Aging of Si3N4-bonded SiC Sidewall Materials in Hall-Héroult Cells: Material Characterization and Computer Simulation

Abstract

Si3N4-bonded SiC materials have become the state-of-the-art sidewall block materials used in aluminium electrolysis cells. Despite its outstanding properties, the material properties degradate in the cell over time. The mineralogical transformations and chemical reactions cause changes in the physical and thermal properties, which influence the thermal balance of the electrolysis cell. Sodium plays an important role in the degradation of the sidewall block, as in the other parts of the cathode lining, and the transportation of sodium in the cathode lining has been in focus in the study in order to understand the degradation mechanisms.

Commercial virgin Si3N4-bonded SiC materials have been thoroughly characterized with respect to microstructure and phase composition. α-Si3N4 was found more pronounced in the area close to the surface of the virgin block than in the center by refinement of X-ray diffraction patterns. The morphology of the phases of the materials was investigated by scanning electron microscopy (SEM). The needle grains in the nitride binder were confirmed to be α-Si3N4 by transmission electron microscopy, while β-Si3N4 was proposed to be hexagonal rod shaped grains. The apparent porosity was lower in the area close to the surface while the density was higher.

The degradation of the Si3N4-bonded SiC materials was studied by investigating spent blocks. The degradation mechanisms in the gas zone (top part) and the electrolyte zone (bottom part) of the spent sidewall block were different due to the different chemical environments in the cell. In the electrolyte zone, the degradation involves reactions with sodium, and the main degradation phase was Na2SiO3 or a mixture of Na2SiO3 and Na2Si2O5. The presence of only Na2SiO3 points to a higher chemical potential of sodium. The formed silicates filled up the pores and dramatically decreased the apparent porosity. The layer structure of the sodium silicate was confirmed by SEM. In the gas zone, the degradation mechanism was more complicated. The material underwent active oxidation, and the main degradation phase was Si2ON2 or cristobalite depending on the partial pressure of oxygen. The apparent porosity and density of the bulk material was not significantly changed compared to the virgin material. At the surface of the block towards the pot, material loss was observed. The remaining material at the surface appeared porous and a significantly higher porosity was inferred. The needles in the binder in the gas zone was found thinner compared to the ones found in the virgin material. HF was proposed to play an important role in the material loss at the surface and the change in the morphology of the binder.

The thermal conductivity of the Si3N4-bonded SiC materials before and after operation was measured by the lasher flash method. The thermal conductivity was found to be higher in the area close to the surface of the virgin block than in the center despite a higher apparent porosity at the surface. Moreover, the thermal conductivity dropped dramatically in both the electrolyte zone and the gas zone of the spent blocks. The variation in the thermal conductivity of the material was modeled taking into account the phase composition, the apparent porosity and the microstructure related properties such as grain boundary resistance, pore shape and orientation factor and finally a thermal resistance layer. Both analytical models and computer simulations showed that the microstructure evolution had a strong influence on the thermal conductivity. Furthermore, the computer simulation demonstrated that the drop in the thermal conductivity of the sidewall block would result in a hotter surface of the block towards the bath and a thinner side ledge.

Sodium diffusion and reaction throughout the whole cathode lining has been investigated by a computer simulation by taking into consideration the different degradation mechanisms in each material of the cathode lining. The simulation demonstrated that the dominant transport of sodium in the carbon cathode was solid phase diffusion. The diffusion in the sidewall block was gas phase diffusion, and the diffusion is to some degree limited by the chemical reactions. Using the diffusion constant of sodium in the refractory layer close to the sodium selfdiffusion in an albite like molten oxide reproduced the sodium penetration depth found in autopsies.

As shown in the work, sodium causes the severe degradation in the cathode lining. Moreover, it “opens” the way for the bath to penetrate into the cathode lining resulting in more severe degradation. Two approaches were applied to estimate the bath infiltration rate into the carbon cathode block. The simulations demonstrated that the infiltration rate was determined by the changes in the wetting property of the carbon material towards the bath, meaning that it was determined by the sodium diffusion rate, which explains that sodium always infiltrates faster than the bath.

A diffusion barrier (like a steel plate or a chemical barrier) is recommended to be placed between the ramming paste (big joint) and the sidewall block, which could inhibit/reduce the sodium infiltration into the sidewall block and so eliminate further degradation. That will not only improve the chemical stability of the sidewall block but also its thermal stability, and thereby maintain the thermal balance during the cell operation. This may also open up for heat recovery from the heat flux through the side walls.