Say you have a bunch of oh, I don't know, extrusion. It's pretty soft as-cast and doesn't flow well. Seems to me you should go looking for some master alloy - 6061 isn't far from 356, if you just add 10% silicon. Well, there's plenty of master alloy available, but damned if you can find it in anything but five-hundred pound drums of the stuff. Screwed? Not a chance!
Rumors deep within the shadows of the internet say an aluminum-silicon thermite reaction is possible. I haven't gotten it to go, not with my coarse sand and aluminum filings anyway. The resulting block of fused alumina with silicon semi-metal trapped inside would be, at best, difficult to seperate and dissolve in your melt, so that's no good anyway. So how else can we get it? Well, after much messing around with salt fluxes and slag, I think I've got it: add pure aluminum, sand and salt to a crucible (SiC or steel, SiC preferred because steel will dissolve; clay-graphite contains silica that will dissolve, although I don't know if the graphite and alumina content will passivate it). Trick is the salt has very slight solubility for the ingredients, ferrying it from the sand grains to the aluminum melt. When it reaches the aluminum, the displacement reaction occurs:
The major reason why the salt is required, however, is that aluminum has this dumbfoundingly tremendous power to passivate itself with oxide. Without the salt flux, even if you could get a sand grain to contact actual metal, it would react as above, forming aluminum oxide, which remains in the exact place it was formed - between the metal and silica. With a slurry of sand in molten salt, the oxide is fluxed and becomes mobile (as a flocculated black colloidial material floating around), allowing the reaction to procede.
I don't have any pictures of the actual reaction, but I have pictures of the degassing I did earlier today. Here I've melted a few pounds of material I've cooked up (which BTW goes remarkably slow at 1400°F, there was quite a bit of sand left in the slag from that two hour-long burn!). The black stuff is slag from the oxidized, misshapen pieces, as well as gunk that was left in the crucible. You can see a shiny spot on the top of the crucible.
FYI, the smoke coming off is salt vapors. Molten salt (calcium and/or magnesium chloride especially) has a high vapor pressure, meaning it evaporates quite quickly, even though the boiling point is quite a ways off; just like alcohol. Salt is also suprisingly watery, although not too suprising considering what it is (an ionized solution). Which also makes a problem of the salt creeping over every surface available through capillary action!
Here I am dipping a fluxing tool, charged with a few grams of Ca(OCl)2, also known as Pool Shock. When the chemical heats up, it releases chlorine, which bubbles through the melt and combines with hydrogen dissolved in the metal. Stay out of the fumes.
Another shot. You can maybe make out the bubbles, kinda large since I didn't hold the tool against the bottom. Oops.
I have to degas because moisture in the salt and sand readily oxidizes the aluminum, giving it a royal case of gas. Yes, liquid salt appears to have good solubility for water, erm steam, just as liquid water has solubility for salt.
The yield: a two foot bar of a few pounds weight, plus some junk that spilled, oh and the slag that was churned up from the crucible. This can be soaked in water (to dissolve the salt) to free up the metal, aluminum oxide, sand and whatever else is in it.
Now I have to assay it to see how much more cooking (with sand) it needs. This is relatively easy because silicon isn't soluble in all but the worst acids, so I just dissolve the aluminum out with some muriatic acid and weigh the silicon bits.