THE IRSID SLAG MODEL FOR STEELMAKING PROCESS CONTROL J. Lehmann, H. Gaye, P. Rocabois. ABSTRACT The optimisation of high temperature reactions taking place during steel elaboration is improved with a precise evaluation of thermodynamic relationships: these are reactions taking place between the metal and the gaseous phases, between the metal and various slags and refractories, as well as the formation of oxide inclusions and of various precipitates. For oxide systems, the type of thermodynamic information concerns slag phase diagrams and component activities in multicomponent systems. A rational use of this information is thus only possible through thermodynamic models. The IRSID slag model is one of these models. It is based on the Guggenheim’s quasichemical approach and has been validated in large temperature and composition domains belonging to the system: SiO2-TiO2-Ti2O3-Cr2O3-Al2O3-Fe2O3-CrO-FeO-MnO-MgO-CaO and for CaF2 contents up to 10-20 wt% and sulphur contents less than a few percents. It has been incorporated in the multiphase equilibrium calculation code CEQCSI, developed at IRSID for specific applications in the field of iron and steelmaking. Some examples of these applications are given concerning various steel grades. INTRODUCTION. The traditional role of slag in Iron- and Steelmaking processes is to eliminate the gangue of the charged materials, and to refine the metal, as major contaminants of the steel can only be removed via the slag phase (S, P, …). Another function has more recently been fully exploited, in the constant drive for higher purity and better and better controlled steel products: the fine adjustment of steel composition by slag treatment in ladle metallurgy in order to promote an adequate inclusionnary state. In recent years, the optimisation of high temperature metal-slag-refractories reactions in order to improve metal purity, and the control of the precipitation of endogeneous oxide and sulphide inclusions in steels have significantly progressed through the use of thermodynamic models of oxide systems (1). To represent industrial processes, thermodynamic models must describe multicomponent systems, as the metal quality will very often be conditioned by minor components of the slag. They should also be reasonably time efficient since the values of slag properties have to be reactualised a large number of times during the calculation, when used in kinetic reaction models or in multiphase equilibrium codes. This paper gives an overview of the models used at IRSID and particularly of the slag model developed at IRSID which are included in the multiphase equilibrium code CEQCSI also developed at IRSID. Two industrial applications are presented. 89 THERMODYNAMIC MODELLING. The IRSID Slag Model. The slag model, originally conceived for liquid oxide mixtures (2), was later on improved in order to introduce a partial substitution of oxygen by sulphur and fluor (3). The properties of the liquid slag are described with a two sub-lattices statistical thermodynamics model. The anionic sub-lattice contains O2-, S2- and (2F)2- ions, and the cationic sub-lattice is filled with the different cations in decreasing order of their charge. The structure is then described in terms of cells composed of a central oxygen (or sulphur, or fluor) surrounded by two cations and noted (i-A-j). The degeneracy of the system is then written by using the Guggenheim’s quasichemical approach (4). For oxide systems, two types of parameters are used to evaluate the energy of the system : - the energies for the formation of asymmetric cells from the corresponding symmetrical ones. For a mixture of oxides only, these energies are noted WijO and correspond to the reaction: ½(i-O-i) + ½(j-O-j)(i-O-j) - the interaction energies between cells, corresponding to one parameter per couple of cations. For a mixture of oxides only, these parameters are noted E Oij and represent the interaction between a symmetrical cell (i-O-i) and an asymmetric one (i-O-j). Thus, only a limited number of binary parameters (two per binary system) are used to describe multicomponent systems, and for the systems investigated so far, they have been assumed to be temperature independent. To the liquid phase model are associated the values of free energy of formation of the various stoichiometric compounds that can be formed at lower temperature, as well as solution models for various solid solutions (spinels, olivine, …). Some of these values have been adapted in order to obtain a best fit of the phase diagrams, while remaining within the range of uncertainty of the experimentally determined thermochemical data. In its present state, the slag model can be used for calculations in the system SiO 2-TiO2Ti2O3-Cr2O3-Al2O3-Fe2O3-CrO-FeO-MgO-MnO-CaO-CaF2-S. More recently, an extension of this model to oxisulphide systems (5) with complete interchange of oxygen and sulphur has been proposed to cover the full range from pure oxides to pure sulphides. The structure is still modelled by the two anionic and cationic sublattices but the cells no more only describe the cationic surroundings of the anions but also their anionic surroundings. Thus a cell contains a central anion, two surrounding cations and six surrounding anions. The parameters of the new model are similar to those of the preceding one. There are only some additional parameters related to the exchange of anions on the anionic shell of the cells. These parameters have been calibrated by using the available data on phase diagrams and sulphide capacities. At this time, the model has been consolidated for the system SiO2-Al2O3-Fe2O3-FeO-MgO-MnO-CaO-S. An example of the accuracy of the 90 model is given by Fig. 1 which compares the experimental and calculated MnS solubility in SiO2-MnO liquid oxides. %MnS model 70 Hasegawa et al. 60 50 40 30 20 10 0 0 20 40 60 80 100 % MnO/(%MnO+%SiO2) Fig. 1: Comparison between the MnS solubility at 1350°C calculated et experimental (6) as a function of the ratio %MnO/(%MnO + %SiO2). Metallic Phases. For the most frequent applications, the properties of the metallic phases are described by the interaction coefficients formalism. This formalism uses the weight percentages as composition variables and the activity of the element i is given by: a i f i .(%i) with log 10 (f i ) e ij .%j j when j represents the different solutes. The interaction coefficients e ij can depend on the temperature. For concentrated solutions, second order coefficients are used: log 10 (f i ) e ij .%j rij, k .%j . %k j j, k Another formalism which is used for some applications, mainly for concentrated or segregated solutions, is the formalism of the sublattice solutions developed by the Royal Institute of Stockholm (7) and used by the Thermo-Calc software. In this formalism, the elements are distributed on one or two sublattices. For instance, the ferrite or austenite structure is described by two sublattices, the first sublattice is occupied by the metallic elements Fe, Mn, Si… and the second one by the non metallic elements C, N and by vacancies. In the liquid metal, on the contrary, all elements are mixed on the same sublattice. The mixing free enthalpy is then written as a function of the site occupancy fractions, ratios of the mole number of a given element on a given site to the mole number of the site. 91 Solid solutions. The nature of the stoichiometric compounds or solid solutions of interest is very diverse: oxides, sulphides, nitrides, carbides, carbonitrides,... The data concerning the thermodynamic properties of the stoichiometric compounds can be extracted from general databanks (SGTE for instance) or are extracted from experiments of equilibrium with metallic phases. They are then expressed in terms of solubility products representing the free enthalpy of formation from elements dissolved in the metal phase. The solid solution models utilised are also very diverse: ideal solutions, regular, subregular solutions, sublattice solutions. If, at high temperatures, the deviations from ideality are usually small, for application at lower temperatures, in the solid metal, they have to be taken into account THE CEQCSI SOFTWARE. The multiphase equilibrium calculation code CEQCSI (8) has been developed at IRSID for very specific applications in the field of Iron and Steelmaking. It is well adapted to the following situations : - calculation of slag crystallisation path in order to determine, for instance, the evolution of its liquid fraction as a function of temperature, - calculation of slag-metal equilibrium in order to determine, for instance, the transfer from the slag to the metal of trace elements Ca, Mg, and in some grades Al, at the level of a few ppm to fractions of ppm, that may influence the nature of inclusions present in the product, -calculation of inclusions precipitation in order to determine, from the overall steel analysis, the amount, composition and origin (ladle treatment, steel cooling and solidification, solid metal) of endogeneous inclusions and precipitates. In this case, the sequence of precipitation of inclusions is calculated by temperature steps from the temperature of liquid steel treatment, to subsequent temperature evolution, during cooling and solidification of the steel. In this situation, an original method has been developed (9) in which the microsegregation equations for elements dissolved in liquid metal, with or without diffusion in solid metal, and the equilibrium conditions between liquid steel and oxide, sulphide, nitride, or carbide precipitates are treated simultaneously. The calculation also gives the liquidus temperature of the oxide inclusions and their crystallisation path during cooling. INDUSTRIAL APPLICATIONS: INCLUSIONS CONTROL IN STEEL. Calcium treatment of Al-killed steels. Calcium additions are widely used in steelmaking to control the morphology of oxide inclusions and sulphide precipitates. In the present application (steel of composition 0.8 %C0.65 %Mn-0.2 %Si-0.13 %Cr with Al contents varying between 150 and 650 ppm), the treatment is performed in order to transform alumina inclusions into liquid calcium aluminates, in order to prevent nozzle clogging during continuous casting. During the treatment, only a small part (about 15 to 20%) of added calcium remains in the metal, the rest being lost by vaporisation. The calcium remaining in the metal at treatment temperature consists of : 92 - Ca contained in oxide inclusions. If this amount is too small, solid aluminates are formed (CaO.6Al2O3, CaO.2Al2O3, CaO.Al2O3) which are even more detrimental than alumina inclusions with regard to nozzle clogging, - Ca dissolved in liquid metal, this amount being larger at higher Al content. Most of this calcium will form precipitates of CaS or (Ca,Mn)S during steel solidification, - possibly Ca contained in calcium sulphide in the liquid steel if the CaS saturation limit is reached. In this case, the CaS precipitates formed in liquid metal can also contribute to nozzle clogging, and the amount of residual calcium may be too small to form liquid aluminates. Figures 2 and 3 summarise, for Al contents of 650 and 150 ppm, respectively, the calculated distribution of calcium among these various phases, as a function of total Ca and O contents fixed in the metal at the end of the calcium treatment. On these diagrams, the heavy lines represent the limits of existence of the various oxides, that is, for increasing calcium content at a given oxygen content, solid aluminates in the domain labelled "Solid Oxides", mixture of CaO.Al2O3 and liquid oxides (from 0 to 100 % liquid) in the domain labelled "CA + liq", and liquid oxides of increasing CaO content in the domain labelled "Liquid Oxides". The dotted lines indexed with S contents indicate the calcium content for which calcium sulphide saturation is reached for the given S content. In the domains "CA + liq", the S content for CaS saturation is practically constant, 90 ppm S for 650 ppm Al and 450 ppm S for 150 ppm Al. Ca total (ppm) 60 650 ppm Al 1545°C 80 ppm S } 90 ppm S Liquid Oxides 40 CA+liq 150 ppm S 20 Solid Oxides 0 0 10 20 30 O total (ppm) 40 50 Fig. 2: Nature of oxide phases formed as a result of Ca treatment of a steel grade containing 650 ppm Al, as a function of Ca and O contents. The lines indexed with S content indicate the minimal value of Ca content leading to CaS precipitation in liquid steels, for the given S content. 93 Ca total (ppm) 60 150 ppm Al 1545°C 150 ppm S Liquid Oxides 40 } 450 ppm S CA+liq 20 Solid Oxides 0 0 10 20 30 O total (ppm) 40 50 Fig. 3: Nature of oxide phases formed as a result of Ca treatment of a steel grade containing 150 ppm Al, as a function of Ca and O contents. The lines indexed with S content indicate the minimal value of Ca content leading to CaS precipitation in liquid steels, for the given S content. These diagrams indicate the windows available for a proper transformation of alumina inclusions into fully liquid aluminates without CaS precipitation in liquid steel. It is apparent that for the steel grade with 650 ppm Al, this window is completely closed for an S content as low as 90 ppm, whereas for the steel grade with 150 ppm Al, an adequate transformation of oxides can be obtained without CaS precipitation for much higher S contents. Inclusions in free cutting steels. In these steel grades with high S contents, two main objectives are sought : - obtain a good distribution of MnS precipitates to limit the chips length, - avoid the contact of the tool with hard abrasive oxides (Al2O3, spinels ...) by avoiding their formation or embedding them in sulphides. In addition, for high speed machining, it is favourable to harden some of the sulphides that will then form protective films on the tool. One way to do this is to add calcium to form (Ca,Mn)S, another way is to obtain oxisulphides in which the hardening effect is created by the partial substitution of S by O. This last solution is applied for free cutting steels. Calculations were made to predict the nature of oxide, oxisulphide and sulphide phases precipitating in liquid steel and during solidification, in a base steel of composition 0.07 %C1.35 %Mn-0.014 %Si containing minute amounts of tramp elements Al, Ca and Mg resulting from the ladle treatment, as a function of oxygen content (20 to 120 ppm). For this application, the extension of the slag model to the oxisuphides has been used. The calculated sequence of inclusions precipitation for an oxygen content of 60 ppm is indicated in Fig. 4 (oxygen distribution in the various precipitates and sulphur distribution in the various precipitates). 94 a) O (ppm) 60 b) Metal liquid fraction 1 50 Oxisulphides 0.8 40 0.6 Liquid fraction S (ppm) Metal liquid fraction 1 3000 Sulphides 2000 0.6 Liquid fraction 30 0.4 0.4 20 0.8 1000 Oxides 0.2 0.2 10 Oxisulphides 0 1550 1530 0 1550 0 1470 1510 1490 Temperature °C 1530 1510 1490 Temperature °C 0 1470 Fig. 4: Computed equilibrium sequence of precipitation (a) of oxide and oxisulphide and (b) of oxisulphide and sulphide inclusions during steel solidification (expressed respectively in terms of ppm oxygen and sulphur fixed in these inclusions) in a free cutting steel grade containing 60 ppm O. A summary of the effect of oxygen content on the amount of the various oxide, oxisulphide and sulphide phases is presented in Fig. 5. It appears that the optimal oxygen content is around 60 ppm, as the amount of oxisulphides is large, and the amount of harmful oxides is minimised. This oxygen range has indeed shown the best free cutting properties in industrial casts. a) b) O (ppm) 140 S (ppm) Oxisulphides 4000 Solid sulphides 120 3000 100 Oxisulphides Liquid oxides 80 2000 60 40 Solid oxides Solid oxides 1000 20 Liquid oxides 0 20 30 40 60 80 O total 100 120 0 20 30 40 60 80 O total 100 120 Fig. 5: Computed distribution of (a) oxygen among oxide and oxisulphide inclusions and (b) sulphur among oxisulphide and sulphide inclusions, as a function of total oxygen content, in a free cutting steel. CONCLUSIONS. At this time, the available thermodynamic data and models allow a good description of the reactions concerning the liquid metal (slag/metal equilibrium and inclusion precipitation during cooling and solidification). Future efforts will be more specifically focused on the precipitation in the solid metal. In this domain, the free enthalpy of formation and the mixing properties of some precipitates remain very poorly known. 95 REFERENCES. 1. Gaye H., Gatellier C. and Riboud P.V.: 'Control of Endogeneous Inclusions in Al-killed and low-Al Steels', Proceedings of the Ethem T. Turkdogan Symposium, The Iron and Steel Society, Toronto, 1994. 2. Gaye H., Welfringer J.: 'Modelling of the Thermodynamic Properties of Complex Metallurgical Slags' Proc. Second International Symposium on Metallurgical Slags and Fluxes, Warrendale, PA, Eds. H.A. Fine and D.R. Gaskell, 1984. 3. Gaye H, Lehmann J, Matsumiya T, Yamada W.: ' A Statistical Thermodynamics Model of Slags: Applications to Systems containing S, F, P2O5 and Cr Oxides' Proc 4th Int Conf on Molten Slags and Fluxes, ISIJ, 1992. 4. Guggenheim E.A.: Mixtures. Oxford, Clarendon Press, 1952. 5. Gaye H., Lehmann J.: 'Modelling of Slag Thermodynamic Properties. From Oxides to Oxisulphides', Proceedings of the 5th International Conference on Molten Slags, Fluxes and Salts, , The Iron & Steel Society, Sydney, 1997. 6. Hasegawa A., Morita K. and Sano N.: 'Phase equilibria for the MnO-SiO2-MnS slag system'. Tetsu-to-Hagane 1995 81 1109-13. 7. Hillert M.: 'Thermodynamic modelling of Phase Diagrams. A call for increased generality in "Computer Modelling of Phase Diagrams' Proceedings of a symposium held at the Fall Meeting of The Metallurgical Society, Toronto, Canada, 1985. 8. Gatellier C., Gaye H., Lehmann J., Zbaczyniak Y.: 'Des outils thermodynamiques pour la maîtrise des réactions métal-laitier et le contrôle inclusionnaire des aciers'. Revue de Métallurgie-CIT 1992 887-8. 9. Wintz M., Bobadilla M., Lehmann J., Gaye H.: 'Experimental study and modelling of the precipitation of non-metallic inclusions during solidification of steel'. ISIJ Int. 1995 35 71522. 96