Note
Carbothermic Reduction of Alumina at 1823 K: On the Role of Molten Iron and Reaction Mechanisms
2016 Volume 56 Issue 7 Pages 1300-1302
Details
2016 Volume 56 Issue 7 Pages 1300-1302
A possible reaction mechanism is presented for the low temperature carbothermic reduction of alumina in the presence of molten iron under inert conditions. This reduction was found to take place in a number of stages that involved the carburization of molten iron, disintegration of alumina into sub-oxide gases, reduction of these gases by the solute carbon followed by the capture of reduced aluminum by molten iron due to its high affinity. With increasing levels of aluminum in the iron, the wettability of alumina with molten metal undergoes a transition from poor wetting (with Fe) to good wetting (with Fe–Al–C) leading to the dissolution and subsequent reduction of alumina in the Fe–Al–C melt. These reaction steps were observed to occur in Al2O3–Fe2O3–C system at 1823 K and atmospheric pressure.
The carbothermic reduction of alumina: Al2O3(s)+ 3C(s)=2Al(l)+3CO(g), has been investigated extensively.1,2) These reduction reactions are known to be thermodynamically feasible only above ~2273 K and are known to proceed in the region of 2473 K at atmospheric pressure.3,4,5) Thermodynamic studies on the system have revealed that the reduction of alumina is controlled by the formation of aluminium sub-oxide vapors.6) Some studies have been reported on experimental investigations on the carbothermic reduction under vacuum conditions.7,8) Frank et al. have reported on the occurrence of carbothermic reduction of alumina in the temperature range 1973 K to 2073 K in the presence of a metallic solvent (Cu and Sn) at pressures between 0.08 to 0.20 atm.9)
Alumina is generally considered to be chemically inert during contact with molten iron at 1823 K.10) During interfacial phenomena investigations on Al2O3-12.9%C/Fe system at 1823 K in Ar atmosphere using the sessile drop method, we however observed clear evidence for chemical reactions taking place in the system.11,12) After 30 minutes of contact, video images of the iron droplet showed intense activity in the form of fine aluminium oxide whiskers emanating from the droplet and on the refractory substrate. Small quantities of iron aluminide intermetallics were recorded as a reaction product in the interfacial region; these reactions were accompanied by the extensive penetration of molten iron into the substrate, the generation of CO gas and carbon pick-up by molten iron. It was concluded that alumina could no longer be treated as chemically inert under these conditions. Further studies were carried out on Al2O3-26.53%C-Fe system at 1823 K. To enhance the contact between the reacting constituents, these were mixed together as a pellet instead of the sessile drop arrangement.13) Detailed x-ray diffraction and SEM/EDS investigations were carried out on reacted assemblies. After 30 minutes of heat treatment at 1823 K, the diffraction pattern contained peaks for Al2O3, Fe3AlC, Fe3Al and C; no peak was observed for pure iron, one of the key reacting constituents. After 60 minutes, alumina peaks were much reduced in intensity and Fe3Al peaks had increased significantly in intensity. All Fe3AlC peaks had disappeared completely indicating it to be an intermediate phase. These studies clearly point towards the carbothermic reduction of alumina taking place in the presence of molten iron at 1823 K and ambient pressure.
In a recent publication, Zienert et al.14) have questioned the possible occurrence of carbothermic reduction at such low temperatures and high pressures, and have suggested that another chemical process must be considered to explain these experimental results. They carried out a numerical simulation on the Al2O3–C/Fe system in a sessile drop arrangement to investigate the diffusion of various species across the interface; the simulation experiment was approximated to 180 equilibrium steps of 1 minute duration. The reaction at every step involved the dissolution of small amounts of alumina in liquid iron and the reaction of dissolved oxygen with carbon to form CO. Alumina was continuously dissolved into iron and 9.75 at% of aluminium was present in molten iron after 180 minutes. It was claimed that no carbothermic reaction was observed between alumina and carbon in the presence of molten iron.
There is however a fundamental mistake in this simulation. The non-wetting behaviour of alumina with iron is very well known and has been investigated extensively. Contact angles ranging between 100° and 141° have been reported in literature in the temperature range 1530–1600°C in inert atmosphere.15,16) Being highly non-wetting, there is no driving force for alumina to dissolve into molten iron. The presence of aluminium in molten iron therefore cannot be accounted for via this mechanism. This note presents possible reaction mechanisms for the observed carbothermic reduction of alumina in the presence of molten iron at 1823 K.
Experiments were carried out on the (Fe2O3, Al2O3, C) system at 1823 K; molten iron was produced in-situ through the reduction of Fe2O3.17) High purity (99.8 pct) fused alumina was mixed thoroughly with synthetic graphite in a 70:30 proportion (blend I). Fe2O3 powder was mixed with C in the ratio 75:25 wt pct (blend II); which was then mixed with blend I in the ratio 20:80 wt pct with 5 wt pct phenolic resin as a binder. This mix was heat treated in the form of powder; its C content was estimated to be 29%C. Additional carbon was added for the in-situ reduction of iron oxide. The sample was heat treated in a graphite crucible at 1823 K for times ranging between 30 minutes to 2 hours. The furnace was continuously purged with 99.99% pure argon with a flow rate of 1.0 L/min. Further experimental details are given elsewhere.11) The furnace was initially purged with argon for 20 minutes with a slight positive pressure (~1.3 atm.) to ensure that no gas was leaking into the furnace; the oxygen levels in the outgoing gases were measured continuously and were below 2 ppm. In-depth x-ray diffraction and SEM/EDS studies were carried out on reacted pellets.
Figure 1 shows detailed SEM/EDS results on reacted pellets after 2 hours of heat treatment; molten iron (seen as bright regions in the back-scattered image) was seen to have nucleated at a large number of places throughout the matrix. Several large bright regions containing metal were also identified; EDS analysis showed these to be composed of Fe, Al and C. Two distinct phases were clearly identified: one the bright shiny outer region and second the dull grey regions embedded within. The relative Al concentration in the grey regions was higher than the corresponding Al concentration in the bright regions. High magnification images of specimens heat treated for 30 minutes are shown in Figs. 2 and 3.
SEM/EDS images of (Fe2O3, Al2O3, C) system after heat treatment at 1823 K for 2 hours. Two distinct phases can clearly be seen in the shiny metallic phases.
A high resolution image showing disintegrated alumina in the form of spaghetti like strands surrounding the metal droplet.
A close-up view showing the disintegration and subsequent dissolution of alumina into the metal droplet. The formation of spidery multiple strands was observed at several places on the droplet surface.
Figure 2 shows a Fe–Al–C droplet that appears to be surrounded by spaghetti like alumina strands; this change in the morphology of alumina seems to indicate its disintegration. Further magnification of these images (Fig. 3) clearly shows this disintegration taking place on the surface of Fe–Al–C droplet. Spider like regions with multiple long strands appeared to emerge out of an alumina particle (Region ‘A’). These structures of varying sizes had nucleated at several places on the metal droplet. While regions ‘B’, ‘C’ and ‘D’ contained Al, Fe and C, the relative proportion of Al/Fe was found to decrease continuously from ‘B’, the region closest to alumina, to somewhat lower level in ‘C’, the region connecting the Fe–Al–C droplet to alumina strands and the region ‘D’ representing the Fe–Al–C droplet. These images provide clear evidence for the disintegration and the subsequent dissolution of alumina into Fe–Al–C droplets. Results for long duration heat treatment samples generally showed the reaction end products (Fe3Al), some unreduced alumina and residual carbon.
The low temperature carbothermic reduction of alumina is expected to occur through the formation of gaseous sub-oxides AlO and Al2O. Although the spontaneous occurrence of these reactions [e.g., Al2O3+2C=Al2O(g)+2CO(g); ΔG (J/mole)=1246 247−528.8T; ΔG (1823 K)=282218 J/mole]9) requires higher temperatures, some disintegration of alumina has been known to take place at 1823 K. During steel casting, these sub-oxide gases have been observed to get re-oxidized resulting in alumina inclusions and clogging of sub-entry nozzles. The key difference in the present study is the inert atmosphere and a complete absence of oxygen which prevents the re-oxidation of Al2O gas to Al2O3. Instead, the Al2O gas is reduced to Al by solute carbon present in molten iron [Al2O(g)+C=2Al(g) +CO(g); ΔG (J/mole)=684370−262.38; ΔG (1823 K)=206045 J/mole];9) and the reduced Al is captured by molten iron due to its high affinity.11) The kinetics of these reactions is expected to be quite slow at these temperatures.6) The overall reaction sequence can be represented as:
With the reaction products Al2O and/or AlO being continuously removed by molten iron, the rate of alumina disintegration gets enhanced leading to the generation of more and more sub-oxide gases. Such an enhancement of reaction kinetics in the presence of molten solvent was also observed by Frank et al. as well.9) With increasing amounts of Al being absorbed/captured, the Fe–C droplet transforms to Fe3AlC alloy. This transformation was found to be complete within 30 minutes as no Fe peak was observed in the XRD spectrum. This transformation also brings about the change in the wettability behaviour of alumina with molten metal. As alumina has good wettability with ferro-aluminium alloys,18,19) it starts to dissolve in this alloy and is eventually reduced by the carbon present. Over extended contact, large amounts of alumina (up to 80%) were consumed and reduced within Fe3AlC alloys to form Fe3Al alloys. The Fe3AlC phase was clearly observed to be an intermediate phase and was not observed in the heat treatment studies for 2 hours.
These results clearly show that the carbothermic reduction of alumina can indeed take place under these operating conditions. Inert conditions and a complete absence of oxygen were an essential prerequisite for these experiments. The key characteristics of molten iron, e.g., the presence of solute carbon which acts as a strong reducing agent, its high affinity with Al19) and its role as a metal solvent, also played an important role in enhancing the reduction kinetics at such low temperatures and pressures. Our experimental results provide clear evidence for various stages of the low temperature carbothermic reduction of alumina in the presence of molten iron.
This research was funded by a Discovery Projects grant from the Australian Research Council (ARC). We gratefully acknowledge their support and UNSW analytical Centre for research facilities.
