The current process used for primary aluminum production, the Hall-Heroult process, is reliable, but it also is expensive, consumes large amounts of energy, and generates significant quantities of greenhouse gas emissions. One possible alternative process is the carbothermic reduction of alumina, wherein aluminum is formed by reducing alumina with carbon at high temperatures. This process, if successful, has the potential for substantial reductions in energy consumption, capital costs, and greenhouse gas emissions. One critical component to making this process successful involves obtaining a better understanding of the thermodynamics. Specifically, the key thermodynamic data are the free energies of the reactions and the thermodynamic activities of the metal (Al-C) and slag systems (Al2O3-Al4C3). These are critical for evaluating and controlling the carbothermic process, but experimental data is extremely limited and much of it was measured many years ago when the experimental techniques available may not have been adequate.
The overall objective for this research was to assess the validity of the thermodynamic data for this process, as well as its suitability for predicting the behavior of the process. This was done through experimental investigations into both the slag (carbide) making reaction and the binary Al2O 3-Al4C3 phase diagram. The comparison of these results, to those expected based on the current understanding for the process thermodynamics (using FactSage along with the ALCO database), assesses the validity of the thermodynamic data.
In this document, the experimental results for investigating the reactions of Al2O3 with carbon are presented. This work involved measuring the operating line for the first step of the carbothermic aluminum process, slag making. This was done using two experimental methods. One involved measuring the evolution of CO from the reactions using a mass spectrometer. The other involved using a vacuum thermobalance (TGA) to measure the weight loss from the reactions. Additionally, two separate reactors were used for the CO evolution measurements. One was carefully designed to minimize the concentration of nitrogen, from air, near the reactants (Reactor B). The other allowed for a significant concentration of nitrogen (Reactor A). The use of these two reactors allowed the influence of nitrogen on this slag making operating line to be determined. Also, experiments were performed making measurements for the binary Al2O3-Al4C3 phase diagram. These included measuring the Al2O3-Al 4O4C eutectic as well as the Al2O3 liquidus line.
In general the measured operating line is close to the predicted line, with the exception being at Al2O3 saturation, where there a significant difference. The measured slag making operating line appears to support the predicted values (temperature and slag composition) through both the single phase liquid and at Al4C3 saturation. The data also supports the temperature for the operating line at Al2 O3 saturation (1948°C), but the slag composition here is measured to be much lower than predicted (5.2 mole % Al4C 3 vs. 7.6 mole % Al4C3). No clear explanation is provided for these lower than expected carbon concentration.
The effect of nitrogen on this slag making operating line is minimal. No discernible difference was observed through both the single phase liquid and at Al4C3 saturation. At Al2O3 saturation, the temperatures were found to be the same, while the composition of the slag was found to be slightly less concentrated with carbon when nitrogen was present.
The eutectic point for the Al2O3-Al4C 3 phase diagram was measured in two separate ways, slag solidification during cooling and slag melting during heating. Both revealed the same temperature (1885-1886°C), which is lower than the predicted value (1908°C). The measured eutectic compositions were slightly different (9.24 mole % Al 4C3 and 10.7 mole % Al4C3), but neither was significantly different than the predicted value (10.1 mole % Al 4C3). These measurements, along with the slag compositions at Al2O3 saturation from the operating line measurements appear to support the idea of a steeper alumina liquidus line. This has implications for the Al2O3-Al4C3 system and thermodynamic models that appear to be unrealistic. (Abstract shortened by UMI.)