Plasma Separation
This may be a great way to get aluminum and other metals at an early Moon base if we could land one of these complete plasma separator units, solar panels and tanks of argon.

I still advocate sulfuric acid leaching at more advanced bases to get calcium sulfate-plaster of Paris.  This can be coated between sheets of tightly woven glass fibers instead of paper to make dry wall for lunar hotels in lava tubes that resists mold and fungus.  The CaSO4 which is separated from silica by electrostatics  can be calcined to get CaO, the active ingredient in cement.  Concrete floors and walls will have use in lava tube towns. The silica will of course be pure glass that can be mixed with CaO, Na2O, Al2O3 etc.

Iron, mild steel, duriron, titanium dioxide, titanium metal, silicon, glass,  aluminum, calcium, magnesium, and perhaps trace metals will be gotten from various processes described earlier.

These LUNAR RESOURCE pages are not the last word.  Many bugs must be worked out, energy requirments, and how to evolve all these things from small payloads sent to the Moon must be thought out.  A bigger booster like the Magnum, Shuttle C, or heck, just bring back the Saturn V, is called for in coming decades when a massive push to colonize space is made by humanity.  What would bring this about?  Fusion breakthroughs that enable the "burning" of helium 3 and the need for more energy from solar power satellites that is environmentally benign when greenhouse gas accumulation in the Earth's atmosphere and fission waste build up becomes intolerable to the world's people; and winds, conservation, hydro, geothermal, biomass, OTEC, water turbines, ground based solar, etc. cannot meet the world's demand for energy and the good life.
good webpages:

http://www.nss.org/settlement/nasa/spaceresvol3/vaporpp1.htm

http://www.nss.org/settlement/nasa/spaceresvol3/plassep1.htm
Plasma deflection depends on plasma velocity, and V depends on temps, pressure, density and mass flow rate.  Since moondust compositions may vary while fed into device plasma V may be varying.  Temp, velocity and pressure sensors could send information to computers and servos that regulate dust flow by controlling Ar pressure on dust in hopper and adjust deflection by regulating ring electrode voltage.  Give the device some intelligence.

It will also have a maze of pipes, pumps, cryo coolers and radiators.  it will be as complex or more as a rocket engine.  The only way to get it to the Moon will be to use Peter Kokh's MUScle strategy.  Massive, unitarian, simple parts like the iron magnets and even some of the titanium jacket should be made on the Moon will complex, light, electronix are upported.