I mean if you melt it without inert gas, etc, you get a cookie that you can break with your bare hands.
And zinc is already a part of some aluminum alloys. Pure aluminum is -really- soft, it's effectively never used
Na. What Nick said. In practice:
Melt scrap aluminum, haul off the dross, pour into a mold to make ingots. Remelt again when you're doing a pour.
Muffin trays work great if you cook them enough to burn off coatings. Heat with a blow torch before pouring in molten friggin metals.
Use greensand with appropriate ratios.
Have two vent holes. One for pouring, one for air/overflow to escape. Your pour/escape vents should be offset. Even after skimming, any remaining dross gets stuck in this part. Hacksaw off the extra bits and the remaining aluminum should be pretty good to go.
You can use a graphite rod to introduce nitrogen, but not necessary unless you are going for a mirror finish. Even flux isn't necessary if you pour within 3-5 minutes of melt. Less is better. Some fluxes add corrosion problems. Because, well, it's salt. Non-corrosive fluxes are expensive or hazmat. Fluoride salts and/or cesium. "High temperatures" and "fluoride" are things that are awesome but only if you're downwind and have an independent air supply. If you have your mold ready to go before you melt your aluminum it's not an issue. You may end up with some pinholes, but not many. Shrinkage is usually the large problem than fluxing or degassing.
It's not hard to find 'scrap' aluminum from metal shops of the alloy you want. Shavings or cutoffs. I prefer cutoffs, and easy enough to find on eBay. I wouldn't use soda cans for an AR15, because of the vinyl and paint. You'd probably be fine, especially if you did an extra ingot stage. But you wouldn't be saving that much money and adding an unknown element.