Lunar Oxygen Therapy

Oxygen is among the most viable near-term resources to be mined and used in space, but where should we get it from?

Two different strategies are emerging on opposite sides of the Atlantic for mining oxygen in space. After flirting with O2 from regolith for decades, NASA is making a dash for water from the lunar poles as a source of oxygen and hydrogen propellant, going as far as creating a budget line item for a water pilot plant in the 2028 timeframe. They’re sponsoring the Break the Ice Challenge that mirrors their internal architecture planning, and recently put out the Lunar Water ISRU Measurement Study. US companies focusing specifically on lunar ice include Lunar Outpost for prospecting and Eta Space for processing.

Across the pond the Europeans have their sights set more on O2 pulled directly from warmed or molten regolith. Two higher profile commercial efforts include Metalysis who were recently awarded an ESA development contract, and HELIOS in Israel who will fly a demo mission with their proprietary tech.

(The distinction isn’t perfectly clear cut: US companies are also working on O2 from regolith, and ESA has the upcoming PROSPECT mission that will land at the lunar south pole).

Is one of these strategies better, and is one side missing something? Here’s some basic pros and cons for each:


O2 from polar ice

Oxygen and hydrogen produced

Extensive prospecting likely required

Low-temperature, straightforward processing

Extremely challenging cold trap environments

No consumables needed


O2 from regolith

Oxygen and metal produced

No prospecting needed

High-temperature, complex reactors needed

Can operate anywhere, relatively benign conditions

Electrodes and electrolytes degrade over time


Not really decisive either way. Let’s look at a more apples to apples comparison in terms of energy costs and resource grades. In the chart below I’ve calculated the energy needed to extract oxygen against the amount of bulk material that has to be processed to produce an equal amount of O2 from different sources. There’s some simplification here but the numbers come out consistent with NASA’s internal trades. I’ve also included asteroid resources both from heating carbonaceous material (clays), and using O2-from-regolith for S-type and E-type asteroids. 

The x-axis implies energy costs in terms of processing bulk material (hauling, etc.). You want to be in the bottom left of this chart, but then two paths diverge in different directions. 

At 1.5-2% water by weight, icy regolith is essentially on par with O2-from-regolith on a joule for joule basis. In other words, if you had a pile of icy regolith already sitting on the surface, it makes sense to throw it out if the grade is less than about 1.5% and extract oxygen directly from the silicate regolith instead. Of course, that ignores the energy costs of prospecting and of operating in cold trap environments, so the break-even is probably a good bit higher than 1.5%. 

This is my current thinking about ice content at the lunar poles:


Surface ice (optical surface)

It’s quite likely the detections from the LOLA, LAMP, and M3 instruments (including the 30% number that gets thrown around) are thin transient surface frosts. You can detect high apparent ice contents from an instrument that only senses to 10 microns below the surface, but if the ice is present as a micron-thick frost it adds up to zilch in terms of volume. 

Near-surface ice (0-1 m)

Here I trust the LPNS neutron data (not the LEND instrument). This gives about 0.2-1 wt.% ice in cold traps within the upper 70 cm or so. But there’s a huge degeneracy with the thickness of a desiccated layer at the surface, and how patchy the ice is.

Deep ice (>1 m)

I’m more bullish on deep ice than ever, but it’s likely many meters down. New interpretations of the LCROSS data suggest >10 wt.% ice about 5-6 m deep or more but almost nothing above that; small crater depth/diameter ratios and surface roughness are highly suggestive of substantial subsurface ice; and my own modeling suggests complex laying with abundant ice that’s meters to tens of meters deep.


Without heavy machinery and big industry in place to get at the deep ice, we’re relying on the surface and near-surface ice in the near term. The surface ice is probably useless and not worth talking about. So that’s where things get interesting: the near-surface ice values hover just below the break-even point with O2-from-regolith. A little higher (2-10% range in localized patches) and it might just be worth mounting a prospecting campaign and braving the cold. On the low end it’s not competitive even if the prospecting, digging and hauling were free.

So both strategies make some sense: if you think you can find enriched areas of near-surface ice or can reach the deep ice (assuming it exists), then it’s worth pursuing water. But the downside risk could easily put companies out of business. Unfortunately, the VIPER rover isn’t going to buy down this risk very much because it won’t go near the large cold traps you’d want to mine from. O2 from regolith has lower downside and greatly simplifies operational logistics, but carries upside risk if there are in fact higher grades of ice at the poles.

Two more notes:

1. Almost everyone working on lunar ice is restricting themselves to techniques that only work in the upper meter or shallower. I think this is misguided: if the LPNS results bear out then there’s really not much ice at these depths. And there shouldn’t be, based on the geologic processes operating at the poles. I want to see plans for prospecting and mining ten meters or one hundred meters down. Rodwells?

2. Quick word on asteroids. CI chondrite-like material at the very high end of water content (20% by weight) trades a bit worse than O2 from lunar or asteroid regolith. But it’s difficult if not impossible to uniquely identify CI-like material using telescopic spectra. CM-like material can have a diagnostic 0.7-micron absorption from Fe-serpentine, but it trades much worse than regolith-based O2 sources. Both require handling a lot of dry bulk material: I think if you want to mine an asteroid, bring the whole thing back and process it at a centralized cislunar facility. From what we’re finding with Bennu and Ryugu, you can crush the stuff like floral foam.