The ion engine

I am designing an ion thruster. This is the first note from that work: what the engine is, where it already flies, why it is weak, and what a reactor would change. For a primer on nuclear physics, read What nuclear means.

What it is

An ion thruster is an electric rocket. A chemical rocket burns fuel and throws the hot gas out the back at about 4.5 kilometres a second. An ion thruster takes a gas, usually xenon, knocks an electron off each atom to make an ion, and then throws the ion out the back with an electric field at 30 to 50 kilometres a second. Ten times the exhaust speed means a tenth of the propellant for the same change in velocity. That is the whole appeal.

They are not science fiction. Hundreds of communications satellites hold their position over the equator with them. Every Starlink satellite raises and holds its orbit with a Hall thruster, a close cousin, first on krypton and now on argon. Dawn used three to visit Vesta and then Ceres, the first spacecraft to orbit two bodies. Hayabusa and Hayabusa2 went to asteroids and came back on them. BepiColombo is on its way to Mercury on four.

The catch

The catch is thrust. Dawn’s thrusters produced 91 millinewtons each, about the weight of a sheet of paper resting on your hand. The ship accelerated for years, not minutes. Thrust is set by electrical power: twice the power, twice the thrust at the same exhaust speed. Dawn had 10 kilowatts of solar panels at Earth’s distance from the Sun, and sunlight falls off with the square of distance, so at Jupiter the same panels give 400 watts, and at Saturn 100. Electric propulsion runs out of power exactly where the distances get long.

What a reactor changes

A fission reactor does not care where the Sun is. A megawatt of reactor power turns the same thruster technology from a sheet of paper into forty newtons, and a ten-tonne ship that needed eight years to gain ten kilometres a second needs a month. That is why nuclear physics is on my reading list, and why the next essay is about it.

General arrangement, explodable. Drag Explode to pull the assembly apart along the thrust axis. Xenon enters at 1, electrons from the cathode at 2 knock an electron off each atom inside the chamber at 3, the grids at 5 and 6 throw the ions out at tens of kilometres a second, and the neutralizer at 7 sprays electrons into the beam so the ship does not charge up. Swap part 8, the power module. Thrust follows power; solar power follows the inverse square of the distance from the Sun; a reactor follows neither.

Why not scale it up and fly cars

One question comes up every time: if it is quiet and clean, why not scale it up and fly cars on it? Because thrust per watt is 2/v. The ion engine’s virtue, a 30 km/s exhaust, is exactly what makes it weak per watt. A hundred kilonewtons, enough to lift ten tonnes, needs 2.4 gigawatts, two large power stations, and at any imaginable power density that plant weighs far more than it can lift. Hovering wants the opposite engine: push a lot of air slowly. A rotor gets fifty times the thrust per watt of any ion device, which is why propellers are right for anything that fights gravity and ion engines are right for anything that has already left it. In air, a gridded thruster does not even work; the grids arc and the ions collide before they accelerate. The atmospheric cousin, ionic wind, flew a 2.5 kilogram plane at MIT in 2018, silently, with no moving parts, and it is twenty times too weak per watt for a useful drone. A reactor fixes power far from the Sun. It does not fix power per kilogram, and lift is entirely a power-per-kilogram problem.

Portable reactors

The next question is whether the small reactors now being built change that. They are real, at the fission end. NASA’s KRUSTY ran in 2018 at a few kilowatts, and the lunar surface power unit it leads to is 40 kilowatts in under six tonnes. The Pentagon’s Project Pele and Westinghouse’s eVinci are 1 to 5 megawatt units that ship in containers, built for remote bases. A submarine reactor is the most mature mobile fission plant there is, and its compartment weighs on the order of a thousand tonnes for tens of megawatts. Count the shield, the cooling and the machinery that turns heat into electricity, and all of them land in the same band: 10 to 50 watts per kilogram today, 100 to 200 as a design goal. A piston engine is about a thousand. A jet engine is five to ten thousand. Nuclear has sat a hundred times below combustion on this axis since the 1950s, when the Air Force tried to build a nuclear bomber and found the crew shield alone weighed tens of tonnes. Against the 2.4 gigawatts the lifter needs, even 200 watts per kilogram is a 12,000-tonne power plant to lift ten tonnes.

Fusion is not yet real at any size. The 2022 shot at the National Ignition Facility put out 3.15 megajoules from 2.05 megajoules of laser light, with about 300 megajoules drawn from the wall to make that light. The companies aiming for net electricity are building plants the size of buildings, and a deuterium-tritium plant needs a shield against 14 MeV neutrons that is heavier than a fission reactor’s. “Portable fusion” is a phrase from press releases.

What portable fission does change

What portable fission does change is the thing this essay is about. Forty kilowatts on the Moon runs through the fourteen-day night, which solar cannot. A megawatt in deep space makes electric propulsion fast where sunlight is gone. And if you want nuclear with real thrust, you skip electricity: a nuclear thermal rocket heats hydrogen in the core and pushes it out directly, with no generator and no radiators. NERVA reached 250 kilonewtons at twice the chemical exhaust velocity in the 1960s. The most recent attempt to fly one, DRACO, was cancelled in 2025. It was always a space engine; nobody will run a reactor exhaust through the atmosphere.

The next essays work through the physics the engine depends on, starting with what nuclear means.