Metal Refining

From John Freudenthal

In regards to economics, there are two broad methods of metal refining; electrolysis and carbothermal. Carbothermal is the current method used for iron, but it works for almost any common metal above a certain temperature threshold.

FeO + CO <=> Fe + CO2

For iron, the equilibrium favors the right side of the above equation above about 800C [L'vov]. In the case of aluminum, the full reduction from Al2O3 has two potential reactions.

Al2O3 + 3C <=> 2Al + 3CO

Al2O3 + 3CH4 <=> 2Al + 3CO + 4H2

The first reaction has a favored equilibrium on the right starting at about 2000C and the later at about 1500C [Halman]. In the case of aluminum, only the electrolysis method (Hall Heroult) is currently used.

In my opinion, thermodynamics favors large scale production due to the high temperatures involved. In order to make small scale refinement feasible two factors are vital; lower temperature and board applicability. Carbothermal reduction has the advantage that its broadly applicable. Almost metal will undergo carbothermal reduction at some temperature, and the use of methane as a feed material almost always lowers the necessary temperature a bit to boot.

So if maintaining temperatures above 1500C seems feasible, then carbothermal reduction might be the best route, but here is where feedstock and ore come in to cause trouble. Every rock just insists on being composed of a disgusting mess of minerals, and pure ores are the only economical source of pure metals. So if using local feedstock is a necessity, then finding universal purification methods is a must. This is where hydrofluoric acid comes to the rescue. HF will oxidize (well, here the colloquial term might be 'flouridize') anything. Fluorides are, in general, more water soluble than oxides (17.2 g/Kl for AlF3 and zilcho for Al2O3) and most importantly, are volatile and will evaporate. This allows for simple distillation purification [Landis], but in this case, low temperature distillation is likely unfeasible, and hence solution phase seperation would be preferable. Once the AlF3 is isolated, it can be electrolyzed with the Hall Heroult process (or any electrolysis method) to form pure Al.

In the Landis paper, all separations are done as low temperature distillations (~-100C) and reductions are done as plasma reductions, which are obviously not possible here.

The trouble is that flourine is the mother of all oxidizers, and upon electrolysis its freed from its AlF3 cage and unleashed upon the world in its full glory. The flourine will oxidize anything; the anode, the chamber, the salt bath, people. In Hall Heroult the feed stock is generally Al2O3 and a carbon anode is oxidized to CO2, but in the case of AlF3 feed stock, the reaction generates free flourine which either oxidizes the anode to CF4 or likely hydrogen would be added to the cell to regenerate the HF which then would be pumped out the of the cell. (Don't quote me on that, I'm having trouble finding the HF bond strength at 2000C, but I think its favored over CF4) The trouble is that the HF still has to be handled extremely carefully or it will simply fluorinate anything it touches.

In a broad sense, what I'm proposing instead is to simply reduce whatever mix of metals you have access to into an alloy, in this case AlSi and then separate the metallic alloys using a second melt electrolysis. Its not really any lower energy, and in the end, both methods still require a high temperature melting. In the case I propose for Al2(SiO3)3, the second melting step and electrolysis is relatively easy because the two metals have drastically different melting points and electrode potentials. In addition, its all done with electrolysis and this single cell could be used for both steps if it was built to withstand 2000C which would require a good deal of forethought.

So overall, no economically viable method exists to reduce a mixed oxide ore to its constituent separated metals. That is what makes this proposition so much fun.

File:Landis.pdf

File:Lvov.pdf