Rocket Physics, the Hard Way: Nuclear Thermal Rockets

Header image: Cherenkov radiation in a nuclear reactor (image source: Argonne National Laboratory.)

“This is not nuts, this is super-nuts.”

Richard Courant, on viewing a test of the Project Orion nuclear propulsion system

Last installment, we delved into the inner workings of chemical rocket propulsion systems. Practically, however, they are limited to a specific impulse of about 450 seconds. This is partially due to the limited energy available from chemical fuels, which pales in comparison to the power of nuclear reactions. In this installment, we will explore the most basic type of nuclear rocket, the nuclear thermal rocket. We will learn about nuclear physics and the role it could play in getting humans to Mars more effectively.

Nuclear Physics 101

Disclaimer: The following paragraphs are extremely simplified explanations of nuclear physics. Exceptions and technicalities abound.

All modern nuclear power relies on the phenomenon of nuclear fission.

Atoms are composed of a tightly bound nucleus of protons and neutrons, orbited by one or more electrons. Protons have a positive charge, electrons have a negative charge, and neutrons have no charge. The densely packed protons repel each other strongly, but are kept together by the strong nuclear force.

Diagram atom
A diagram of a lithium atom (image source: AG Caesar, Wikimedia Commons, CC BY-SA 4.0.)

Despite its great strength, the strong nuclear force has a short range. It has a harder time binding bigger, heavier nuclei with more protons and neutrons. As a result, heavy elements like uranium and plutonium have unstable nuclei. At any given moment, there is a small chance that a heavy nucleus will spontaneously decay into a lighter, more stable element by emitting a particle (like an electron or positron), a helium nucleus, or a gamma ray. This emission of energy through particles is known as radioactivity.

nuclear chain reaction
A diagram of an example nuclear chain reaction (image source: Wikimedia Commons.)

There is an extremely small chance that instead of simply decaying into a lighter nucleus, the nucleus will spontaneously fission: it will randomly break up into two or more smaller ‘daughter’ nuclei, emitting neutrons and gamma rays in the process.

Uranium-235 is the workhorse of modern nuclear power. When it spontaneously fissions, it breaks up into two lighter daughter nuclei and two or three neutrons. If these neutrons strike another uranium-235 nucleus, they can cause it to fission as well, releasing another two or three neutrons which strike other nuclei, and so on. If there are enough uranium-235 atoms packed together in a small space, the rate of fission multiplies exponentially in a self-sustaining chain reaction1, producing massive amounts of energy. When this occurs, we say that the uranium is critical.

Fission reactions are extremely energetic, which heats up the uranium and causes it to emit neutrons. The heat and the neutron radiation can be used to heat a working fluid, allowing it to be harnessed for energy – or propulsion. For contemporary nuclear power reactors, this working fluid is usually high-pressure water. Water is pumped through an array of uranium rods known as a reactor core, which causes the water to boil. The reactor core can be brought to criticality (where it will produce the power we need) by manipulating the amount of neutrons allowed to participate in fission reactions.

We can use the same principle to heat propellant for a nuclear rocket.

Nuclear Thermal Rockets (NTRs)

As was explained in the installment on chemical rocket engines, gases expand when heated. Rocket engines rely on this expansion to generate thrust. Nuclear thermal rockets (NTRs), use a fission reaction as the source of this heat. NTRs were first demonstrated in the 1960s as part of NASA’s Project NERVA (Nuclear Engine for Rocket Vehicle Application.)

Here’s how they worked:

Instead of a combustion chamber, the engine had a nuclear reactor core. When the engine was activated, the reactor core would become critical. Liquid hydrogen was then pumped through the core, where it would heat up to approximately 2,000 °C. The hydrogen would then expand and rush out through the nozzle, generating thrust.

nuclear rocket NERVA
A cutaway diagram of a NERVA engine (image source: NASA.)

The first NERVA engine was fired in a ground test in 1964 (five years prior to the Moon landing), where it achieved an astounding specific impulse of 825 seconds – the very best hydrogen-oxygen engine, on the other hand, achieves about 453 seconds.

nuclear rocket test
A ground test of an NTR (image source: US National Archives2.)

The temperatures in a NERVA engine (~2,000 °C) are much cooler than in a typical chemical rocket engine (well over 3,000 °C). Nevertheless, it enjoys a performance advantage because it does not need an oxidizer, meaning that the exhaust is composed entirely of lightweight hydrogen molecules.

NTRs also work with propellants other than hydrogen – methane, ammonia, carbon dioxide, and even water have been considered. These have fewer storage problems than hydrogen, as explained in Rocket Physics, Extra Credit: Rocket Fuels. While this comes at the cost of worse performance because the molecules are heavier, a nuclear rocket can still achieve impressive specific impulses with them compared to what chemical rockets can.


How would a nuclear engine be started up and shut down? To answer this question, we need to dive deeper into the concept of criticality. As introduced earlier, there are two kinds of nuclear reactions that are important for nuclear rockets:

  1. Spontaneous decay: When an atomic nucleus randomly decays into smaller nuclei and emits neutrons.
  2. Chain reactions: When a neutron strikes an atomic nucleus, causing it to fission into smaller nuclei and more neutrons.

If only a small amount of uranium is present, fission events will occur from time to time, but too many neutrons escape the material before they can strike another nucleus. To start a self-sustaining chain reaction, or to reach criticality, we need to make it more difficult for neutrons to escape the uranium. Here are two ways we can do this:

  1. Put more uranium closer together: If a large enough mass of uranium is packed together into a small space, neutrons will be more likely to collide with other nuclei, making it easier to start a chain reaction. The minimum mass that a sphere of radioactive material needs to have to achieve criticality is known as its critical mass. Uranium-235 has a critical mass of 52 kg, which is a sphere about 17 cm in diameter.
  2. Surround the uranium with a neutron reflector: Neutrons tend to bounce off materials like graphite, steel, and tungsten. Surrounding a piece of uranium with a neutron reflector ensures that fewer neutrons escape, reducing its critical mass.

If we’re not careful, the rate of fission reactions can spiral out of control, eventually overheating the reactor. In the Chernobyl disaster, the heat from the reaction caused the pressure in the reactor to become so high that it burst. Eventually, the uranium rods became so hot that they melted into a puddle. It’s not difficult to see that a reactor core meltdown on a nuclear rocket would destroy the engine.

Nuclear weapons are an extreme example of this being done on purpose. Modern nuclear weapons implode a sphere of uranium or plutonium with explosives, making it smaller and denser. This makes it more difficult for neutrons to escape, triggering a chain reaction. The reaction proceeds so rapidly that the uranium turns into ionized gas at millions of degrees Celsius.

So how can we control a nuclear reaction? We can’t change the mass or shape of the uranium rods during operation, but we can change how strongly the engine’s walls reflect neutrons. In the original NERVA design, the reactor core was surrounded by control drums. One side contained a neutron-reflecting material, while the other side contained a neutron-absorbing material (or a neutron poison.)

Initially, when the engine is turned off, the neutron absorbers face inwards to the reactor core. The reactor core is slightly radioactive and emits some neutrons, but too many are absorbed to sustain a chain reaction.

Then, when power is needed, the drums are rotated so that the neutron reflectors begin facing inwards. The neutrons emitted by the reactor core are now reflected back, which triggers more chain reactions, which in turn produces more neutrons, bringing the reactor to criticality. Liquid hydrogen propellant is pumped in, which begins to heat up and vaporize.

neutron control drums
A diagram of the control drums, looking through the bore of the engine (image source: Los Alamos National Laboratory, Nuclear Propulsion for Space.)

Controlling the reactor is no mean feat. The drums need to be precisely aligned so that they absorb and reflect just the right amount of neutrons. If they absorb too many, the amount of neutrons (and hence power produced) will plummet until the reactor shuts down. If they reflect too many, the neutron count (and power produced) will begin rapidly spiralling out of control until the reactor destroys itself.

Powerful chemical rocket engines like the Space Shuttle RS-25s take up to eight seconds to reach full power, because the turbopumps need time to reach full speed. NTRs require an even longer startup period for two reasons.

Firstly, time is needed for the rate of fission reactions to ramp up to the required rate in a controlled manner. Starting up the reactor requires the control drums to reflect a lot of neutrons, causing the neutron count to rise until the reactor reaches the desired power level. Once it does so, the drums need to be precisely realigned to maintain a stable power level. If reactor power ramps up too quickly, the control drums may not be able to react quickly enough.

Secondly, the reactor’s fuel rods are prone to cracking if their temperature changes too quickly – a phenomenon known as thermal shock. For this reason, the startup process is kept slow.

Another important effect that helps the reactor to start is known as neutron moderation. The neutrons emitted by fission reactions are extremely fast, travelling at thousands of kilometres per second. For complex quantum mechanical reasons, fast neutrons are less likely to collide with nuclei and cause fission reactions, and so need to be slowed down. To accomplish this, the liquid hydrogen propellant helpfully acts as a neutron moderator. The neutrons slow down as they pass through the hydrogen, and so become more effective at starting fission reactions – the hydrogen thereby moderates the neutrons. As a result, the rate of reaction increases. Together, the hydrogen and neutron reflectors work together to bring the reactor to full power.

Impressively, many of these complex issues were resolved during the NERVA program decades ago.

Nuclear waste

In addition to a longer startup, NTRs also take longer to shut down.

As the engine runs, uranium is used up, leaving behind nuclear waste.

Initially, these waste products are extremely radioactive and unstable. As they decay into lighter elements, they emit additional neutrons, which help keep the reaction going and slow the shutdown process.

Furthermore, shutting down the engine is not as simple as merely exposing the neutron absorbers and shutting off the hydrogen (not if you want to use the engine more than once, at least.) Even with the neutron absorbers rotated fully inward, the reactor core will still be radioactive and generate power due to the presence of fission waste products. These need time to decay into less radioactive elements, which means that the engine needs several minutes to shut down. In the meantime, this residual heat needs to be removed or the reactor core will melt.

The solution used during the NERVA program was to put the engine through a ‘cooldown’ cycle before full shutdown. This was done by rotating the control drums to expose the neutron absorbers, then flowing hydrogen through the engine at a low rate for several minutes. The ultracold hydrogen would cool the reactor core while the nuclear waste products decayed. In this case, its effectiveness as a coolant overpowers its effectiveness as a neutron moderator.

Astute readers familiar with nuclear power may notice a possible improvement to the system. Nuclear power plants generate electricity using the same fundamental principles as nuclear thermal rockets: by using nuclear fission to heat fluids to high temperatures. What if the same reactor core used for propulsion could also be used to power the spacecraft’s electrical systems?

This idea is known as the bimodal nuclear thermal rocket (BNTR). When propulsion is not needed, the reactor is used to generate electricity. Then, when maneuvers are required, the reactor is used to heat hydrogen for propulsion. The advantage of doing this is that since the reactor is always ‘hot’, it can ramp up to the power level required for propulsion more quickly. Finishing a maneuver is also less difficult, since the reactor can just revert to power-generation mode without needing to shut down completely.

For example, this design proposes pumping a helium-xenon working fluid through the core to generate power when the engine is not in use:

NASA nuclear rocket diagram
Schematic of a BNTR (image source: NASA.)

A further improvement is the addition of a third mode called LOX-Augmented Nuclear Thermal Rocket (LANTR), making a trimodal nuclear thermal rocket. In this mode, LOX, or liquid oxygen, is sprayed into the exhaust, where it combusts with the hot hydrogen. This increases the mass flow rate and boosts the exhaust temperature, improving the engine’s thrust at the cost of specific impulse. This means that when high thrust is needed and efficiency is less important, the rocket switch to LANTR mode.

NASA nuclear rocket diagram
Schematic of a LANTR (image source: NASA.)


Nuclear thermal rockets are not without their problems.

First and foremost, the exhaust tends to carry radioactive particles with it, which poses environmental hazards if the engine fires in an atmosphere. Under modern regulations, such engines would need to be ground-tested in specialized closed facilities. In transport to orbit, a catastrophic failure of the launch vehicle could break the reactor core and scatter radioactive material over a wide radius. While these risks could be feasibly mitigated given sufficient time and effort, the true challenge is political acceptance.

Not only is the exhaust radioactive, the reactor itself is also radioactive. Usually, a nuclear reactor only emits low levels of harmful radiation when powered down. However, when operating at criticality as an engine or as a power generator, it is extremely radioactive. It mercilessly bombards everything around it with neutrons and penetrating gamma rays. Some of these neutrons may get absorbed by atomic nuclei in the spacecraft’s structure, making the spacecraft itself radioactive – a process known as neutron activation. These neutrons may also knock atoms in the structure out of alignment, making parts of the reactor warped or brittle, as well as making them moderately radioactive.

If the vehicle is crewed, radiation poses multiple health risks. This subject is complicated enough to warrant its own article, but the United States Environmental Protection Agency provides a good overview of the health effects here. There are two ways to minimize radiation exposure:

  1. Increase the distance between the crew and the reactor: The reactor will emit radiation in all directions. Due to the inverse-square law, increasing the distance to the reactor by a factor of two will decrease the radiation intensity by a factor of four. However, making the ship longer means making it heavier, because it means more structural mass is needed.
  2. Put radiation shielding between the crew and the reactor: Using hydrogen-rich shielding like water or polyethylene to stop the neutrons, in combination with dense shielding like lead or tungsten to stop the gamma rays, reduces the crew’s radiation dosage. However, shielding is also heavy.

The design of a nuclear vehicle depends on the mission. Let’s delve into a case study.

Case Study: Terra Nova

In the 2007 Canadian miniseries Race to Mars (originally aired on Discovery Channel Canada), an international team of astronauts crosses the interplanetary gulf between Earth and Mars aboard the nuclear-powered crew transfer vehicle, Terra Nova. Terra Nova’s design is closely based on the work of Borowski, Dudzinski, and McGuire at the NASA Glenn Research Centre: Vehicle and Mission Design Options for the Human Exploration of Mars/Phobos Using “Bimodal” NTR and LANTR Propulsion.

According to this design, the crew transfer vehicle (CTV) would be equipped with three bimodal nuclear thermal rockets. These are fuelled by uranium and use liquid hydrogen as a propellant.

nuclear rocket NASA
An artist’s impression of the nuclear-powered CTV (image source: NASA.)
Terra Nova rendezvous
CTV Terra Nova rendezvousing with the ascent/descent vehicle Gagarin (image source: Race to Mars website.)

The CTV would ferry the crew to Mars, where they would dock with an ascent/descent vehicle placed in Martian orbit by a previous uncrewed cargo mission. After landing on Mars and completing the surface mission, they would ascend back into orbit to meet with the CTV, which would return them to Earth.

Terra Nova diagram
A diagram of the CTV Terra Nova (image source: Race to Mars website.)

The vehicle is quite long, mainly to accommodate the massive liquid hydrogen tanks required. This length also allows it to be spun end-over-end like a baton to generate artificial gravity. However, crucially, the length also serves to distance the crew from the reactor, reducing radiation exposure by the inverse-square law.

In the original Borowski design, each nuclear engine is equipped with a ‘shadow shield’ that absorbs neutron and gamma radiation. Its main purpose is to protect the rest of the spacecraft from being damaged and overheated by the engine’s intense radioactivity. While this is sufficient for the cargo vehicles, the crewed vehicles require a heavy additional 3.24-ton external shield on each engine to attenuate the radiation flux further.

Nuclear engine diagram NASA
Nuclear engine schematic for the Borowski design (image source: NASA.)

While nuclear thermal rockets can achieve greater specific impulses, they also have poorer thrust-to-weight ratios due to the weight of the reactor and shielding. The power achievable by the reactor is also limited by the maximum temperature of the fuel rods, further reducing thrust. This means that nuclear rockets would most likely see use in heavy orbital spacecraft (where the reactor weight has a smaller impact and thrust is less important), which would only be used in space. Nuclear engines are less feasible for launch vehicles, because they require very high thrust-to-weight ratios.

That hasn’t stopped engineers from designing nuclear launch vehicles through some clever workarounds. This article has focused on designs from the NERVA program – designs that are over 50 years old – providing a rather narrow perspective. In the intervening decades, numerous new nuclear engine designs have been created, each aimed at solving a different subset of the problems that plague nuclear rocketry. However, to understand them, it’s a good idea to start with NERVA.

The Nuclear Future

NASA planned to put humans on Mars by 1982 with a gargantuan 3,000-ton spacecraft powered by nuclear thermal engines. However, Project NERVA, after years of budget cuts, was officially terminated in 1973. With the increase in public opposition to nuclear power, it seems unlikely that nuclear thermal rockets will be tested anytime soon, let alone flown. Perhaps the more advanced concepts, ones that have fewer environmental issues and safer designs, may bring them back into the realm of acceptability.

Or perhaps it will be the dream of Mars that turns the tide.

Footnotes and further reading

1 In nuclear power reactors, we also need to account for the radioactivity of the daughter nuclei, as they are usually also radioactive. They may also decay and emit neutrons, helping to sustain the reaction. These are known as delayed neutrons, because they are emitted some time after the original reaction. On the other hand, the two or three neutrons emitted by a fissioning uranium nucleus are known as prompt neutrons. Nuclear power reactors rely on delayed neutrons to keep the reaction going. On the other hand, nuclear bombs force the nuclear fuel to become so critical that it is flooded with prompt neutrons, causing it to explode. There isn’t enough time for delayed neutrons to be produced.

2 Watch the full documentary on nuclear propulsion here:

Visit Winchell Chung’s invaluable page, Atomic Rockets, for a motherlode of information on nuclear rockets and space travel in general:

Leave a Reply

Your email address will not be published. Required fields are marked *