Further study of the electrochemical deoxidation of induction-stirred copper melts

Previous studies by Oberg, Friedman, Boorstein and Rapp on the electrochemical deoxidation of induction-stirred copper melts have been extended by the measurement and interpretation of the pumping current and the emf of an auxiliary galvanic cell to monitor oxygen contents at the pumping surface. Imperfect seals in the high-temperature apparatus allowed the pickup of gaseous oxygen by the melt to counteract partly the electrochemical deoxidation of the melt. The observed pumping current was divided into fractions accounting for this oxygen pickup as well as the oxygen removed from the bulk of the melt. Consistent with the deoxidation mechanism of Oberg,et al, the deoxidation rate is limited by ionic conduction in the zirconia electrolyte crucible at high oxygen contents, and by the mass transport of oxygen in the induction-stirred melt at low oxygen contents. A mass transport coefficient for oxygen was calculated by several methods.     

Further study of the electrochemical deoxidation of induction-stirred copper melts

Previous studies by Oberg, Friedman, Boorstein and Rapp on the electrochemical deoxidation of induction-stirred copper melts have been extended by the measurement and interpretation of the pumping current and the emf of an auxiliary galvanic cell to monitor oxygen contents at the pumping surface. Imperfect seals in the high-temperature apparatus allowed the pickup of gaseous oxygen by the melt to counteract partly the electrochemical deoxidation of the melt. The observed pumping current was divided into fractions accounting for this oxygen pickup as well as the oxygen removed from the bulk of the melt. Consistent with the deoxidation mechanism of Oberg,et al, the deoxidation rate is limited by ionic conduction in the zirconia electrolyte crucible at high oxygen contents, and by the mass transport of oxygen in the induction-stirred melt at low oxygen contents. A mass transport coefficient for oxygen was calculated by several methods.     

Deoxidation of molten copper with a rotating graphite cylinder

The kinetics of deoxidation of molten copper by the “vacuum-suction degassing” (VSD) method is investigated. The molten copper is deoxidized by a rotating porous graphite tube immersed in the copper bath. The inside space of the porous graphite tube is evacuated so that the CO gas formed at the graphite-metal interface is removed through the tube wall. The experimental results suggest that the mass transfer of oxygen in the metal phase controls the reaction rate. The kinetic data are arranged with a first-order rate equation. At (ppm O)≥10, the rate constant increases by decreasing the porosity of the graphite and increasing the thickness of the tube wall. This result suggests that the suction of CO gas weakens CO bubble stirring and, thereby, the mass transfer at the tube-melt interface. However, when the rate of CO suction becomes comparable to or larger than the CO gas evolution rate, the effect of CO stirring becomes negligible. This situation appears under the conditions of high porosity and large wall thickness at (ppm O) ≥ 10. At the low oxygen concentration range of (ppm O) ≤ 4, the effect of CO stirring becomes negligible, regardless of the CO suction condition, because of the considerably low CO formation rate. The achievement of deoxidation by the VSD method is evaluated in connection with the final oxygen concentration.     

Kinetics of deoxidation of liquid copper by graphite particles during submerged injection

Liquid copper can be deoxidized by submerged injection of inert gas in the presence of graphite particles. This paper describes the results of experiments in which approximately 20 kg copper melts were deoxidized by injection of N₂, CO₂, and air in the presence of low sulfur graphite particles. The apparent kinetics of the deoxidation process are first order with respect to the concentration of dissolved oxygen, and the concentration of oxygen in the melt could be reduced to less than 10 ppm by weight after less than 30 minutes of injection. The kinetics are consistent with mixed rate control with both mass transfer and chemical reaction rate affecting the rate of deoxidation. In these experiments, the rate of deoxidation under a graphite covering was slower when particles were injected with the gas stream than when gas was injected alone, although this result may have been influenced by the small size of the melt. Y. W. Chang, formerly Graduate Student with the Department of Civil Engineering, Mechanics, and Metallurgy, University of Illinois at Chicago This paper is based on a presentation made in the T.B. King Memorial Symposium on “Physical Chemistry in Metals Processing” presented at the Annual Meeting of The Metallurgical Society, Denver, CO, February, 1987, under the auspices of the Physical Chemistry Committee and the PTD/ISS.     

Electrochemical deoxidation of induction-stirred copper melts

Induction-stirred melts of copper contained in zirconia-base electrolyte crucibles were deoxidized by electrochemical pumping of oxygen. The rate-controlling step for the deoxidation reaction was found to be transport of oxygen in the electrolyte in the higher oxygen concentration ranges, and transport across a boundary layer in the melt at the melt/electrolyte interface for the lower concentration ranges. The mass transfer coefficient for transport of oxygen across the boundary layer at the melt/crucible interface was determined to be αco^u = 1.7 . exp (-12,900/äT) cm/s The optimization of the experimental parameters for this new deoxidation method are discussed.     

Mass transfer between a horizontal,submerged gas jet and a liquid

The rate of reaction between a horizontal, submerged gas jet and a liquid has been measured in a model system under conditions where mass transfer in the gas phase is rate limiting. The gas was 1 pct SO₂ in air, and the liquid was a 0.3 pct solution of hydrogen peroxide in water. SO₂ absorption rates were measured as a function of jet Reynolds number (10,000 < N_Re < 40,000) and jet orifice diameter (0.238 < d₀ < 0.476 cm). The product of the gas phase mass transfer coefficient and the interfacial area per unit length of jet trajectory, k_SO2 α was found to increase linearly with increasing Reynolds number and to be a strong function of the orifice diameter. The ratio of k_so2 α to volumetric gas flow rate was shown to be independent of Reynolds number for a given orifice diameter. Extrapolated values of k_so2 α are lower than the coefficients measured for vertical CO jets blown upward through liquid copper. Extrapolation of the measured mass transfer data to the jet conditions in copper matte converting and in the gaseous deoxidation of copper has indicated that the gas utilization efficiencies in these processes should approach 100 pct if gas phase mass transport is rate controlling.     

Desulfurization and deoxidation of Cu-S-O alloy in induction melting and solidification under argon and their rates of elimination in vacuum induction melting

Solid Cu-S-O alloy (about 150 g) containing 0.05 pct S and 0.04 pct O (by weight) was heated to 1473 or 1573 K under argon at 6.5 to 101 kPa. Desulfurization of 64 to 80 pct and deoxidation of 75 to 82 pct were observed. The extent did not depend on argon pressure, and this desulfurization and deoxidation probably occurred during melting. In the course of solidification and remelting of molten Cu-S-O alloys containing sulfur and oxygen at approximately the same concentration,i.e., 0.027 pct S and 0.026 pct O, and 0.0456 pct S and 0.0416 pct O under argon at 1.33 to 102 kPa, the percentage of sulfur removed was 49 to 75, the percentage of oxygen removed was 63 to 81, and the extent did not depend on argon pressure. The alloys containing more oxygen than sulfur before cooling,i.e., 0.0216 pct S and 0.081 pct O and 0.0185 pct S and 0.15 pct O, 73 to 90 pct of the sulfur was eliminated. Most desulfurization and deoxidation probably occurred in the course of solidification, because vigorous boiling was observed at this stage. The rates of removal of oxygen and sulfur from molten Cu-S-O alloys maintained at 1373, 1473, and 1573 K under vacuum were expressed by a first-order rate equation, and the dependence of their rate constants on the reciprocal of temperature was determined. In alloys in which pct S ≃pct O before evacuation and pct S ranges from 0.0202 to 0.042 and pct O ranges from 0.023 to 0.048, 79 to 96 pct desulfurization and 66 to 87 pct deoxidation were observed. For the alloys with pct O > pct S initially,i.e., 0.081 pct O and 0.0216 pct S, and 0.15 pct O and 0.0176 pct S, 85 to 94 pct desulfurization was observed and was close to the observed levels of desulfurization in solidification and remelting of the alloys with pct O > pct S under argon.     

Quantitative evaluation of inclusion in deoxidation of Fe-10 mass pct Ni alloy with Si, Ti, Al, Zr, and Ce

The composition, size, number, and morphology of the inclusions in deoxidation of Fe-10 mass pct Ni alloys with 0.2 mass pct M (M=Si, Ti, Al, Zr, and Ce) containing mostly 60 to 130 mass ppm oxygen were studied as a function of holding time at 1873 K. It was found that the initial primary inclusions contained FeO and the FeO content decreased with an increase of deoxidation power and holding time. The mean spatial diameter of inclusions tends to increase with increasing holding time, but did not show the systematic trend with respect to the deoxidation power. The number of particle sections per unit area decreased with decreasing deoxidation power and increasing holding time. In the absence of stirring of melt, the growth rate of the mean spatial diameter of inclusions and the removal rate of particle sections per unit area decreased rapidly with increasing holding time and approached a constant after 10 minutes. The morphology of inclusions was found to be spherical, polyhedral (except Si), and irregular including cluster, and the mean spatial diameter of these inclusions increased with increasing holding time.     

A study of the reaction of CO on liquid iron alloys

In all previous studies of the CO reaction on liquid iron the rate was controlled by liquid phase mass transfer. In this study an isotope exchange technique was used to eliminate liquid phase mass transfer as a possible rate-controlling mechanism. The rate in the present work is over an order of magnitude faster than previously measured. The isotope exchange rate for C¹³O¹⁸ in normal CO on liquid iron was not a function of temperature or sulfur content up to 0.43 pct S, but was a function of gas flow rate, indicating that the rate is controlled by gas phase mass transfer. the mass transfer parameter for the specific experimental condition was determined and the calculated rate for gas phase mass transfer of C¹³O¹⁸ in CO controlling the rate is in agreement with the experimental results. At low temperatures (1523 K) and sulfur contents greater than 0.015 pct the rate may be controlled by mixed control, gas phase mass transfer, and chemical kinetics in series. The chemical rate constant is estimated to be 1.5×10⁻⁵ moles/cm² sec atm for these conditions. S. ANTOLIN, formerly Granduate Student at Carnegie Mellon University     

Influence of surface tension-driven flow on the kinetics of oxygen absorption in molten copper

Absorption measurements have revealed that surface tension-driven flow, the Marangoni effect, can enhance the liquid phase mass transfer coefficient by at least an order of magnitude during the top jetting of molten copper with oxygen. This factor has been established by comparing the mass transfer coefficient for copper to the value measured for molten silver which does not display the surface phenomenon under similar jetting conditions. The spontaneous motion on the copper surface was found to arise under conditions of starvation oxygen transfer in the gas phase in which copper oxide forms in a localized region beneath the lance tip and spreads radially outward at about 30 cm s^-1. The mass transfer coefficient was correlated satisfactorily to unsteady state diffusion theory using measured spreading velocities. Sulfur initially present in the copper bath at concentrations of 0.01 to 0.5 pct and silicon at 20 ppm had no effect on the mass transfer coefficient under the spreading conditions. By virtue of the size of the enhancement factor and the lack of influence of dissolved sulfur and silicon on mass transfer, it is suggested that the Marangoni effect may play an important role in the transfer of oxygen into process baths. R. G. BARTON formerly Graduate Student.     

Thermodynamics of calcium and oxygen in molten titanium and titanium-aluminum alloy

The deoxidation equilibrium of molten titanium and titanium-aluminum alloys saturated with solid CaO has been measured in the temperature range from 1823 to 2023 K. The equilibrium constant of reaction CaO (s)=Ca (mass pct in Ti,Ti-Al)+O (mass pct in Ti,Ti-Al) and the interaction parameter between calcium and oxygen were determined for Ti, TiAl, and TiAl₃. The standard Gibbs energy of reaction for TiAl was obtained as follows: $$\Delta G^\circ = 279,000 - 103TJ/mol$$ The possibilities for the deoxidation of titanium and titanium-aluminum alloys by using calcium-based fluxes are discussed.     

Deoxidation of molten titanium by electron-beam remelting technique

The deoxidation of molten titanium was attempted by aluminum suboxide evolution in an electron-beam remelting furnace. Titanium aluminum alloy was deoxidized to about 0.05 to 0.01 wt pct oxygen in a few minutes by adding aluminum to the melt. Aluminum activity in the alloy was estimated by the evaporation rate of aluminum. formerly Graduate Student, Institute of Industrial Science, The University of Tokyo     

Solidification of the Cu-35wt pct Fe Alloys with Liquid Separation

The microstructures of the Cu-35wt pct Fe alloys were investigated by melt-fluxing in combination with cyclic superheating and melt-spinning technique, respectively. Using the melt-fluxing with cyclic superheating technique, it was found that a complicated sub-microstructure formed in the separated minor phase, when the undercooling was 120 K (120 °C). The processes of the phase transformation from a liquid state to room temperature for undercooled Cu-35wt pct Fe alloys were discussed, in order to understand the solidification with metastable liquid separation. By means of melt-spinning technique, it was indicated that the microstructure of solidification for Cu-35wt pct Fe alloys could be refined due to the high cooling rate.     

Selective oxidation of copper from liquid copper-silver alloys

In this work, the oxygen refining of liquid copper-silver alloys with a borosilicate slag was studied. First, a comprehensive thermodynamic analysis was performed using the data available in the literature. The results indicate that since silver oxide is relatively unstable in silicate-based slags, then it should be thermodynamically feasible to oxidize copper from copper-silver alloys with a very low silver loss to the silicate slag. In actual practice, although relatively low copper levels can be achieved in the metal phase, the silver losses to the slag are excessive. Therefore, in the present work, both kinetic and equilibrium experiments were performed on a molten copper-silver alloy containing 12.68 mass pct silver in order to elucidate the mechanism of silver loss to the slag. The kinetic experimental results indicated that copper levels of less than 2 mass pct could be achieved with silver recoveries of about 95 pct after relatively short refining times of 15 minutes. In the equilibrium experiments, the copper contents of the metal were less than 1 mass pct, and these values were in good agreement with those which were calculated from the data of previous researchers. In order to explain the relatively high silver losses to the slag, a model was developed which is based on the transport of silver from the metal phase to the slag phase both in metallic form and as silver oxide in the copper oxide oxidation product. The copper and silver oxides and the metallic copper-silver alloy are all transported into the slag by the oxidizing gas bubbles. It is proposed that once in the slag, the silver oxide is unstable and decomposes into metallic silver which is not easily recovered in the metal phase. Also, the transfer of the copper-silver alloy into the slag, by the gas bubbles, promotes the slag-metal exchange reaction, which again results in the generation of silver particles in the slag.     

Deoxidatlon rate of copper droplet levitated in Ar-H2 gas stream

Levitated copper droplets, 5 mm in diameter with initial oxygen contents of 0.036 to 1.9 wt pct, were deoxidized at about 1970 K in an Ar-H₂ gas stream. The Ar-H₂ gas mixture having hydrogen partial pressure less than 4 kPa was introduced into a silica reaction tube of 11-mm ID at gas flow rates up to 2 x 10^-4 Nm³s^-1. The effects of initial oxygen content of the droplets, hydrogen partial pressure, and gas flow rate on the deoxidation process were examined. A mixed control model for the deoxidation rate involving both gas and liquid film mass-transfer resistances was combined with a thermodynamic relationship for the dissolved species in molten copper. The value of 2 × 10^-4 m 73x00D7; s^-1 was assigned to the liquid film mass-transfer coefficient of dissolved oxygen throughout all experimental conditions. Under the experimental conditions of low initial oxygen content and high hydrogen partial pressure, the liquid film mass-transfer resistance was significant. When a droplet of high initial oxygen content was deoxidized, transition phenomena from gas to liquid film mass-transfer control were noticed in the later stage of reaction. It was deduced from the present model that the accumulation of dissolved hydrogen was indispensable to these phenomena. Formerly Student     

Desulfurization of bath smelter metal

The purpose of the present work was to determine the mechanism and optimal conditions for desulfurizing bath smelter metal with a CaO-CaF₂ flux. The minimum silicon (0.1 pct), or aluminum (0.3 pct), contents in the metal for optimal rates were determined. It was found that 8 to 10 pct CaF₂ at 1450 °C is required and that the rate below the CaO-CaF₂ eutectic temperature (1360 °C) is very slow. It is proposed that a liquid phase at the surface of the CaO particles is required, which is provided by the addition of CaF₂. The Si or Al is required to reduce the number of phases for the reaction from three, when carbon is controlling the oxygen potential, to two when Si or Al is; two-phase reactions are inherently faster than those involving three phases. For the optimal conditions, the rate is controlled by mass transfer of sulfur in the metal to the CaO-CaF₂ surface. A simple model for continuous desulfurization indicates 95 pct desulfurization can be achieved at high production rates for metal containing 0.10 to 0.15 pct Si using a CaO-10 pct CaF₂ flux at 1450 °C. Formerly Research Associate, Department of Materials Science and Engineering, Carnegie Mellon University     

Desulfurization kinetics of molten copper by gas bubbling

Molten copper with 0.74 wt pct sulfur content was desulfurized at 1523 K by bubbling Ar-O₂ gas through a submerged nozzle. The reaction rate was significantly influenced not only by the oxygen partial pressure but also by the gas flow rate. Little evolution of SO₂ gas was observed in the initial 10 seconds of the oxidation; however, this was followed by a period of high evolution rate of SO₂ gas. The partial pressure of SO₂ gas decreased with further progress of the desulfurization. The effect of the immersion depth of the submerged nozzle was negligible. The overall reaction is decomposed to two elementary reactions: the desulfurization and the dissolution rate of oxygen. The assumptions were made that these reactions are at equilibrium and that the reaction rates are controlled by mass transfer rates within and around the gas bubble. The time variations of sulfur and oxygen contents in the melt and the SO₂ partial pressure in the off-gas under various bubbling conditions were well explained by the mathematical model combined with the reported thermodynamic data of these reactions. Based on the present model, it was anticipated that the oxidation rate around a single gas bubble was mainly determined by the rate of gas-phase mass transfer, but all oxygen gas blown into the melt was virtually consumed to the desulfurization and dissolution reactions before it escaped from the melt surface.     

On the Formation of non-metallic inclusions in fe-50 pct Ni type alloys by deoxidation with Mn,Si, and Al

This investigation deals with deoxidation experiments in 30 g lab melts of Fe-50 pct Ni alloys. After deoxidation with different amounts of Mn, Si and Al and their combinations the samples were quenched into water at different times. Metallographic studies comprising light microscopy, scanning electron microscopy, electron microprobe and image analysis were performed. Classical nucleation theory was used for computation of the different supersaturation with oxygen or the deoxidant necessary for homogeneous nucleation. The different deoxidation reactions and the transformation of inclusions due to diffusion of oxygen, or the deoxidant, from or into the inclusions was treated for the different cases of deoxidation. Most deoxidation reactions take place within some seconds. The experimental results were to be used to estimate the pertinent interfacial tensions between the oxides and the melt and the values obtained for the different oxides seemed to be reasonable. The diffusional computations were successfully used for predicting the different transformations taking place. For example, in deoxidation with 0.03 pct Si the oxygen solubility is controlled by the equilibrium with liquid FeO ⋅ SiO₂. The time taken to reach equilibrium is determined by the number of inclusions and the particle size. In deoxidation with 0.1 pct Si or more, the equilibrium is controlled by SiO₂ inclusions and the time taken to reach equilibrium, less than 1 s, is much shorter compared to the samples with 0.03 pct Si. The deoxidation reactions with aluminum were treated in the same way, and it was shown that the number of particles determined the time elapsing before equilibrium with respect to the formation of FeOAl₂O₃ or A1₂O₃. It was further shown that transformation of primary liquid FeOAl₂O₃ with high contents of FeO into solid FeOAl₂O₃ was expected to occur within one second. However, the experiments showed that it took somewhat longer, due to formation of solid FeOAl₂O₃ around the liquid FeOAl₂O₃ inclusions, thereby preventing the diffusion of aluminum into the particles.     

Vacuum distillation of copper matte to remove lead,arsenic, bismuth,and antimony

Vacuum-refining experiments were carried out on copper matte melts, containing 35 to 73 pct Cu, to measure the removal rates of lead, bismuth, arsenic, and antimony over the temperature range of 1373 to 1523 K under pressures in the range of 50 to 130 Pa. High rates of refining, controlled by mass transport in the liquid phase, were achieved for all impurities in melts containing up to 65 pct Cu and for chamber pressures less than 100 Pa. After 40 to 60 minutes of treatment, lead elimination was between 70 and 96 pct, bismuth elimination was between 88 and 98 pct, arsenic, elimination was between 60 and 93 pct, and antimony elimination was between 40 and 92 pct. The overall mass transfer coefficients for vacuum refining for the impurities considered fell in the range of 5×10⁻⁵ to 2×10⁻⁴ m s⁻¹. The values were insensitive to small changes in melt temperature but decreased with increasing pressure above 250 Pa. Also, the rates of refining were seen to be influenced by the sulfur and oxygen activity in the melt. The evaporation and subsequent elimination of the impurities were mathematically modeled by extending previously published models beyond consideration only of evaporation of monomers to consideration of the evaporation of monomers and dimer compounds. The model showed that due to the contributions to the refining by evaporation of each of the metallic, oxide, or sulfide vapor species, arsenic and antimony exhibited a maximum in refining rates atP _{o 2} andP _{s 2} potentials corresponding to matte grades of about 55 pct copper, whereas lead showed a minimum at about the same matte grade, and bismuth showed a continuously decreasing rate of refining with increasing matte grade. The model was also used to simulate the refining behavior of copper matte melts. An example of a commercial-scale operation is given.