Light the Nuclear Candle!

Old technology, new century: Can we ride an atomic rocket to space?

In 1968, over the no man's land of the Nevada desert rose a towering flame hundreds of metres tall. It marked the test of Phoebus 2A, a nuclear thermal rocket: A technology from the past that could yet deliver us to the future.

Nuclear thermal rockets are a strange remnant from the dawn of the Atomic Age, where the Las Vegas strip would occasionally stop, paying tribute to a rising mushroom cloud in the distance, and open-air tests of mankind's most potent technological avatar were commonplace. It was a forgotten age, backlit by the optimism and dread of the atomic halo. It was a time of new fears, new hopes and the seedling potential for man to shrug off the limits of Earth and make way, wings spread, to the stars.

It was also a time when, unthinkable as it seems, the Los Alamos research laboratory was happy to run an open gas-cooled nuclear reactor, flinging superheated hydrogen from the hot core straight into the shimmering desert air. And so, Phoebus 2A let a reactor core roar open to the world, like a rocket.

You simply can't visualize something like that being done now. Why the hell did they do it? Was it sheer mischief?

No, it was a prototype of a deep space propulsion system of spectacular potential, and one that may yet take us to Mars. Let's take a journey through… ominous pause… the reactor core!

1: The hydrogen cooled heart.

Unlike a chemical rocket, which combines fuel & oxidiser and burns them to generate heat and power, a nuclear thermal rocket just heats a light molecular mass substance directly by passing it through a reactor core. The reaction mass (typically helium or hydrogen) cools the core, which heats the hydrogen in turn, letting it expand through a nozzle to generate thrust. Hey, presto! Nuclear rocket.

To understand why anyone would do something so convoluted, we must first understand the Tsiolkowsky rocket equation… no, don't go! I promise it's not difficult!

The purpose of a rocket is to accelerate a mass (the payload) through a change in velocity (‘delta-V’). In the case of a schoolyard bottle rocket, this velocity change is pretty small. In the case of transferring a payload from low Earth orbit to, say, Mars, it's a rather more impressive 3.6 kilometres per second (plus the roughly ten kilometres per second needed to get into Earth's orbit in the first place).

These huge velocity changes are inconvenient, because rockets function on the basis of conservation of momentum: If an astronaut floating about in space pushes an object of his (or, because I get to pick, her) weight away at one metre per second, then she will push off in the opposite direction by one metre per second. If she wants to achieve that effect with something one tenth of her weight, then she needs to fling it at ten metres per second, and so on.

If you replace the thrown toolkit with chemical propellant then you have a rocket, and if you integrate the conservation of momentum terms over a given change in rocket velocity then you get the Tsiolkowsky rocket equation, which tells you how much of your rocket's mass needs to be propellant for a given exhaust velocity and a given change in rocket speed. The unfortunate outcome of this is that it is exponentially linked to the rocket's velocity change, and the exhaust velocity is of prime importance in reducing this.

Easy-peasy. Told you it was simple.

Anyway, this is why for all the fire and sophistication of a SpaceX Falcon 9 rocket, which burns kerosene and oxygen with a vacuum exhaust velocity of about 3km/s, the amount of actual payload it can deliver to orbit is tiny next to the bulk of the rocket & propellant itself: Typically under 4%. If you want this to increase, without doing something quirky like using an air-breathing scramjet, you need your rocket exhaust velocity to go up.

Why not burn hydrogen? Hydrogen-oxygen burning rockets can get exhaust velocities of about 4.5km/s, which is why a lot of deep space missions use them, but hydrogen is a low density fuel that's a pain to handle.

And that's about as far as you can go for chemical fuels. To go one better we need to try something different, which is where our nuclear rockets come in…

Now might be a good time to explain the concept of Specific Impulse (Isp), which we will be referring to quite a bit. It's an efficiency measure obtained from dividing the exhaust velocity by standard Earth gravitational acceleration. In specific impulse terms, a SpaceX Merlin engine achieves 310 seconds, a hydrogen burning engine can get up to 455 seconds and a nuclear thermal engine…?

About 800 seconds!

What this means in practice, if you compare a nuclear thermal to a hydrogen burning rocket, is that with the nuke you can equip a round-trip transfer stage to Mars from Earth orbit that’s half the size of its chemical rocket equivalent, due to lower propellant requirements. Conversely, you could have a transfer stage that’s the same size, but carries twice the payload! This is unnecessary if you’re just interested in flinging down survey robots and scientific experiments, but if you want to haul something bulky like, say, a colony, the nuclear engine shows promise. 

But how?

What you might be thinking is that this is because nuclear engines burn hotter than chemical rockets, but that would be wrong: In fact, due to the need to keep the enriched fuel safely contained and jacketed, nuclear thermal rockets actually run substantially cooler than their chemical equivalents, at 2,700 Kelvin for the nuke versus over 3,400 for the hydrogen burner. So how is it that nuclear engines manage to create high exhaust velocities, and therefore a higher specific impulse?

The reason is the difference in molar weight between the output of the rockets. A hydrogen-oxygen rocket produces water exhaust, whereas a nuclear rocket superheats hydrogen or helium directly. At a given temperature, all gases have the same average kinetic energy in their component molecules, so because kinetic energy at a given temperature is constant, lighter molecules must logically sustain higher velocities. H2 (molecular hydrogen) is a lot lighter than H2O (water) and moves faster when you heat it to a given temperature, allowing for the exhaust to be expanded to high average speed more effectively. -And a faster exhaust means higher rocket efficiency!

So let’s get down to brass tacks… what are the challenges of a nuclear rocket, and what can it give us in return?

2: NERVA and the nuclear sunrise.

We’ll start with a US program of the swinging sixties: The NERVA (Nuclear Engine for Rocket Vehicle Application) was the culmination of a multi-decade program of research into the various sub-systems of nuclear rocket design, manifesting in the testing of several prototypes, such as the Phoebus motor we opened with, and the creation of a design, NERVA, which could tap the prodigious promise of atomic propulsion. 

It also resulted in the NERVA XE, a physical prototype that was just about as close as possible to a flight-ready engine, a 1,140 megawatt nuclear reactor integrated into a mock-up of a nuclear rocket flight engine, which was tested in a simulated space vacuum and achieved a chamber temperature of 2,270 Kelvin. It achieved a specific impulse of 710 seconds (more than twice that of a vacuum- optimised SpaceX Merlin), producing 250 kiloNewtons of thrust in a 28 minute full power run in 1969. 

Overshadowed by the global broadcasting of the moon landings, and its own inevitable secrecy, NERVA XE ran twenty-eight times in total, giving humanity a sly glimpse of a distant spacefaring future that may yet arrive. 

The core was split into an array of hexagonal fuel elements made of a composite fuel matrix, with each element swiss-cheesed with thru-holes within which ran hydrogen gas. The fuel elements were themselves assembled into ‘fuel segment clusters’, an array of six hexagonal elements surrounding a central ‘support element’ with a much larger hydrogen flow path, plus reactor moderating elements. 

The path of the hydrogen coolant was complex: From the storage bottles, hydrogen was first fed around the rocket nozzle, cooling it and pre-heating the hydrogen, some of which was then tapped to run a turbopump that would pump the hydrogen in the first place. For an understanding of this particular chicken-&-egg propellant pumping dynamic, I wrote a separate article on rocket engine design which you might be interested in…

After cooling the nozzle, the gaseous hydrogen would then run up through the central support element before reaching the top of the core, reversing direction and powering down through the fuel elements themselves, superheating it to its final (up to 2,750 Kelvin) temperature before expanding through the rocket nozzle and thrusting the beast through space. 

There were alternative designs, of course: Pratt 7 Whitney’s XNR2000 design, based on broadly similar principles, zoned the core into cooler outer regions and a hot inner core and used a ‘folded’ hydrogen flow path that ran through the outer core, boosting its temperature from 103K-210K to 1,660K and then tucking, rolling and thrusting its way through the inner core where it would leave at a rocket-ready 2,670 Kelvin. 

The USSR, characteristically, had a stranger design again, using ‘twisted ribbon’ fuel bundles whose flat spiralling architecture allowed much better heat transfer. The Soviets, no slouches in nuclear engineering, also varied their core fuel composition in the axial direction along the hydrogen flow-path, changing between fuel alloys as the temperature rose. This, in combination with the high heat transfer efficiency of the twisted ribbons, allowed for a core whose highest temperatures were strongly localised within fuel bundles, allowing the rest of the core to run cooler, simplifying the design. It could also have allowed for hydraulic profiling of the hot hydrogen flow for better overall efficiency. The Soviet approach allowed for easier testing and qualification by allowing for fuel assembly testing in research reactors with a hydrogen irradiation loop, meaning they didn’t need whole-core testing to assess design performance, unlike the US NERVA.

Know and respect thine enemy, for they are not stupid…

So the work has been done and we know that nuclear thermal rockets can run. If we focus on, say, Mars, what can we do with them? 

We've already covered mass: A nuclear candle means twice the payload or half the propellant, approximately, but it can also mean speed, which may be useful if you're sending humans into the high radiation environment of deep space, where charged particles from solar storms can wreak havoc on an unprepared crew. 

Mass or velocity? Nuclear power means that you can have both, but they are tradable in other ways: High payload mass could mean, for example, lots of water for an incipient colony, which would double as radiation shielding.

But if it's raw velocity you want then the difference between chemical and nuclear expands prodigiously. A leisurely one-way trip length of 210 days to Mars in a ‘fast conjunction’ return mission would demand a chemical rocket craft that's three quarters payload, to the nuclear engine's one-half. That's big enough, but try to power through for a fast trip and expense jumps exponentially, such is the tyranny of the Tsiolkowsky rocket equation. A trip of 120 days means that the nuclear engined craft will need to be about 60% propellant, while a chemical rocket shifts up to 85% propellant or more. This means that the amount of travel mass left for the craft, payload etc shrinks to 40% for the nuke engine and under 15% for a chemical rocket.

Get that trip to 90 days by brute force and a 15 km/s delta-V (velocity change) budget and even the nuclear rocket craft is now over 80% propellant, but the chemical rocket is now over 97% propellant, meaning that there's barely any room for anything else on board (though you could use multiple stages to boost things a little).

And remember: This is after you've already blasted these big boys into low Earth orbit.

So, there's value in atomic engines, and some irony too: If we can get over our superstitious fear of the glowing rocks and master the nuclear thermal engine then it becomes a tool to lower the radiation exposure of our future astronauts.

As long as you shield them from the engine itself, that is.

But remember, as a species we stopped developing the nuclear thermal rocket fifty years ago and there must have been a reason.

What's the catch..?

3: Nervous about NERVA.

One pragmatic answer to the lack of nuclear candles thudding through our skies is that we don't really need them right now. The inquisitive robots we're sending to other worlds aren't in a rush, they don't need radiation protection and they're not bringing snacks along for the trip. Things only get tricky when you're sending big, heavy, sweaty meatbags who get precious about things like toilets and clean drinking water.

Even then there might be more conventional solutions. If you can synthesise fuel on a new world by capturing carbon and electrolyzing water, as might be possible on Mars, then you can make your own propellant, at which point the failings of chemical rockets become a little less acute. You need lots of energy to do this of course, so in some guise we'll probably be shipping nuclear reactors around anyway if we want to colonise new worlds, even if it's not for propulsion.

But the allure of a nuclear rocket is unmistakable, and if we keep increasing our distance of travel we'll need to use it one way or another. Are we ready?

Yes and no. The experiments of the US NERVA program, and its Soviet equivalent, proved the concept, but they also unveiled a few challenges which would need finessing away if we are to harness the power of the atom for an interplanetary road trip.

Light the nuclear candle! Convoy coming through!

One of these is materials science. A fundamental problem when running superheated hydrogen past fuel cladding is that hydrogen is the very devil when it comes to corrosion and wear: Its tiny molecules insinuate themselves deep into the matrix of the cladding material, where they cluster at crystal boundaries and form stress raisers, initiating micro-structural cracks. This process, known as ‘hydrogen embrittlement’, is one I encountered myself in the nuclear industry: Large primary circuit pressure vessels would often need extended heat treatment in massive autoclaves to rid themselves of molecular hydrogen if poorly controlled arc welding interacted with moisture in the air or on surfaces.

So as you might imagine, blasting hot gaseous hydrogen at over 2,500 Kelvin straight through a reactor core might present… issues.

One of the main outputs of the NERVA program was the characterization of fuel cladding materials for the extreme environment of a nuclear thermal rocket.

And it’s not just about the hydrogen embrittlement: Nuclear rocket fuel cladding assemblies need to be resistant to high temperatures, melting, evaporation, neutron bombardment, thermal shock and mismatched swelling or expansion. That’s quite a list of requirements!

Fuel cladding materials can be constructed as a coated particle matrix, a composite material, Cermets or carbides. A variety were trialled during the NERVA program, and between 1965 and 1969 the measured rates of cladding erosion by hot hydrogen decreased tenfold as new materials were trialled.

A graphite matrix cladding comprises a graphite substrate, which is a high temperature resistant substance with excellent neutron moderation, peppered throughout with uranium carbide fuel pellets, each with a pyrocarbon coating. The entire assembly is then subject to a Zircium carbide or Niobium carbide surface coating. Graphite matrix fuels were a solid starting place, but endurance in heated hydrogen is lower than many other fuel cladding materials, and rapidly drops after 2,200 Kelvin. 

Composite fuel cladding performs slightly better with high temperatures, but is still fundamentally limiting. Carbide fuel alloys & cladding are potentially much better for high temperatures, but can be unstable in hydrogen, which draws a question mark on their use in nuclear rocketry. 

Indeed, throughout the NERVA program the only completely stable fuel & cladding material that was found was Tungsten-matrix Cermet. ‘Cermet’ literally means ceramic-metal composite, and comprises Uranium dioxide fuel particles in a Tungsten matrix, with a CVD-deposited tungsten surface coating. To create the Cermet matrix a few processes can be used, but one is assembly of a molybdenum can and hexagonal mandrils, into which Tungsten-coated uranium dioxide particles are laid in. The can is then sealed, evacuated, heated and exposed to hot isostatic compression, a high temperature sintering process that compacts the particles and tungsten coating together, merging them and eliminating cracks in-between. Once completed, the molybdenum components can be leached away and the newly-revealed Tungsten-UO2 cermet can be coated with a final Tungsten layer by chemical vapour deposition, forming the fuel element. 

That’s complex, but we want this to survive an atomic furnace at the end of the day.

So how far did they get? Well, the NERVA program demonstrated that a tungsten-UO2 cermet fuel element can run almost a hundred cycles and many tens of hours while keeping cumulative weight loss below 1%. That’s still hardly ideal for anything that runs on nuclear fission, but you could design a Mars bus that way, even a re-usable one… and that’s with 1969 technology! Modern materials science should be able to run straight past that. 

Do this, and we can light the candle that will illuminate the heavens.

4: Lighting the Orrery.

So what do you think: Would you book a ticket on a nuclear cruise ship? And what if the destination was another world?

“Hello everyone and welcome aboard the Irish Ferries candleship Icarus. We will shortly set sail for the Martian colonies of Arcadia Planitia and Hobbiton. Please make your way to your bunk rooms while we finish embarkation. Once underway there will be light breakfast and a drinks reception in the solarium for our first class passengers. I am your pilot-module Michael 57, if you have any requests during your journey please remember to JustAskMike! Welcome aboard.”

Or something like that. After all, a sixth of all commercial aircraft are already leased through Ireland, so why keep that going when we go interplanetary? Tax knows no borders…

Anyway, the nuclear thermal rocket is certainly a tempting solution for such a thing, but not the only one. Compare, for example, the four propulsion solutions in this graph, comparing manned mission options for a trip to Mars with total mission duration set against the crew vehicle mass in Earth orbit. 

We've already discussed nuclear thermal rockets (NTP) and hydrogen burning chemical rockets, but there are two new entrants that demand explanation: Nuclear Electric Propulsion (NEP) and Solar Electric Propulsion (SEP). 

Electric propulsion involves the acceleration of ions such as Xenon in electric fields instead of core-heated gas. The advantage of this is very high exhaust velocities and correspondingly high propulsive efficiency: For example, while chemical rockets max out at a specific impulse of 465 seconds, nuclear thermal rockets could potentially reach 900 seconds and electric propulsion could achieve a dizzying 4,000 seconds… or maybe even more! This means less propellant and more cargo, or a lighter ship. 

Sadly there's a trade-off, which is acceleration. Electric thrusters are crazily fuel efficient but pack pathetic amounts of punch, and that feeble thrust means that it takes a long, long, long time to get to cruising speed. Over savagely long distances and velocities (a pleasure cruise to Saturn, perchance?) that doesn't matter, but for a middle-distance interplanetary voyage like Mars that leaves it outperformed by the big thermal candle for many mission profiles… if you're interested in speed, that is.

Perhaps you aren't. Maybe you're travelling steerage, or just sending bulk goods, in which case you might be happy to wait awhile.

The two alternative electric propulsion systems, NEP and SEP, are separated by their power generation strategies. SEP uses solar energy and is correspondingly puny with heavy loads, making it fine for scientific probes with infinite robotic patience, but poor for humans. An NEP system by contrast uses a closed-circuit nuclear reactor: A recent NASA study looking at getting a 50 ton return payload (or 200 ton one-way) to Mars arrived at a design with a highly enriched lithium-cooled reactor powering a supercritical CO2 secondary cycle which powers four 500 kW Brayton cycle turbines with heat management provided by large radiators. 

Phew! Toasty.

A nuclear electric solution could run at much lower temperatures than a nuclear thermal rocket and avoid hydrogen erosion issues, making it truly re-usable, but to counteract that it may need Xenon gas propellant, which is a hundred times more expensive than hydrogen and unlikely to be synthesised on Mars itself.

Hmm…

Perhaps you could use Argon, a common welding gas, instead? Not quite as efficient as Xenon but orders of magnitude cheaper, which becomes a big deal if you're shipping entire colonies with re-usable reactors.

As is always the case in engineering, it's all trade-offs. Do we go nuclear thermal, nuclear electric or just say “screw it!”, shove on our Biggles goggles and blast our way into heaven the old-fashioned way: By burning stuff.

In the cool and un-inspirational twilight of reality we need to think about Earthly concerns, like the safety factor of flinging entire nuclear reactors, full of highly enriched fuel, into orbit on top of a mighty rocket. Before we do anything big with nuclear rockets, we’ll need to know that the conventional sort that’s lifting it is as reliable as it can be. You wouldn’t want to be downrange of one of these beasts if something went wrong… not with a mini-Chernobyl on board. Qualifying the nuclear candle would be a serious feat these days as well, as the 1960s light-it-up-in-the-Nevada-air approach would raise eyebrows. A simulated, enclosed vacuum testbed would be needed, which would be an engineering feat extraordinaire to start with. 

Still, bigger problems have been conquered before. We’d be fools to dismiss something so useful out of hand, when it might be a key that opens up interplanetary space to us sweaty mammals. 

So, again: Nuclear thermal, nuclear electric or chemical rockets? Can we pick a winner?

It’s an un-answerable question right now, because nobody has any idea what a Martian colonial society would look like, if it’s possible at all, or if we’re aiming at the wrong place. Perhaps it’ll be cities on the moon, or captured asteroids or who-knows-what, in which case chemical rockets are probably A-OK. Or maybe we’ll listen to the proverbs and poison of the nitwits who crowd under the political banner of ‘Green’ and decide that we shouldn’t subject space to nasty creatures like ourselves, and that we should all look to higher things instead, like cultivating our own backyard root vegetables… and using hemp. 

But hopefully we’ll choose to put on our civilizational Big Boy trousers and be ambitious. Red planet, here we come! 

If we do, our propulsion system will depend on advances in technology and how many journeys we’re making. As Elon Musk, for all his eccentricities, has shown us, re-usable stuff beats single-use stuff most of the time, so if we can’t crack the problem of hot hydrogen erosion then nuclear thermal rockets won’t find much use. -Especially since that’s not their only issue; shut down a nuclear thermal rocket and decay heat alone with still be blasting out 5% of max thermal power for a while, which needs effective cooling: A NERVA design can be producing one degree of heating per second even an hour after shutdown!

If the nuclear candle can extend its life from a hundred hours to thousands or more then it’s likely to challenge conventional rocketry in the interplanetary gulf and win, since hydrogen can be synthesized anywhere in the solar system that can support us (by electrolysing water), and you’ll get a lot more bang for your buck if you feed it into a nuclear engine and leave the released oxygen for us meatbags.

Likewise, in instances where you need bulk transit with a reusable interplanetary truck and speed isn’t a factor then perhaps nuclear-electric is the way to go, with a suitably cheap propellant gas. 

“Welcome to Arcadia Planitia! Passengers in steerage, your belongings may arrive on a separate flight…”

The road technology takes us on isn’t obvious and somehow it’s only foggy up ahead: In the rear-view mirror, hindsight renders the landscape serene and sun-kissed, but in front of us lies mystery. Maybe we’ll make it up there into the endless gulf and touchdown, naïve and hopeful, on a cold, red desert world where we can write our own stories, lit more by optimism than sunlight. 

And if we manage that, wouldn’t it be grand if we did it with a reborn relic from the Swinging Sixties, dragged into this century to give succour to a new world?

Your helmet fog clears up, and you look down at booted feet leaving imprints in red-orange dust. Dust that has never touched a human and has only ever seen the light of the heavens, where a new star now shines faintly, far above.

Your eyes adjusting to the gloom and your head a-fizz with new ideas, you salute the good ship that brought you here. 

Light the nuclear candle. God speed!

Images and diagrams used are credit of Eric Proust’s 2021 lecture series on Space Nuclear Power & Propulsion Systems.