manganese killed steel

In the case of silicon manganese killed steels, the oxygen content that would be in equilibrium with the elements and the oxides (according to Reactions 1–4) were calculated for Si, Mn and Fe (Fig. 6).

Fig. 6.

Calculated values of oxygen concentration in equilibrium with Si and silica. Silica activity calculated using computational thermodynamics. Dashed line shows measured values for each heat.

(0.09MB).

Fig. 5 presents the comparison of the measured oxygen content with the oxygen calculated according to the Si/SiO2 equilibrium. As the silica activity in the slag is very low, silicon becomes an excellent deoxidant.

However, the actual oxygen values achieved in the LF deoxidation process are significantly higher than those calculated from the Si/SiO2 equilibrium.

The values for the Fe/FeO and Mn/MnO calculated equilibria are presented in Fig. 7.

Fig. 7.

Calculated values of oxygen concentration in equilibrium between Mn and MnO and Fe and FeO. Oxide activities calculated using computational thermodynamics. Dashed line shows measured values for each heat.

(0.09MB).

The expected values for oxygen in equilibrium with Mn/MnO and with Fe/FeO are of the same order of magnitude as the experimental values of oxygen content. Furthermore, evaluating the evolution of the content of MnO and FeO in the LF slag (Fig. 8) indicates that the heats approach the FeO/MnO equilibrium and reach lower values than those shown in Fig. 5, for Al-killed steel. This may be due to the strong slag deoxidation performed with CaC2 during tapping from EAF, in these Si–Mn steels. Direct slag deoxidation seems to be quite effective in lowering its content of the less stable oxides.

Fig. 8.

Evolution of FeO and MnO during the treatment in the LF for various Si–Mn deoxidized heats, using CaC2 additions during tapping of the EAF. Samples from the same heat are on a single line. Arrow indicates evolution with time. Dashed line indicates equilibrium FeO/MnO relation proposed by Turkdogan [3].

(0.11MB).
3.3Summary of the observations

In both cases (Al-killed and Si–Mn deoxidation), the equilibrium oxygen contents would be significantly lower than the values measured at the end of the LF process. In the case of Al–killed steels, much lower soluble oxygen contents were reached than in the case of Si–Mn steels. This is reasonable as the equilibrium with a lower oxide activity (in the slag) is not critical to an effective deoxidation with Al. The measured values are in apparent equilibrium with an alumina activity of aAl2O3=0.60±0.17, between 1 (no effect of the slag) and the calculated value for the slag composition.

In the case of the Si–Mn deoxidized steel, the measured soluble oxygen content deviates significantly from what would be expected. It is, however, much lower than the value expected if equilibrium was established with pure silica, that would be in the range of 120–140 ppm.

There are two plausible explanations to the observed phenomena. First, it could be postulated that the processing time was not enough to achieve metal–slag equilibrium. However, during the processing of the Al-killed heats, for instance, many heats required further additions of aluminum to keep this concentration within the specified range. Furthermore, times as long as 30–40 min were used in some Al-deoxidized heats, which should approach equilibrium, based on information acquired elsewhere in similar conditions, but under vacuum (e.g. Ref. [6]). This seems to indicate that the reduction of less stable oxides is not effective in the process and/or reoxidation by air is playing a relevant role in the process. While FeO and MnO are reduced during the processing of both steels, it seems that in the processing of the Si–Mn steel, the practice of directly deoxidizing the slag is positive for oxygen control. However, in the processing of both steels, the deviations suggest that reoxidation due to exposure to air may be playing a relevant role. This indicates that the formulation of a kinetic model for the LF process is of paramount importance for the further development of the understanding of the deoxidation–reoxidation process. This understanding will have a significant impact on quality of steel and cost control in the melt shop.

4Conclusions

Contrary to steels processed for longer times and under vacuum, equilibrium between deoxidants in the steel and oxides in the slag was not achieved during the regular ladle furnace processing of Al-killed BOF steel and Si–Mn EAF steels. In the case of Al-killed steels, aluminum clearly controls the oxygen potential in the steel and levels of oxygen close to those expected by equilibrium are achieved. In the case of Si–Mn steels, with somewhat shorter LF processing times, the deviation is quantitatively larger. The use of direct slag deoxidation with CaC2 in the Si–Mn steels was shown to be effective in reducing the content of less stable oxides in the slag but was not enough to make possible slag–metal equilibrium at the end of the process.

Conflicts of interest

The authors declare no conflicts of interest.

Acknowledgements

The authors thank GERDAU and CSN for support during the performance of this work. André Costa e Silva thanks CNPq, CAPES and FAPERJ for support during different stages of this work.

References
[1]
The making, shaping and treating of steel, casting volume, 11th ed., pp. 1000 pp
[2]
H. Gaye.
Inclusion formation in steels.
The making, shaping and treating of steel, casting volume, 11th ed., pp. 65-84
[3]
E. Turkdogan.
Principles of steelmaking.
The Institute of Materials London, (1996), pp. 331 pp
[4]
A. Costa e Silva, F. Beneduce Neto, R.R. Avillez.
Liquid iron deoxidation by aluminum: a brief review of experimental data and thermodynamics description.
Technol Metal Mater Min, 12 (2015), pp. 267-273
[5]
A. Costa e Silva.
Calcium and magnesium thermodynamics in steel and its impacts on secondary steelmaking: a computational thermodynamics approach.
Rev. Metal. — CIT, 105 (2008), pp. 181-193
[6]
A. Costa e Silva.
Non-metallic inclusions in steels — origin and control (review).
J Mater Res Technol, 7 (2018), pp. 283-299
[7]
K.J. Graham, G.A. Irons.
Toward integrated ladle metallurgy control.
Iron Steel Technol, 6 (2009), pp. 164-173
[8]
D.G. Robertson, B. Deo, S. Ohguchi.
Multicomponent mixed-transport-control theory for kinetics of coupled slag/metal and slag/metal/gas reactions: applications to desulphurization of molten iron.
Ironmak Steelmak, 11 (1984), pp. 41-55
[9]
H. Sun, K. Mori.
Oxidation rate of aluminum in molten iron by CaO-SiO2-Al2O3-FeO-MnO slag.
ISIJ Int, 36 (1996), pp. s34-s37
[10]
G. Okuyama, K. Yamaguchi, S. Takeuchi, K. Sorimachi.
Effect of slag composition on the kinetics of formation of Al2O3-MgO inclusions in aluminum killed ferritic stainless steel.
ISIJ Int, 40 (2000), pp. 121-128
[11]
D. Sichen.
Modeling related to secondary steel making.
Steel Res Int, 83 (2012), pp. 825-841
[12]
A. Costa e Silva.
Applications of multicomponent databases to the improvement of steel processing and design.
J Phase Equilib Diffus, 38 (2017), pp. 916-927
[13]
H. Gaye, C. Gatellier, M. Nadif, P. Riboud, J. Saleil, M. Faral.
Slag and inclusions: control in secondary steelmaking.
Clean steel 3, IOM3, London, (1987), pp. 137-147
[14]
J. Björklund, M. Andersson, P. Jönsson.
Equilibrium between slag, steel and inclusions during ladle treatment: comparison with production data.
Ironmak Steelmak, 34 (2007), pp. 312-324
[15]
M.-A. Van Ende, I.-H. Jung.
A kinetic ladle furnace process simulation model: effective equilibrium reaction zone model using factsage macro processing.
Metall Mater Trans B, 48 (2017), pp. 28-36
[16]
A. Costa e Silva.
Controle de inclusões não-metálicas em aços: aspectos termodinâmicos e físico-químicos.
2009 workshop clean steel CARBOOX heraeus-electronite, (2009),
[17]
B. Sundman, B. Jansson, J.O. Andersson.
The Thermo-calc databank system.
CALPHAD, 9 (1985), pp. 153-190
[18]
TCAB.
SLAG3 database.
TCAB, (2012),
[19]
H. Gaye, J. Welfringer.
Modelling of the thermodynamic properties of complex metallurgical slags.
H.A..Fine, D.R.Gaskell. 2nd international symposium on metallurgical slags and fluxes, (1984), pp. 357-375
[20]
A. Costa e Silva.
Thermodynamic aspects of inclusion engineering in steels.
Rare Metals, 25 (2006), pp. 412-419
[21]
TCAB.
TCOX7 database.
TCAB, (2017),
[22]
M. Hillert, B. Jansson, B. Sundman, J. Agren.
A two-sublattice model for molten solutions with different tendency for ionization.
Metall Trans A, 16A (1985), pp. 261-266
[23]
M. Selleby.
Thermodynamic modelling and evaluation of the Ca-Fe-O-Si system. Doctoral dissertation.
KTH Stockholm, (1996), pp. 19
[24]
H. Ohta, H. Suito.
Activities in CaO-SiO2-Al2O3 slags and deoxidation equilibria of Si and Al.
Metall Mater Trans B, 27B (1996), pp. 943-953
[25]
H. Ohta, H. Suito.
Activities in MnO-SiO2-Al2O3 slags and deoxidation equilibria of Mn and Si.
Metall Mater Trans B, 27B (1996), pp. 263-270
[26]
R.P. Batista, A.A. Martins, A. Costa e Silva.
The effects of deoxidation practice on the quality of thin foil low-carbon steel.
J Min Metall Sect B Metall, 53 (2017), pp. 357-363
Journal of Materials Research and Technology

Subscribe to our newsletter

Article options
Tools
Cookies policy
To improve our services and products, we use cookies (own or third parties authorized) to show advertising related to client preferences through the analyses of navigation customer behavior. Continuing navigation will be considered as acceptance of this use. You can change the settings or obtain more information by clicking here.