Nuclear Propulsion: From NERVA to the Return of Nuclear Thermal Rockets

The fundamental limitation of chemical propulsion is thermodynamic. A rocket engine produces thrust by expelling hot exhaust gas; the energy comes from combustion. The exhaust velocity — and therefore the specific impulse — is limited by the energy content of the propellant combination and the temperature the combustion chamber can sustain before the structure fails. Liquid hydrogen/liquid oxygen, the highest-performance chemical propellant, achieves Isp of approximately 450 seconds in the best engines. This ceiling is essentially fixed. No chemical propellant combination can deliver substantially more.

Nuclear thermal propulsion breaks this ceiling by separating the energy source from the propellant. A nuclear reactor heats a propellant — typically liquid hydrogen — which expands through a nozzle to produce thrust. The energy is nuclear, not chemical; the temperature is limited by the structural properties of the reactor, not the energy content of the propellant. The result: Isp of 800–900 seconds, roughly twice the best chemical rockets, using hydrogen as propellant.

This doubling of Isp has profound mission implications through the rocket equation. For a mission to Mars requiring 5 km/s of delta-v, a nuclear thermal rocket with Isp of 900 s (exhaust velocity ~8,800 m/s) needs a propellant mass ratio of e^(5000/8800) = 1.76:1 — propellant plus spacecraft mass 1.76 times the dry mass. The same mission with LH₂/LOX at 450 s Isp (exhaust velocity ~4,400 m/s) requires a ratio of e^(5000/4400) = 3.11:1. The difference in required propellant for a 50-tonne Mars-transit vehicle is approximately 67 tonnes versus 105 tonnes — a 38-tonne savings, directly translatable to more payload, shorter transit time, or smaller launch vehicle.

Key parameters

Parameter	Chemical (H₂/O₂)	NERVA NTP	NTP (target 2030s)
Specific impulse	~450 s	~825 s	~900 s
Thrust	High (MN class)	~73 kN	~25–100 kN
Propellant	LH₂/LOX	LH₂	LH₂
Power source	Chemical	Fission (1,100 MW_th)	Fission

NERVA: The Programme That Proved It

The Nuclear Engine for Rocket Vehicle Application (NERVA) was the product of a joint NASA-Atomic Energy Commission programme that ran from 1955 to 1973. It was not a paper concept or a design study — it was a hardware programme that produced and tested engines in ground operation.

The NERVA programme built on the Kiwi series of reactor tests at the Nevada Test Site’s Jackass Flats facility throughout the early 1960s, and progressed through the Phoebus and Pewee reactor designs before converging on the NERVA flight design in the late 1960s.

Phoebus 2A (1968): The most powerful nuclear rocket ever tested. Output: 4,100 MW thermal. Thrust: 930 kN (compared to 1,960 kN for an F-1 engine on the Saturn V first stage). Run time: 30 minutes at full power. Specific impulse: 820 s in actual test conditions. Specific impulse projected for flight conditions: 825–850 s.

XE Prime (1969): A flight-configuration NERVA engine, tested at full simulated altitude conditions. Demonstrated restart capability and throttling — critical for mission operations.

The design: liquid hydrogen is pumped through the graphite reactor core, heated to approximately 2,500 K by the fission chain reaction, and expelled through the nozzle. The reactor operates at a power density of approximately 450 MW per cubic metre of core — far higher than any terrestrial power reactor. The fuel elements are uranium carbide dispersed in graphite, cermet, or metal matrix composites, chosen for their ability to retain integrity under thermal shock and hydrogen corrosion at extreme temperatures.

NERVA was cancelled in January 1973 — not for technical failure, but for a combination of budget constraints following the cancellation of Apollo 18–20 and the deferral of crewed Mars missions, and political concerns about nuclear propulsion in the post-Apollo environment. At cancellation, the programme had achieved all its technical goals. The technology worked.

The Fundamental Challenge: Hydrogen as Propellant

Liquid hydrogen is simultaneously the ideal nuclear thermal propellant and an extremely difficult engineering fluid.

Isp scales inversely with the square root of molecular weight. Hydrogen (M = 2 g/mol) is 18 times lighter than water and 7 times lighter than nitrogen — the lightest possible propellant, maximising exhaust velocity for any given temperature. This is why hydrogen is used despite the handling challenges.

The challenges are real:

• Boiling point: -253°C (20 K), 3 degrees above absolute zero, requiring cryogenic insulation throughout the propellant system
• Storage density: 71 kg/m³ at liquid state — about 1/14th the density of water, requiring large tanks
• Ortho-para conversion: Hydrogen exists in two quantum spin states; warm liquid hydrogen slowly converts from ortho to para, releasing heat that causes boiloff. Propellant storage for months-long Mars transits requires managing this conversion
• Embrittlement: Hydrogen diffuses into metals and causes hydrogen embrittlement, requiring specific alloy choices throughout the propulsion system

For Mars transit missions (180-day cruise), the boiloff of liquid hydrogen stored in moderately insulated tanks would be substantial — perhaps 15–20% of loaded propellant. Active zero-boiloff systems using cryocoolers can reduce this to near zero but at significant power consumption.

Nuclear Electric Propulsion: The Alternative Architecture

Nuclear thermal propulsion produces high thrust (hundreds to thousands of Newtons) for relatively short burns — useful for fast interplanetary transfers and acceleration out of Earth orbit. A complementary approach is nuclear electric propulsion (NEP): a nuclear reactor generates electrical power, which drives an ion thruster (Hall thruster or gridded ion engine).

NEP provides very high specific impulse (3,000–10,000 s, similar to solar-powered ion drives) but the thrust is low — tens to hundreds of Newtons. For a Mars mission, NEP would require years of continuous thrusting rather than short burns, significantly extending transit time but dramatically reducing propellant mass. For outer solar system missions where solar power is unavailable, NEP enables sustained high-Isp thrust that solar electric propulsion cannot.

The tradeoff between NTP (high thrust, moderate Isp) and NEP (low thrust, very high Isp) is mission-dependent. A common proposal for crewed Mars missions is the “bimodal” design: a nuclear reactor operates in both thermal (for thrust) and electric (for spacecraft power) modes, providing ~800 s Isp for propulsive burns and several hundred kilowatts of electrical power during cruise. Bimodal nuclear propulsion eliminates the large solar arrays that otherwise dominate power generation in the inner solar system.

DRACO and the Current Revival

After three decades of inactivity, nuclear thermal propulsion is under active development again. DARPA and NASA jointly announced the DRACO programme (Demonstration Rocket for Agile Cislunar Operations) in 2022. The stated goal: demonstrate a nuclear thermal rocket in space by 2027.

DRACO uses a low-enriched uranium (LEU) fuel design rather than the highly enriched uranium (HEU) used in NERVA. The shift to LEU (<20% U-235 enrichment, compared to NERVA’s ~93%) is driven by proliferation concerns and the regulatory environment for launching nuclear materials. The performance penalty is real but manageable — LEU core designs achieve somewhat lower power density than HEU, and the fuel temperature limits are similar.

The DRACO flight demonstration will operate at a modest thrust level — current specifications are approximately 10–20 kN, far below NERVA’s 930 kN. The objective is not operational Mars transit propulsion but demonstration of startup, operation, and shutdown in space, validation of the thermal management system, and accumulation of in-space operating data for regulatory and safety purposes.

General Atomics and X-Energy were selected in 2023 for concept development. BWX Technologies is developing the LEU fuel. Lockheed Martin and Blue Origin are part of the industrial team.

Safety and Launch Considerations

The most often-raised concern about nuclear thermal propulsion is the consequence of a launch failure — an accident that disperses the nuclear fuel into the atmosphere or ocean. NERVA-era designs used solid fuel elements that would remain intact in a propellant fire or launch explosion. The DRACO fuel approach similarly prioritises fuel integrity in accident scenarios.

Standard practice for nuclear space systems is to launch the reactor subcritical — unable to sustain a chain reaction — and only achieve criticality after the spacecraft has reached a sufficiently high orbit that any re-entry would take decades or centuries, allowing radioactive decay. For DRACO, operation is planned at altitudes above 700 km.

The radiological risk from fuel dispersal in a launch accident, given the fuel mass and enrichment level of a DRACO-class system, is extremely small compared to, for example, the Cassini RTG controversy of 1997. Nonetheless, the public communication of nuclear launch safety will likely be a non-trivial challenge for any actual crewed Mars mission using nuclear propulsion, as it was for Cassini’s launch with ~33 kg of Pu-238.

The Delta-V Case for Deep Space

The practical argument for nuclear propulsion becomes compelling beyond Mars. Jupiter missions with nuclear thermal propulsion — as studied in the cancelled JIMO (Jupiter Icy Moons Orbiter) programme — could achieve round-trip times of 5–6 years compared to 12+ years for chemical or solar electric trajectories. Neptune missions, which have a single probe visit (Voyager 2, 1989) and no dedicated orbiter, would become feasible at scientifically useful timescales with 900 s Isp propulsion.

The delta-v budget for a crewed Mars round trip using nuclear thermal propulsion is achievable with a vehicle of reasonable launch mass. Without it, the propellant mass required pushes crewed Mars missions into the category of requiring extremely large initial mass in Earth orbit — achievable with Starship refuelling but fundamentally mass-limited by chemistry.

The technology to double the efficiency of deep-space propulsion was demonstrated in 1968 and abandoned in 1973. Whether the current revival reaches operational hardware before the first crewed Mars mission — or arrives too late to matter for the first generation of missions and becomes standard for the second — depends on engineering, budgets, and schedule pressures that are impossible to forecast. The physics is settled. The propellant efficiency is real. What remains is the decision to build it.