The math that makes refueling from the Moon appealing is pretty simple. "As a rule of thumb," write the authors of the new study on the topic, "rockets launched from Earth destined for [Earth-Moon Lagrange Point 1] must burn ~25 kg of propellant to transport one kg of payload, whereas rockets launched from the Moon to [Earth-Moon Lagrange Point 1] would burn only ~four kg of propellant to transport one kg of payload." Departing from the Earth-Moon Lagrange Point for locations deeper into the Solar System also requires less energy than leaving low-Earth orbit, meaning the fuel we get there is ultimately more useful, at least from an exploration perspective. //

the researchers decided to focus on isolating oxygen from a mineral called ilmenite, or FeTiO3. It's not the easiest way to get oxygen—iron oxides win out there—but it's well understood. Someone actually patented oxygen production from ilmenite back in the 1970s, and two hardware prototypes have been developed, one of which may be sent to the Moon on a future NASA mission.

The researchers propose a system that would harvest regolith, partly purify the ilmenite, then combine it with hydrogen at high temperatures, which would strip the oxygen out as water, leaving behind purified iron and titanium (both of which may be useful to have). The resulting water would then be split to feed the hydrogen back into the system, while the oxygen can be sent off for use in rockets.

(This wouldn't solve the issue of what that oxygen will ultimately oxidize to power a rocket. But oxygen is typically the heavier component of rocket fuel combinations—typically about 80 percent of the mass—and so, is the bigger challenge to get to a fuel depot.). //

The team found that almost all of the energy is consumed at three steps in the process: the high-temperature hydrogen reaction that produces water (55 percent), splitting the water afterward (38 percent), and converting the resulting oxygen to its liquid form (5 percent). The typical total usage, depending on factors like the concentration of ilmenite in the regolith, worked out to be about 24 kW-hr for each kilogram of liquid oxygen. //

Obviously, we can build larger arrays than that, but it boosts the amount of material that needs to be sent to the Moon from Earth. It may potentially make more sense to use nuclear power. While that would likely involve more infrastructure than solar arrays, it would allow the facilities to run around the clock, thus getting more production from everything else we've shipped from Earth.