Three independent lines
Radar, neutrons, and thermal models on laser altimetry.
Arecibo found anomalously bright radar patches near both Mercurian poles in 1991, and the interpretation was contested for two decades. MESSENGER settled it in 2012 on three independent lines at once: radar brightness, hydrogen measured by the neutron spectrometer, and thermal models run on Mercury Laser Altimeter topography, which showed the bright patches sitting exactly where a surface never sees the Sun.
The deposits are dominantly water ice. That is as firm as a remote-sensing result gets without a lander, and it is firmer than the corresponding lunar result was at the same stage. Ice grade — concentration, form, and whether it is a coherent slab or a frost dispersed through regolith — is not known at either pole, and it is one of the two things a first Mercury lander would be sent to find out.
How much, and where
Comparable to the Moon, on the planet nearest the Sun.
| Body | Reserve | Coldest floors | Cover |
|---|---|---|---|
| Mercury | 10¹¹–10¹² t | <100 K | Exposed in the coldest floors, organic lag elsewhere |
| The Moon (south pole) | 10⁹–10¹⁰ t | 25–40 K | Dispersed in regolith, no comparable lag |
Mercury’s zero obliquity is what makes this possible. A tilt of 2.04 arcminutes means a shadowed crater floor near either pole has been shadowed continuously for the age of the solar system, with no seasonal libration to admit sunlight. The Moon’s 1.54-degree tilt is 45 times larger, and the lunar cold traps are correspondingly less rigorous even though the Moon sits at a tenth the solar flux. Rigor beats distance.
The lag is the carbon mine
Tens of centimeters of organic-rich material over the ice.
In the coldest floors the ice is exposed. Slightly equatorward it sits beneath an insulating lag of dark material tens of centimeters thick — probably complex organics delivered by the same comets and volatile-rich asteroids that delivered the water, and left behind as the more volatile component sublimated away. The ice you can see is a minority of the ice that is there.
Mining the covered deposits therefore means excavating a meter-scale organic overburden, and that overburden is itself the only substantial carbon source on a planet with almost none. Crustal graphite runs 1 to 4 weight percent in the low-reflectance terrain, but the polar lag is carbon that has already been concentrated and delivered to the one place industry is going to be working anyway.
This is the hazard-as-resource inversion in its cleanest form: the thing in the way is the thing you wanted. Carbon matters at Mercury because the crust is pre-reduced and oxidizer-poor, which closes off the carbon-bearing routes every other inner-system body uses.
Why the floors are 100 K, not 40
A cold trap with hot walls.
Lunar shadowed floors at Shackleton reach 25 to 40 kelvin, colder than Pluto’s surface, and that is the physical basis of the cold-compute argument on the Moon pages. Mercury’s coldest floors are modeled below 100 kelvin — cold enough to hold ice for billions of years, and nowhere near 40.
The reason is that a crater floor in permanent shadow still sees its own sunlit walls, and at Mercury that surrounding terrain is receiving 6.7 times the flux. The floor is cold because it never sees the Sun and warm because everything around it does. So the cold trap works and the cryogenic sink does not: superconducting logic at 77 kelvin has no margin against a 100-kelvin floor, where at Shackleton it has 37 kelvin of it.
The cold-compute half of the Shackleton argument does not transfer to Mercury. Water, cryogenic propellant storage and shielding all transfer intact. The 40-kelvin floor does not, and no amount of engineering recovers it, because the heat is arriving by radiation from terrain that has to be there for the site to be valuable.
What the water is for
Shielding, oxidizer, and a little propellant.
Water at Mercury is a local utility with a large reserve, a low delivered cost on site, and no export edge at all. That is a shape the water topology does not otherwise contain: every other large reserve on the map is either a commodity with a corridor attached or a reserve nobody can reach.
Three local uses. Shielding, which is the design answer to the compressed-magnetosphere event and is already underfoot at any shadowed-floor site. Oxidizer, which matters more here than anywhere because Mercury’s crust cannot supply it — there is no iron oxide to reduce. And propellant for the small amount of chemical work that has to happen on site, since the bulk export is thrown electromagnetically rather than flown.
The ice-mining volume implied by a hydrogen-oxygen propellant path is a large part of why Mercury’s launcher is electromagnetic rather than chemical.