Spaceship propulsion types: chemical, ion, nuclear, sail and fusion compared

If you play space games or read science fiction, you have met a dozen kinds of engine: plasma drives, fusion torches, solar sails, nuclear rockets. Real spacecraft use only some of them, and the ones that exist trade away something you take for granted in fiction. The short version is that every real propulsion type is a bargain between thrust, which is how hard the engine pushes right now, and efficiency, which is how long a given amount of propellant keeps pushing. Chemical rockets are strong and thirsty. Ion engines are frugal and feeble. Nuclear thermal sits between them and has never flown. Sails need no propellant at all and push with almost nothing.

The one number that sorts them: specific impulse

Engineers compare engines using specific impulse, usually quoted in seconds. You can read it as “how long one unit of propellant can produce the same unit of thrust”. Higher is more efficient. It is not the same as power or speed, and a high value does not mean the engine is good at leaving a planet. Keep that in mind, because most of the surprises below come from engines that are efficient and weak.

All the values in this article are approximate ranges taken from NASA, JAXA and company statements, and real engines vary with design and operating mode. Treat them as a map, not a spec sheet.

Chemical: loud, strong, and the only way off the ground

Chemical rockets burn a fuel with an oxidizer. Typical specific impulse in vacuum runs from the low 300s of seconds for kerosene to the mid 400s for hydrogen, with methane in between; one reference range puts hydrogen and oxygen at about 421 to 460 seconds and kerosene at 300 to 340, though sources vary on sea level versus vacuum. What chemical engines have that nothing else has is thrust: they can push harder than the vehicle weighs, which is the only way to leave the ground. Every launch you have watched, including SpaceX’s Falcon 9 and Starship, is chemical, and the fuel choices (kerosene, hydrogen, methane) are trade-offs between efficiency, tank size and cost. Our piece on Falcon 9 booster reuse shows what that looks like in practice.

The weakness is the equation. Because exhaust speed is limited by chemistry, most of a rocket’s launch mass is propellant. That is why Moon plans lean on tricks like refueling in orbit.

Electric and ion: weak push, long patience

Electric engines use electricity, usually from solar panels, to accelerate propellant far faster than combustion can. NASA’s Dawn mission is the cleanest example. Its xenon ion engines had a specific impulse of about 3,100 seconds and a maximum thrust of about 90 millinewtons per engine, roughly the weight of a couple of sheets of paper. Over its mission, according to NASA technical records, the engines ran for around 51,000 hours and changed the spacecraft’s velocity by almost 11.5 kilometers per second, a record for an on-board propulsion system, using about 416 kilograms of xenon. That is enough speed to orbit two different bodies in the asteroid belt, Vesta and Ceres, which a chemical probe of that mass could not have done.

Japan’s Hayabusa2 used the same idea at a smaller scale. Its microwave ion engines each produced up to about 10 millinewtons, and with three firing together the thrust was about 28 millinewtons, at a specific impulse near 3,000 seconds. The spacecraft used them to reach, leave and return from the asteroid Ryugu.

Closer to home, SpaceX’s Starlink satellites use electric thrusters to raise and hold their orbits. SpaceX has said its V2 Mini satellites use argon Hall-effect thrusters rated at about 170 millinewtons of thrust with 4.2 kilowatts of power and a specific impulse around 2,500 seconds, which is the company’s own figure. It chose argon over xenon or krypton because the heavier gases are scarcer and more expensive.

The catch is plain. These engines cannot lift anything off a planet, and an ion-driven ship takes months to build speed. In a game, an ion drive that lets you dodge missiles is fiction; in reality it is a patient engine for probes and satellites. If you want to see how weak thrust changes steering and combat, our explainer on how spaceships turn covers the related problem of rotation.

Nuclear thermal: the one that was built and shelved

A nuclear thermal rocket uses a reactor to heat a light propellant, usually hydrogen, and expels it through a nozzle. It is not a bomb and not a reactor generating electricity; it is a very hot kettle. NASA’s description is that specific impulse of around 900 seconds or more is possible, about twice the best chemical engines, because the exhaust is lighter. The idea is old. The Rover and NERVA programs ground-tested around twenty reactors between 1955 and 1972, and a NERVA engine reached roughly 841 seconds in vacuum in tests, according to data tables and NASA histories. The program was ended in 1973, and no nuclear thermal rocket has flown.

It nearly came back this decade. NASA and the Pentagon’s research agency DARPA picked Lockheed Martin in 2023 for the DRACO flight demonstration, but DARPA later cancelled it, with an official citing infrastructure barriers to ground testing. A 2021 National Academies study assessed nuclear propulsion for Mars with a target of at least 900 seconds, which tells you where the interest is: shorter crew trips to Mars. Right now, it is a technology with a history and no ride.

Solar sails: free fuel, almost no force

A sail uses the pressure of sunlight. The force is minuscule, but it never runs out of propellant. Japan’s IKAROS, launched in 2010 alongside the Akatsuki Venus probe, unfolded a sail of about 196 square meters in interplanetary space and became the first spacecraft to rely on solar sailing for propulsion. The Planetary Society’s crowdfunded LightSail 2 declared success on 31 July 2019, after raising its orbit using sunlight alone; the gain over several days was about 2 kilometers in apogee, which sounds tiny and was the point. Sails suit patient, light spacecraft and they cannot take off. They also work only where there is sunlight to push on, and they cannot go back toward the Sun without tricks.

Fusion: still a theory for ships

Fusion propulsion is the one in your favorite sci-fi, and no fusion engine has flown or been built as a working prototype. Concepts exist on paper, such as pulsed or magnetically confined designs, with claimed exhaust speeds far above any chemical or nuclear thermal rocket. We have no verified performance numbers, because there is no hardware to measure. Anything you read that gives a precise thrust for a fusion drive is a design study or a plot device. We cover the long distances such ships would need to beat in how long interstellar travel would take.

Putting them side by side

TypeSpecific impulse (approx.)ThrustStatus
Chemicalabout 300 to 460 svery highAll launches
Ion and Hall electricabout 2,500 to 3,100 s and upmillinewtons to fractions of a newtonDawn, Hayabusa2, Starlink
Nuclear thermalabout 840 to 900 shigh, but not for launch from EarthGround tests only, 1960s to 1972
Solar sailno propellanttinyIKAROS, LightSail 2
FusiontheoreticalunknownConcept only

What a game or a story usually gets wrong

The common mistake is giving one engine every virtue. Real ships tend to carry two or three kinds: a chemical stage for getting out of a gravity well, and something efficient for cruising. Dawn used a chemical rocket to launch and ion engines for everything after. Another quirk is that thrust and efficiency pull apart, so a ship that accelerates hard cannot also be frugal. Fiction that wants fast, maneuverable ships usually hides this with a power source that does not exist. We make a browser space game, and its ships are fiction, not a physics model (we say so rather than pretend). For a more rigorous take on ship behavior in a story, see our Expanse ship combat piece.

Key sources: NASA’s Dawn ion propulsion page, NASA on nuclear thermal propulsion, The Planetary Society on LightSail 2.