Development of a Novel Non-Consumable Anode for Electrowinning Primary Aluminum

The principal goal of the project was to determine through theoretical considerations and from chemical and electrochemical laboratory studies the technical and economic feasibility for the substitution and retrofitting of an SOFC-type anode for today's carbon anode in a cell for electrowinning primary Al. However, solubility measurements showed that no value of cryolite ratio can exist where the solubilities of the solid electrolyte components (zirconia and especially yttria) would be small relative to the alumina solubility. Therefore, the utilization of the proposed SOFC-type anode cannot be realized for any cell involving a cryolite-base solvent. However, the project suggested that the SOFC-type anode scheme might be successful if the solvent/electrolyte for electrowinning Al could be changed to a fused sulfate melt. During the solubility experiments, electrochemical probes were developed, and a bath characterization was defined, to measure quantitatively the acid-base character of cryolite melts. The measured acid-base behavior was then used to correlate the alumina solubility in cryolite over a wide range of cryolite ratio at 1300K. A mathematical modeling of the alumina solubility as a function of basicity identified three solutes of Al{sub 2}O{sub 3} in cryolite-base melts: Na{sub 2}Al{sub 2}OF{sub 6}, Na{sub 2}Al{sub 2}O{sub 2}F{sub 4}, and Na{sub 4}Al{sub 2}O{sub 2}F{sub 6} as acidic, neutral and basic solutes, respectively. For the first time, the stereochemistry (geometries) of these complex solutes was clarified. For the non-oxygen containing Al-F complex anions, Na{sub 3}AlF{sub 6} and NaAlF{sub 4} were also considered as solutes, and some NaF (but no AlF{sub 3}) could remain in the melts. The previously suggested solute Na{sub 2}AlF{sub 5} was found to be unstable. The strong complexing in the cryolite/alumina system means that the bath is highly buffered so that a significant shift in basicity is not possible and therefore the alumina solubility does not vary greatly. The maximum solubility for alumina occurs at a cryolite ratio of about four. The method used for theoretical modeling of the alumina solubility in the NaF-AlF{sub 3} system involved the simultaneous solution of all possible equilibria in the bath coupled with element balances, similar to the software program SOLGASMiX. Such an analysis identified the dominant complex oxyfluoride solutes in the system and provided a quantitative evaluation for their stabilities. With these new values added to the thermodynamic data bank, the solubilities of other oxides in cryolite could be analyzed. Thus new papers by other authors on the solubilities of NiO/NiAl{sub 2}O{sub 4}, FeO/FeAl{sub 2}O{sub 4}, and TiO{sub 2} in cryolite were interpreted differently than the original authors to identify the solute ions and provide quantitative data for their stabilities in these systems: Na{sub 2}NiF{sub 4}, Na{sub 4}NiF{sub 6}, Na{sub 2}FeF{sub 4}, Na{sub 4}FeF{sub 6}, FeF{sub 2}, Na{sub 4}TiO{sub 2}F{sub 4}, NaTIOF{sub 3}, and Na{sub 3}TiO{sub 3}F. Again the stereochemistry for these solutes was described by simple geometric (octahedral and tetrahedral) arrangements of large anions about a smaller cation. In the case of the solubilities for the oxides of iron and nickel, the data and explanations would be useful in understanding the dissolution of the proposed Fe,Ni oxide composite for the oxygen-evolving inert anode. The research on this project has demonstrated the correct method to analyze the complicated equilibria in such a complex solution, and exposed as inadequate (incorrect) the existing (log-log) method of treating only one or two equilibria in isolation. In response to a high priority from the 2002 Aluminum Roadmapping report, two sensors were proposed to measure the dissolved alumina content in cryolite. Because of the high cost of the necessary BN hardware, however, confirming experiments were not undertaken. Hole-in-tube type sensors were designed and demonstrated for the measurement of Mg activity/concentration in binary Al-Mg melts and for Li activity/concentration in binary Al-Li melts. In each case, these probes provided accurate and reproducible data over the commercial ranges of temperature and composition. The probes were tested and performed well in industrial shop conditions, and the sensors are considered suitable for industrial commercialization and in-line service at this time