Belt Water & Propellant

Where the water is

Belt Water Comes in Two Forms, and They Cost Different Amounts

Free ice needs heat. Bound hydroxyl needs a furnace.

Water in the belt exists in two chemically distinct states and the distinction decides the whole extraction economy. Free ice sits as a solid in cold traps and beneath the surface, and it comes out with nothing more than heat — below roughly 145 K it is stable in vacuum indefinitely, and above that it becomes mobile. Bound water is locked into the crystal structure of phyllosilicates as hydroxyl groups, and driving it out requires 600 to 1,000 K, which is not warming ice but decomposing a mineral.

Most of the belt’s water is the second kind. Carbonaceous bodies make up roughly three-quarters of the population by number and carry their water largely as hydrated clays. Both returned samples confirm it: Ryugu and Bennu material is dominated by hydrated phyllosilicates, and the Bennu samples also turned up phosphates and other evaporite minerals that formed in liquid water on the parent body long ago.

Free ice does exist and it is concentrated in specific places. Ceres has hundreds of permanently shadowed crater floors at its poles, with ice detected in several including Juling, where the deposit was observed to change over months. The main-belt comets — 133P/Elst-Pizarro, 311P/PANSTARRS and a few dozen others — develop comae near perihelion because they still hold buried ice. Observed

No belt body has had its water grade measured at a site where a machine could work. Every figure here is either a whole-body inference from density and spectra, or a laboratory measurement of material from a different body entirely.

Grade decides the energy

Ore Grade, Not Temperature, Drives Water Extraction Cost

Ceres beats the Moon on water energy despite being far colder, because the grade is far better.

The energy to extract a kilogram of water is mostly the energy spent heating the rock that came with it. At a grade of 25%, three kilograms of regolith ride along with every kilogram of water; at 5%, nineteen kilograms do; at 1%, ninety-nine. That ratio dominates everything else in the calculation.

Work it through for a cold-trap deposit at Ceres. Warming the ore from 160 K to the melting point is roughly 113 K of rise for both the rock and the ice, and then the ice itself has to be melted:

heat the gangue: 3 kg × ~800 J/kg·K × 113 K ≈ 270 kJ
heat the ice: 1 kg × 2,050 J/kg·K × 113 K ≈ 230 kJ
melt the ice: 334 kJ/kg
———
≈ 830 kJ/kg ideal, about 0.63 kWh/kg at realistic system efficiency

The same arithmetic at lunar polar grades gives 2.75 kWh/kg at 5% ice and 12.14 at 1%. Ceres comes out 4 to 19 times better than the Moon, and not because it is warmer — it is 60 K colder. It is better because the grade is up to 25 times higher. Temperature enters the calculation as a modest sensible-heat term; grade multiplies the whole thing.

Which means the site question is the whole question, and it is the same fight as at Shackleton: the good ice is in the permanently shadowed polar floors, the sunlight is elsewhere, and the two have to be connected. See the Ceres site contest.

Splitting costs ten times

Electrolysis Is 6.30 kWh per Kilogram, and Thermodynamics Sets the Floor

No process improvement gets below 4.41. The bond energy is the bond energy.

Splitting water is not an engineering cost that clever design reduces — it is the enthalpy of formation, run backwards. Water releases 285.8 kJ per mole when it forms, so putting it back into hydrogen and oxygen requires the same amount:

285.8 kJ/mol ÷ 0.018 kg/mol = 15,870 kJ/kg = 4.41 kWh/kg thermoneutral
at 70% cell efficiency → 6.30 kWh/kg

Set that against 0.63 kWh/kg to get the water out of the ground in the first place. Making propellant costs ten times what mining it costs, and the ten times is a physical constant rather than a technology gap. Every kilowatt-hour of that is supplied at 177 W/m² of sunlight, or by a reactor that had to be flown from Earth along with its fuel.

Size a modest plant to see what that means. One tonne of water a day requires 6,300 kWh a day, or 263 kW running continuously. At the belt’s array specific mass of roughly 51 kg/kWe averaged — 59 at aphelion, where it has to be sized — that is on the order of 13 to 15 tonnes of solar array for a single tonne of propellant feedstock per day, before liquefaction, storage or any of the plant itself. Modeled

Ship water, crack it in the light

The Belt Exports Water as Water

Electrolysis belongs where sunlight is cheap, and sunlight at Earth is 7.7 times cheaper.

Two costs, two places. Extraction is cheap and has to happen at the body, because that is where the ore is. Electrolysis is expensive and is indifferent to location — it needs only water and power. Solar flux at 1 AU is 1,361 W/m² against 177 at Ceres, a factor of 7.7, so the same array mass produces nearly eight times the propellant in the inner system as it does in the belt.

So the belt does the step that must be done locally and exports the intermediate product. Water is shipped as water and split where the sunlight is. That also suits the cargo: water is dense, non-cryogenic, storable indefinitely without power, and safe to handle, which is the opposite of liquid hydrogen in every respect.

Departure economics reinforce it. Ceres escape velocity is 516 m/s, so bulk water leaves the surface on an electromagnetic launcher at 0.037 kWh/kg without any propellant expended — see The Ceres Mass Driver. The energy to throw a kilogram of water off Ceres is about 6% of the energy to extract it and about 0.6% of the energy to split it.

The exception is propellant for local use. A tug operating at Ceres needs propellant at Ceres, and hauling it 15.5 months from the inner system to save on electrolysis is absurd. The rule is about export, not about self-supply: split locally what you burn locally, export the rest as water.

Storage is free

At 160 K, Water Ice Keeps Forever Without a Refrigerator

The cold that makes extraction awkward makes storage cost nothing.

A cryogenic depot in the inner system fights a losing battle against heat. Liquid hydrogen boils at 20 K, liquid oxygen at 90 K, and both require active cooling, multilayer insulation and sunshades, with boil-off measured as a continuous percentage loss. That is the recurring cost that makes propellant depots hard everywhere closer to the Sun.

The belt does not have the problem. Ambient temperature is 100 to 160 K and water ice sublimates negligibly below about 145 K, so a stockpile of ice sitting in shadow loses effectively nothing over years. No power, no insulation beyond a sunshade, no active system to fail. A belt operation can bank propellant feedstock across launch windows for free, which matters enormously in a regime where the window spacing is 15.3 months and the spares queue is 30.8.

That changes what a depot out here is. Instead of a tank of cryogens losing a few percent a month, it is a pile of ice with a shade over it, converted to propellant on demand when a vehicle actually arrives. The conversion plant and the storage are decoupled, and only the plant needs power. Modeled

What else water does

Water Is Also the Shielding, the Thermal Mass and the Life Support

Propellant is the use that gets discussed. It may not be the one that matters most out here.

Hydrogen is the most effective shielding material per kilogram against galactic cosmic rays, because light nuclei fragment incoming particles without producing the heavy secondary showers that dense materials generate. Water is the practical way to carry hydrogen, and the belt’s radiation environment is chronic with no acute component and no relief — the dose accumulates every day of a multi-year stay. The crewed architecture for this regime shields a spun habitat with locally extracted water rather than with rock, which makes water the enabling material for human presence rather than a convenience.

Beyond that, water is drinking supply and life support feedstock, thermal mass for a facility with no atmosphere to buffer it, a working fluid, and reaction mass in its own right — electrothermal and solar-thermal thrusters run on water directly at modest specific impulse, trading performance for the ability to refuel anywhere ice exists. Several small spacecraft already fly water thrusters for exactly that reason.

It is also the feedstock for the oxygen in the carbon monoxide that the carbonyl metallurgy route recycles, which ties the water plant and the metal plant together at a chemical level. See Nickel-Iron & Carbonyl Metallurgy.

Who is going after it

Water Is the Thesis the Commercial Sector Actually Believes

Sell it in space, never bring it down.

Karman+ raised a $20 million seed round on carbonaceous targets and volatiles, with a demonstration mission aimed at February 2027, and its stated model is selling propellant in space rather than returning anything to Earth. TransAstra works on the capture end — enclosing a small body in a bag and using concentrated sunlight to sublimate the volatiles out of it, which sidesteps drilling and anchoring on a surface with no gravity to speak of. AstroForge is the exception, pursuing metals, and its first vehicle was lost after launch in February 2025.

The state-funded pointer runs the same direction. China’s Tianwen-2 collects 200 to 1,000 grams from the near-Earth object Kamoʻoalewa through April 2027, returns the capsule at an Earth flyby in November 2027, and then continues to 311P/PANSTARRS, a main-belt comet, arriving January 2035. That target is volatile-bearing by definition, and it is the only funded main-belt arrival anyone has. The water-rich end of the population is where the actual hardware is going, on both sides.

Nothing has yet extracted a gram of water from any body. The total mass of asteroid material returned to Earth is about 127 grams, all of it carbonaceous, all of it analyzed rather than processed. Observed

Record

Provenance and Data Sources