Nuclear Thermal Propulsion

Unlike most things at Beyond NERVA, the nuclear thermal rocket (NTR) has been built and tested by both the US and the USSR. Both programs looked at a solid core NTR, the simplest type of nuclear rocket. In a nuclear thermal rocket, the heat from a reactor is used to directly heat a propellant, which is pushed out a nozzle at the back of the spacecraft. This propellant isn’t burned, just heated, so there’s no need for the fuel to even be able to burn. However, the lighter the propellant, the more efficient the rocket, since the same amount of energy can make a smaller molecule travel faster than a larger one, and this is proportional to the atomic mass of the molecules that make up the propellant. Because of this, hydrogen is usually chosen since it’s very lightweight, but ammonia, CO2, and others have been proposed over the years as well.

NTRs are usually broken up by the state of matter of the fuel within the reactor: solid, liquid, gas, and some in-betweens like vapor core reactors.

Options Abound

Solid Core NTR (NTR-S)

Schematic view, Finseth

NERVA A6 Reactor, Finseth 1991

Solid fueled NTRs, like in the rockets tested in Project Rover (US) and the RD041X (USSR), are closest in form to the reactors used on Earth. However, solid fuel has limitations, mainly with how hot it can get before melting down. There’s lots of variants on this, using different shapes and materials for fuels, but in general they aren’t as efficient as a rocket that could reach higher temperatures. They are still significantly better than chemical engines, with the NERVA NRX getting a specific impulse of 825 using H2 propellant, and the Russian RD0411 getting an isp of about 900. There are other designs,  like pebblebed reactors, that can do even better: the Project Timberwind rocket was supposed to break 1000 seconds of isp, and still other designs offer improvements here as well.

Recent NASA research into NTP (nuclear thermal propulsion, their preferred acronym) has focused on lowering the enrichment required for the fuel by using a new form of fuel, a ceramic metal or CERMET. This is just one of several different types of fuels that can be used in an NTR-S, which are looked at more on the main page.

For more information on NASA’s current work in nuclear thermal propulsion, click here!

Liquid Core NTR (NTR-L)

Liquid fueled NTRs have been proposed as a way around the temperature limitations of solid fuel: don’t fight melting down the reactor, do it on purpose and plan for it! This is more difficult to pull off, but there are some designs that have been proposed over the years that make this possible. These designs can run much hotter than a solid core NTR, and so are more efficient, but the materials you make the reactor itself out of are going to melt at some point if you make it too hot, so there’s still limits to how good this rocket can be.

This is a rare type of reactor, so any information about NTR-L reactor architectures I don’t have on the main page, please leave in the comments!

Closed Cycle Gas Core NTR (NTR-GC)

Overall Configuration, McLafferty

Nuclear Lightbulb Concept, McLafferty and Bauer 1968

To achieve even higher temperatures, sometimes the fuel is turned into a gas, or sometimes a plasma. This is a gas-core nuclear thermal rocket, and there’s two varieties with very different advantages and disadvantages. The first is called the closed cycle gas core NTR, also sometimes called the “nuclear lightbulb.” This was one of the first nuclear rocket designs proposed, first suggested by Robert Bussard in 1957. Here, the gas is contained in a quartz vessel that looks like a lightbulb, and is squeezed by a piston to make it dense enough to sustain a chain reaction. The reason quartz is used is that X-rays can’t go through it, so this could possibly be used to launch from the surface of planets. The downside is that at some point the quartz is going to break down, so you can’t get this reactor as hot as if the gas wasn’t physically touching anything. Despite these limitations, a nuclear lightbulb engine could theoretically reach a specific impulse of over 2000 seconds, and with the additional shielding from the quartz bulb and a thrust-to-weight ratio of greater than one (some designs claim as large as 10!), this is one of the only designs that can be used to take off from the surface of the Earth… and possibly land as well.

Open Cycle Gas Core NTR (NTR-GO)

gascore14 NASA

Open Cycle Gas Core Concept, NASA

This next concept, where the fuel is a charged plasma and held in magnetic fields, is the open cycle gas core NTR. There have been a number of designs over the years, to help maximize fuel burnup and avoid having unburned uranium swept away by the passing hydrogen propellant. Because the gas doesn’t physically touch anything, these rockets can get much hotter than any other design, but since there’s nothing physically separating your propellant from your fission reaction, fission products and quite a bit of unburned fuel go out the nozzle and into the exhaust, making it highly radioactive. This is NOT something that you want to be near when it lights up, except behind some very good shielding at a fair distance away. However, this is the first engine on the list that meets the common impression of nuclear spacecraft: with a specific impulse of over 9000, thrust of up to 3 MN, and a thrust-to-weight ratio of over 30, this is a rocket that can propel truly massive spacecraft very fast across the solar system.

Vapor Core NTR (NTR-VC)


Vapor Core, Diaz INSPI

There are other options as well, where the fuel isn’t quite one state or another, such as in the vapor core rocket, where the fuel is held in tubes similar to what you’d find in a solid NTR, but instead of fuel pellets the fuel is uranium tetra- or hexaflouride vapor, which gives some advantages for fuel burnup and temperature. A rocket of this type can reach a specific impulse of about 1000, and a thrust-to-weight ratio of greater than 5, and some variants can do even better.

Molten Core NTR (NTR-MC)

Reactor Core CS

LARS, Powell e.a., 1973

Another in-between concept is the molten core NTR, where the fuel only partially melts, like in the Liquid Annular Reactor System (LARS) by Dr. Powell and colleagues at Brookhaven National Labs in 1973; this allows the material that the reactor is made out of to better handle the heat of the reactor, but causes problems when trying to figure out the behavior of the chain reaction in the fuel. 

Pulsed NTR (NTR-P)


Pulsed NTR via Wikimedia

There is one final NTR design that is quite new, and doesn’t quite fit into any of these categories: the pulsed NTR, proposed by a team headed by Dr. Francisco Arias at the Polytechnic University of Catalonia, Spain, in 2014. The inspiration for this reactor is a terrestrial design called the TRIGA (Training, Research, Isotopes, General Atomic) reactor, originally designed by Freeman Dyson and  Edward Teller. While the recket design uses solid fuel in the form of plates, the reactor was designed specifically to handle large pulses of power safely (up to 22,000 MW). Also, the fact that this is a pulsed system, rather than a steady-state system, makes it unusual for an NTR, but it generates thrust by heating hydrogen propellant and accelerating it out of a nozzle, so by definition it’s an NTR. The really nice thing about this design is that it has two modes: Thrust amplification, for orbital insertion burns and other orbital events, and specific impulse amplification, which has a lower thrust but is much more efficient. Calculating thrust and specific impulse on this engine is not simple, but the isp that it is theoretically capable of reaching makes it a potential candidate for interstellar probe missions. 

Propellants for NTRs


H2 via Wikimedia

The propellant of choice for an NTR is hydrogen: its low atomic mass means that the atoms are accelerated more per unit energy than any other atom, and even H2 (molecular hydrogen) still weighs less than helium does. However, its low atomic mass also leads to additional complications that are far harder to deal with than other propellants:

  1. Hydrogen will seep into, and through, metal fuel tanks and pipes (carbon composite is an even bigger problem), leading to boiloff. There are designs to mitigate or eliminate this, but they are still in the development phase.

  2. Cryogenic transfer of fuels is difficult, even after spinning up both tanks for the transfer. Experiments on the Space Shuttle and the International Space Station have demonstrated this capability, but it has never been done in bulk on a regular basis. Testing is ongoing, and will continue to be investigated for chemical as well as nuclear engines.

  3. Hydrogen is incredibly reactive, not just combustible, with many different materials. Especially in the reactor core and the throat (smallest constriction) of the nozzle, hot hydrogen erodes most substances very quickly.

  4. Hydrogen is incredibly transparent to most EM radiation, meaning it’s very difficult to transfer much heat per unit volume. This means that your propellant mass flow has to be quite high compared to how much heat is actually being captured from the reactor. Additional, or better, turbopumps for the hydrogen are needed than for other propellants.

  5. Hydrogen is very bulky, even in cryogenic storage, due to its low atomic mass. This is less of an issue for spacecraft that don’t have to worry about atmospheric drag, but larger tanks mean more mass to construct them, especially for hydrogen (see #1).

So, why use hydrogen at all? The same cause of all these headaches (atomic mass) also means that it can develop the highest exhaust velocities, and therefore is the most efficient propellant to use. Furthermore, when H2 is heated to above 1000 K, the hydrogen dissociates, making it even lighter and therefore far more efficient. Effectively, this adds an extra 1000 seconds of isp to your engine.

Other Options: What, Where, and Why?

One nice thing about an NTR is that if it’s able to be a gas while it’s in the reactor, it could theoretically be used for propellant. Many propellant gas possibilities are easier to handle and more dense and stable. The reactor design isn’t significantly different from a nuclear powered ramjet design, so in a place with an atmosphere (and no concerns about potentially irradiated exhaust) the NTR can use whatever’s on hand.

There are three common suggestions for other propellants, although there’s a table of a wide variety of gas possibilities at the Atomic Rockets page. The most common suggestions are: Carbon dioxide, methane, and ammonia. In fact, the USSR’s first design for an NTR was an ammonia-based first stage, but the thing scared Korolev (probably for very good reason) and the program was cancelled shortly after work had started. However, with such a large trade-off in specific impulse using these larger molecules for propellants, and the increased chance of other chemical reactions, the trade-off needs to make sense for the mission that it’s being used on.

Carbon_dioxide_3D_spacefill wikimedia

CO2 via Wikimedia

Carbon Dioxide (CO2) is probably the best example of where the trade-off makes sense. This is probably best illustrated by Dr. Robert Zubrin’s Nuclear rocket using Indigenous Martial Fuel (NIMF). There’s two varieties of this rocket, a winged one and a ballistic one, but the ballistic one seems the most versatile of the designs so we’ll focus on that. The rocket has compressors that gather the Martian atmosphere and pressurize it. After its propellant tanks are full, the reactor warms up and then uses the on-board pressurized Martian atmosphere to power itself on a ballistic trajectory, land, and start the repressurization process. There’s no need for propellant depots, or complicated infrastructure, to have rapid, worldwide reach. There are some radiological concerns about nuclear-powered landers (radiation backscatter, flux coming off the mostly unshielded reactor, and others) that were at least partially addressed in the design, but the use of the atmosphere for propellant is incredibly attractive. One other disadvantage to CO2 is that at high temperatures, it dissociates, becoming carbon monoxide and oxygen, and then to just carbon and oxygen. This carbon tends to plate itself out on surfaces (this is a way that bulk graphite is made, after all), including exhaust or cooling tubes. Therefore, it requires more maintenance to clean the irradiated carbonized gunk out of the increasingly radioactive reactor. This isn’t nearly as big a problem as it sounds (the process would largely be similar to swabbing out a cannon, could be automated, and most of the radiation coming off the reactor is alpha and beta, which are easily shielded against), but it does decrease your engine’s efficiency (and neutronics) over time.


CH4 via Wikimedia


Methane is another common suggestion, this time for use on Titan. Methane is even heavier than carbon dioxide, and also dissociates into carbon and hydrogen, which means carbon buildup is also a problem. However, it can also be stored far more densely with far less difficulty than either CO2 or H2, and may be plentiful in the outer solar system and in some carbonaceous asteroids.


NH3 via Wikimedia

Ammonia is another option for propellant. One major advantage it has is the fact that when it dissociates, it does so into nitrogen and hydrogen, both gasses, so while chemical reactions remain a concern there aren’t solids plating out on your reactor’s exhaust tubes. Ammonia is also potentially common in asteroids and the outer solar system, and is easily synthesized and stored.

Propellant Doesn’t Have to be One Gas

Using just one gas in a rocket is not the norm in modern rocketry. While using only one propellant is an advantage in that it adds simplicity, it’s also possible to combine more than one substance to use as propellant. There are two common examples of this concept.

Liquid Oxygen Augmentation

LANTR Borowski 1991

LANTR Schematic, Borowski 1994

First is the LANTR, or LOX (liquid oxygen) Augmented Nuclear Thermal Rocket. This is effectively an afterburner on a solid core NTR (but the design can be adapted to other types as well): Liquid oxygen is carried as an additional propellant, injected into the hot hydrogen exiting the nozzle, and ignited, giving a big boost of thrust at a cost of both isp and mass for the additional systems in the spacecraft. However, having high thrust is a good thing for a space tug or Moon ferry, where going faster is all about having a lot of thrust for a rapid change in velocity. 

Propellant Seeding

Second is the possibility of having the propellant absorb more of the heat. For some engines (the open-cycle gas core, for instance) this isn’t a matter of convenience but of necessity as part of the cooling system: if most of the heat passes through the gas, it’s going to melt itself down. To deal with this, tungsten microparticles are added to the hydrogen gas stream (melting point 3625 K, boiling point 6203 K), although tantalum halfnium carbide has been proposed as well (melting point 4488.15 K, highest known melting point). The better your propellant absorbs heat, the smaller your radiators have to be on your spacecraft (due to the high temperatures being generated, most designs need large radiators or other heat rejection mechanisms to not melt).

Bimodal Nuclear Rockets: Have Your Cake and Eat It Too!

BNTR Schematic NASA

BNTR Schematic via NASA

Many NTR designs also incorporate electricity production capability. While this adds mass and complexity to the reactor system, it also eliminates the need for additional systems to provide electrical power for the spacecraft. Many different ways have been proposed in the past, and they’ll be looked at more in depth as time goes on.

The most common type of bimodal design is meant to use both thermal and electric propulsion in the same spacecraft using the same reactor. This way, the higher thrust thermal propulsion can be used for major burns, such as leaving or entering orbit around a planet or moon, and the electric propulsion can be used in what would normally be the cruise phase to decrease overall mission time.

Whether this type of rocket engine is a good choice depends on how much mass the electric propulsion system adds, and the challenges of integrating additional cooling capability into the reactor can be daunting. Many different options have been proposed over the years, each addressing these tradeoffs and challenges differently.

Other Ways to be Bimodal

One additional way to change the way an NTR operates has already been looked at, the LOX-augmented NTR (LANTR). These systems increase thrust, but decrease the specific impulse of the rocket. Since we’ve looked at this before, we’ll mention this concept here and move on.

NTER Inductor Cutaway, Dujarric

Induction Heater for NTER, Dujarric e.a. 2013

Another current design looks to overcome the thermal limitations of solid fuel to increase the specific impulse of an NTR, by adding a thermal induction heater downstream of the reactor core. This is the Nuclear Thermal Electric Rocket, under development by Dr. Dujarric and his team at ESA. Other designs have looked at using electrically heated thermal rockets run off a nuclear reactor, but often those are tentative designs for reaction control systems on nuclear powered space stations and the like, and have tended to be small systems. However, this sort of bimodalism is very rare.

Trimodal NTR

There is one final type of hybrid NTR to mention, and that’s the Trimodal NTR. This is mainly explored in the TRITON concept, but there have been some other proposals. Here, a relatively typical bimodal NTR with both thermal and electric porpulsion systems has an oxidizer afterburner in the same way that the LANTR has. This allows for a wide range of thrusts and specific impulses, but at the cost of adding the mass and complexity of not just one additional propulsion system, but two, and an additional independent cryogenic storage, pumping, and emission system as well.

Coming Up

This page will be updated from time to time with additional changes, links, and the like. However, most of the additional content will probably be published on the links for the different types of reactors. Eventually, cryogenics and other engineering considerations will be addressed as well, in Dealing with Physics.

Keep checking back for more updates!