1 t of aluminium requires spending of 600 kg of graphite electrodes, ru-wiki says.
Alternatively, you can reduce it chemically from chloride with sodium or potassium amalgam (the historicaly first process, used in mid-XIX), much more expensive.
Or you can reduce it with CO at 2000+ C temperatures, which is also not the easiest and the cheapest way.
The alumina together with silica are one of the most stable compounds in the universe, that's why the rockey planets consist mostly of them.
So, as Mercury is solid, the direct splitting of the alumina with solar power looks having problems.
Iirc, to the date the real achievement was to split several hundred grams of stones, and just to extract the oxygen.
As the solid waste would consist of aluminium, magnesium, titanium, and iron, and be polluted with sulfur, so their further separation will result into something not very different from the current process. Maybe also, they will need to oxidize them to separate the stable Al2O3 from others, lol.
"To produce 1000 kg of raw aluminium, you need 1920 kg of alumina, 65 kg of cryolite (Na3AlF6, sodium hexafluoroaluminate, the deposists are very rare, but is produced from fluorite), 35 kg of aluminium fluoride (still requires fluorite), 600 kg of anode graphite electrodes, and about 17 MWh = 61 GJ of electric energy."