A Thermodynamic Database for Copper Smelting and Converting SERGEI A. DEGTEROV and ARTHUR D. PELTON The thermodynamic properties of the slag, matte, and liquid copper phases in the Cu-Ca-Fe-Si-O-S system have been critically assessed and optimized over the ranges of compositions of importance to copper smelting/converting based on thermodynamic and phase equilibria information available in the literature and using the modified quasichemical model. A thermodynamic database has been developed, which can be used for the calculation of matte-slag-copper-gas phase equilibria of interest for the production of copper. The model reproduces within experimental error limits all available experimental data on phase diagrams, matte-alloy miscibility gap and tie-lines, enthalpies of mixing, and activities of Cu and S in the matte and liquid alloy. The calculated solubilities of Cu in both S- free slag and slag equilibrated with matte are also in good agreement with experiment under all studied conditions, such as at SiO 2 saturation, in equilibrium with Fe, Cu, or Cu-Au alloys, at fixed oxygen or SO 2 partial pressures and at different contents of CaO in the slag. Sulfide contents (sulfide capacities) of the slags are predicted within experimental error limits from the modified Reddy–Blander model, with no adjustable parameters. As an example of the application of the database, the stability field of matte/slag equilibrium is calculated, and the matte and slag compositions are plotted vs iron to silica ratio in the slag at various SO 2 pressures over this field. The matte-slag two-phase field is limited by the calculated lines corresponding to precipitation of copper, silica, and magnetite. I. INTRODUCTION to be as accurate as or, in certain cases, even more accurate than direct experiments because they are based on experi- THE present study is concerned with the development mental data of good quality obtained for less complex sub- of thermodynamic models and a thermodynamic database systems, which are much easier to study. For example, the of model parameters for the matte, slag, and blister copper difference between the measured losses of copper in the phases in the Cu-Ca-Fe-Si-O-S system. When used with the slag and the calculated copper solubility will represent the Gibbs energy minimization software and other databases of amount of entrained copper. the FACT thermodynamic computing system, these data- Under conditions of interest for copper smelting and con- bases will permit the calculation of matte-slag-copper-gas verting, the concentrations of Ca and Si in the matte and phase equilibria that take place during copper smelting liquid copper alloy are very small. Hence, the presence of and converting. these components was neglected. The oxygen content in the In particular, the copper, sulfur, and magnetite contents matte is reported [3,4,5] to decrease with increasing matte grade of the slag can be accurately calculated over wide ranges (wt pct Cu in the matte). For example, the oxygen concentra- of temperature, iron to silica ratio, percentage of lime, and tion is about 1.1  0.2, 0.8  0.2, and 0.2  0. 1 wt pct oxygen potential. The solubility of iron in the matte as well for 50, 60, and 78 wt pct Cu in the matte, respectively. The as that of sulfur, iron, and oxygen in blister copper can matte grade is normally higher than 50 wt pct Cu in present also be computed. Furthermore, one can calculate liquidus day plant practice. Therefore, the presence of oxygen will temperatures and conditions for precipitation of solid phases have only a small effect on the matte-slag-copper-gas equi- from the slag. All these calculations can be readily used libria considered in the present study. Consequently, oxygen to develop a better understanding of modern smelting and was not taken into account in the matte model. converting processes and to optimize the operating Oxygen impurity in blister copper after converting may conditions. be of some importance, and so oxygen was taken into It should be noted that direct experimental study of the account as a component of the liquid copper model. All gas-matte-slag-copper equilibria is subject to experimental six components of the Cu-Ca-Fe-Si-O-S system were errors because of the complexity of the system and because considered for the slag model. Gold was also incorporated of problems with sampling, quenching, and analysis of the into the liquid copper model in order to treat numerous phases. Entrainment of matte in the slag can also affect the experimental data on matte/slag/Cu-Au-alloy equilibria dis- accuracy of measurements. [1] Taking into account the fact cussed subsequently. that operating conditions of most smelters and converters The modified quasichemical model [6–9] is used to describe are not far from equilibrium, [2] one can expect calculations the thermodynamic properties of slag, matte, and liquid cop- per alloy. Strictly speaking, there is just one liquid phase in this system with miscibility gaps. Under certain conditions, SERGEI A. DEGTEROV, Senior Research Associate, and ARTHUR D. which are far from those used in copper smelting and con- PELTON, Professor of Metallurgy, are with the Centre de Recherche en verting, these phases completely dissolve in each other. In Calcul Thermochimique, E ´ cole Polytechnique de Montre ´ al, Montre ´ al, PQ, the present study, however, slag, matte, and alloy are treated Canada H3C 3A7. Manuscript submitted August 10, 1998. as separate phases. This is sufficient for thermodynamic METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 30B, AUGUST 1999—661