This is a subject that's been stewing for a while now. I often see debates in comment sections over whether or not nuclear electric power is feasible in space. Only rarely do those arguing hold the same assumptions about what nuclear power actually means. As a result, these debates rarely convince anyone of anything beyond the stubborn natures of their opponents.
The goal of this post is to briefly cover the range of commercial, military and scientific nuclear power systems ranging from a few kilowatts to over a gigawatt. I will follow up the (hopefully) useful background information in a later post with some fanciful projections and my usual call for unlikely investments in space.
Read on after the break so you can be armed with facts for your next debate thread.
Nuclear energy is produced by the fission (splitting) of certain heavy atoms. This fission produces radiation which becomes heat which is then turned into electricity. The leftover heat and spent nuclear fuel must be dealt with. Shielding must be provided.
I won't get too deep into this subject, but there are several types of radiation. All of these types create challenges for material designs, since most materials become brittle with exposure to radiation. (Would you like to know more?)
- Neutrons are nuclear particles emitted during fission; a certain amount of neutron radiation is needed to start up most nuclear reactors. Neutrons can be either fast (high energy) or thermal (fast neutrons that have been slowed by smashing into a moderator). Neutrons are a form of penetrating radiation; they are a neutral particle so electrical interactions have no effect, which means they can penetrate deep into many materials. Neutrons can also 'activate' other materials; once a neutron has been slowed down by many collisions with atoms, it eventually gets slow enough to be captured. This neutron capture process can form radioactive isotopes of common materials like iron or nitrogen. The best shielding for neutrons is either a lot of hydrogen (usually as water or polyethylene) or layers of neutron reflectors (lead, bismuth, beryllium; see below). It's important to note that the neutron environment inside a reactor must be carefully controlled for efficient operation, and there is definitely a lower limit as well as an upper limit for workable designs.
- Gamma rays are very high energy photons (electromagnetic energy) produced either directly during fission, indirectly after a positron (anti-electron) is released and then annihilated with an electron, or indirectly by a beta particle colliding and emitting bremsstrahlung. Gamma is undesirable in a reactor because it is penetrating, very harmful and can be activating. Gamma rays can trigger the fission of deuterium, for example, causing the release of a moderate-energy neutron. The best shielding for gamma is a heavy metal like tungsten, but often a conductive liner (steel) and a bulk absorber (very thick concrete) are used.
- Other particles (protons, alpha particles and heavier fission fragments) have different typical energy levels but are largely the same as far as a reactor is concerned. They are typically charged, can be slowed or stopped efficiently with metals and eventually become troublesome atoms trapped inside the fuel or coolant. Higher-speed fragments will also emit bremsstrahlung as they slow down, so essentially all nuclear reactors produce some level of gamma radiation.
The simplest fission fuel is an unstable isotope that spontaneously decays. Plutonium-238 is probably the most common example; this is used in RTG (radioisotope thermoelectric generator) units and radioactive heater units on deep space probes. Strontium-90 is another example, widely used in the Soviet Union in space and on Earth as a reliable power source for remote outposts like lighthouses. Some additional possibilities are Polonium-210 (powerful, dangerous, short life) and Americium-241 (long life, relatively high penetrating radiation output). These decay fuels are usually used as a simple source of heat, either maintaining operating temperature for some other device or powering a thermoelectric generator. The ideal unstable fuel would be something that decays only into alpha particles and stable products, producing no penetrating or activating radiation while having a decay rate high enough to be reasonably energy-dense yet low enough to operate for a few decades. No such material is known.
Next is fissile material. A fissile isotope is one that can capture a low-energy neutron and then split. The four main examples are uranium-235 (naturally occurring), uranium-233 (bred from thorium-232), plutonium-239 (bred from uranium-238) and plutonium-241 (bred from plutonium-239 by way of Pu-240). Fissile material is useful for making nuclear weapons, so the production and use of these isotopes is very tightly controlled. Inefficient early reactors couldn't use natural uranium because the fissile content was too low; the U-235 had to be separated (enriched) to produce a fuel that would work properly. The same technology is used to make highly-enriched material for weapons, so again enrichment technology is tightly controlled. More modern reactor designs are more neutron-efficient, so they can use fuel that is less enriched or not enriched at all. Note that highly-enriched fissile material is very dangerous to handle or transport; too much of it in one place or accidentally exposed to neutron flux could lead to a chain reaction, a sudden spike in radioactivity and heat.
Last is fertile material. A fertile isotope is one that can capture a neutron and convert into a fissile isotope, which can then be split with another neutron. Examples are uranium-234 (natural, makes U-235), uranium-238 (natural, makes U-239), thorium-232 (natural, makes U-233), plutonium-238 (artificial, makes Pu-239) and plutonium-240 (artificial, makes Pu-241). Fertile materials are relatively stable; they are not particularly radioactive nor will they do anything dangerous if you put a lot of it in one place. Most of them are flammable metals, but that is a chemical hazard rather than a nuclear hazard; burning U-238 is no more dangerous than burning magnesium (though the results are a bit more toxic). Fertile materials (including natural uranium) are far easier to transport safely than fissile or unstable materials.
The fuel by itself is only part of the story. The full fuel cycle is important to consider. Earth-based commercial power reactors can rely on an extensive infrastructure of mining, refining, enrichment, fabrication, reprocessing and disposal. Space-based reactors will have none of those advantages.
Most commercial reactors and some military reactors are thermal, meaning their fast neutrons are moderated down to an energy level that allows for efficient capture in fissile fuel. Most such reactors require enriched fuel, which means fuel elements would be shipped from Earth until nuclear materials processing infrastructure is established in space. This is politically, economically and environmentally difficult, so Earth-style thermal reactors are not likely to be used in space for a long time if ever. One possible exception is CANDU, a heavy water moderated thermal reactor that can burn natural uranium (and a lot of other radioactives) as fuel. Interestingly, ice on Mars is significantly richer in heavy water than on Earth thanks to atmospheric losses over the eons; this might be a reasonable medium-term approach, particularly since the design does not require massive pressure vessels.
Many research and medical reactors and some military reactors are fast, meaning their neutrons are used as they are produced. Fast reactors are often called breeder reactors, because they turn fertile material into fissile material which is then split for energy. An initial 'spark plug' of fissile material is used to generate enough neutrons to get the reactor going, then the majority of the fuel is natural uranium, natural thorium or some other fertile material. The earliest breeder reactors were used to generate fissile plutonium for the production of nuclear weapons, but current designs using thorium are specifically intended to prevent any application to weapons (proliferation-safe). Small-scale research and medical reactors are used to irradiate materials to make useful isotopes for medical imaging, cancer radiation therapy and RTG power cores. Thorium-based reactors are particularly interesting for space colonization since they could be fueled using rudimentary refining techniques and produce little waste.
Moderators, Coolants, Poisons and Reflectors
The neutron environment inside a reactor is critically important to safe and efficient operation. Four types of materials are present in most reactors and all of them affect how neutrons behave. Many materials have more than one property from this group.
A moderator is some material that can absorb energy from neutrons without stopping them entirely. A coolant is something that can carry heat efficiently and hopefully is not too corrosive or degraded by radiation. By far the most common material in both cases is plain water thanks to its high hydrogen content, excellent heat capacity and reasonable thermal conductivity. Commercial power reactors are almost exclusively thermal, either pressurized water or boiling water types, which use purified light water to moderate neutrons and to carry heat out of the core. Care must be taken that the design is passively safe; that is, if the coolant were to boil suddenly then the reactor should naturally reduce its power output without intervention. For an example of passive safety, check out TRIGA (training, research, isotopes, General Atomics) reactors; operating safely since 1958 these are the only reactors licensed for unattended operation.
The two other moderators in common use are heavy water (water made of oxygen and deuterium) and graphite (pure carbon). A third used in a handful of experimental and military reactors is lithium-7 (with or without beryllium), typically as part of a molten salt.
The main heavy water reactor design is CANDU, which uses it as both moderator and coolant. Derivative designs use separate light and heavy water systems, with the heavy water providing mostly moderation and the light water providing mostly cooling. Heavy water is used because the hydrogen already has an extra neutron and is much less likely to capture another one. It does happen, so heavy water reactors produce small amounts of tritium.
Graphite always uses a separate coolant since it is a solid. Graphite was used in the first reactor (the Chicago pile) and in many others since then due to its stability, mechanical strength, incredible temperature tolerance and ready availability. As a solid, graphite is susceptible to lattice defects called Wigner energy; this led to the Windscale fire before it was understood, though most modern reactors operate above the annealing temperature of carbon so this is not a concern.
Beryllium is a suitable moderator if you only look at physics. Unfortunately it's expensive and extremely toxic, so it is not normally used on its own. In a mix with lithium-7 and fluorine it forms the coolant/moderator FLiBe used in molten salt reactors.
Fast reactors need to have as little moderation as possible (or at least a predictable and controllable amount) inside the core. That means they need to use coolants that are poor moderators or are neutron-transparent. Common materials are sodium and lead (yes, lead; it's great at absorbing gamma radiation but it tends to reflect neutrons). Some molten salt reactors are also fast reactors and may use zirconium and sodium fluorides instead of beryllium and lithium fluorides in the salt mix. It's worth noting that some graphite-moderated reactors are cooled with molten lead or sodium, since using a coolant that is a poor moderator means the reactor's behavior is more predictable during transient problems with coolant flow.
Carbon dioxide has been used as a coolant (with moderating properties) in the past, and may be used again as a supercritical fluid. This requires fairly high pressures, but learning how to handle supercritical CO2 would have useful applications for cooling or refrigeration elsewhere in space.
Helium has also been used as a coolant and is proposed to be used in some very high temperature reactors as both the coolant and the working fluid for the turbine. Because it resists activation, if a reactor core uses fuel elements that trap their own fission products then the helium can pass directly through the core and into the generator turbine with no intermediate heat exchangers; this requires very high temperature turbine materials but leads to superior efficiency and compact, simple design.
Zirconium is nearly transparent to neutrons. Many fuel assemblies use Zircalloy, an alloy that is at least 95% zirconium, to allow fast neutrons to escape the fuel pins and to allow moderated neutrons back into the fuel to trigger more fission. A common fuel is uranium zirconium hydride, with zirconium alloyed for structural strength and hydrogen adsorbed for inherent moderation.
A poison is some material that absorbs neutrons very efficiently. Examples include lithium-6, boron, hafnium, xenon-135 and gadolinium. These are used in control rods and safety systems or are produced naturally by nuclear reactions within the core. Over time, neutron poisons build up in the fuel; the dynamics of this are complex but neutron poisons are the main reason why uranium fuels only burn about 2% of their potential in one pass through a reactor. The poison byproducts have to be removed for the fuel to become usable again. Xenon is the most important of these over short timescales.
Hafnium, boron and gadolinium are common materials for control rods. These devices allow operators to precisely control how many neutrons are flying around at a given time inside the core and can also be used as an emergency shutdown device. Control rods may be suspended above the core by electromagnets; during a loss of electrical power the rods will naturally fall into the core and stop primary activity. Soluble boron salts are used as an emergency shutdown tool in water-moderated reactors; the salt is injected into the moderator or coolant loop, causing an immediate and dramatic reduction in neutron flux and stopping the reactor's primary activity. Radioactive byproducts will still produce significant heat and radiation for hours to days, so additional safety features like auxiliary cooling are required.
A reflector is a material that reflects (elastically scatters) neutrons. Primary examples are beryllium, graphite, steel, lead and bismuth. This is another reason why graphite was used in early reactors: a layer of solid graphite blocks around the outside of the pile reflected neutrons back into the core, reducing the required size of the core and reducing the required neutron shielding.
Many reactor designs intended for use in space rely on controllable reflectors rather than controllable poisons; the reactor core would be safe (subcritical) by design, only able to operate when neutron reflectors were properly placed. That allows a reactor to be launched before activation, meaning the potential radioactive release during a launch accident would be minimized.
Some other designs use reflectors to boost reactivity near the end of life for a given batch of fuel, or otherwise as an alternative to poisons for control. An example is the SSTAR design, which would use a movable reflector to move the active region of the reactor through a fuel load over the course of 30 years rather than refueling every ~18 months. If the reflector were to fail then the reactor's output would taper off to nearly nothing over a few days. By relying on reflectors rather than poisons, the reactor requires a lower level of neutron flux to operate and can use less efficient (less or not enriched) fuels.
Turning heat into electricity
Once you have a steady supply of heat, you have to put it to use somehow. The laws of physics are singularly unforgiving about energy conversion. For every useful unit of electricity produced you will have to deal with two to five units of waste heat in any practical design. Less efficient options are always available.
In space we don't have access to free-flowing rivers or oceans of water to use as coolant; without conduction or convection we can rely only on radiation. Thermal radiators are significantly more efficient at high temperature, so the higher our core reactor temperature the better for a free-flying spacecraft. (Radiative output scales as the fourth power of temperature, so a small increase in temperature causes a very large increase in radiator output.) The temperature limit for a reactor is usually based on either the primary coolant or the fuel material, around 900-1000 °C for zircalloy cladding and possibly higher for ceramic or carbide fuel elements. Molten salt or gas-cooled reactors could go higher, while water-cooled reactors are a fair bit lower. (Water-cooled reactors use water at high pressures, so the boiling point of the coolant is typically several hundred °C.) I won't get into the physics and mechanics of radiators here other than to say they are similar to solar panels in terms of areal density, pointing and deployment. The size of a radiator system depends very strongly on the temperature of the coolant and whether there is a large hot object (like Earth) nearby.
For a surface base with access to a large thermal mass (dirt, ice, etc.) there may be the option of process heat. Some of the waste heat from the reactor can be used to do useful work like melting ice, heating greenhouses or powering thermochemical reactions like the sulfur-iodine process for producing hydrogen. From the perspective of the electrical generation system this is still waste energy, but these uses increase the overall efficiency of the system. This kind of cogeneration greatly increases the required radiator area in free space, so although it seems counterintuitive it may not be mass-efficient to use waste heat for chemical processes on an orbital station. Rather, it may actually take less mass to produce electricity (at 20-30% efficiency, but with high-temp radiators) and use it directly in electrochemical processes vs. thermochemical processes. Each individual mission / craft / architecture is unique and may come down on either side of the line.
So, with a source of heat (reactor coolant loop) and a sink of heat (radiator coolant loop) we can put a heat engine between the two and extract useful energy. The most basic approach is to use the thermoelectric effect (like a Peltier cooler), directly converting heat into an electric current. These devices typically have no moving parts and are highly reliable, but are poorly scalable and only modestly efficient. RTGs use these, as have some flown reactors on Soviet satellites.
By far the most common method on Earth is to use a steam turbine in the Rankine cycle. Heat from the reactor loop boils water into steam in a steam generator, which is passed through a turbine to rotate a shaft. The depleted steam is recondensed into water, passing low temperature waste heat into the cooling loop. This would be extremely inefficient in space as the low waste temperature would require enormous radiators.
A promising technique is to use the Brayton cycle in a reactor with a gas coolant. The most likely of these is helium, since it is very stable and nearly impervious to neutrons. A space-optimized Brayton cycle reactor (see for example project Promethius) would circulate helium through the core and pass it directly through the turbine, with no intermediate loops or heat exchangers. This is possible only because helium does not become radioactive inside the core, but it also requires that the fuel elements contain all fission products; any fuel leak would contaminate the turbine. A cycle using steam without a condenser and boiler is also possible.
Surface bases with abundant heatsink potential could use a Combined cycle. This is a high-temperature Brayton cycle turbine whose waste heat is still high enough to run a Rankine cycle turbine of one or two stages. The Rankine cycle exhaust heat is quite low temperature and would have to be rejected into a body of water (or some other liquid) or pumped into the ground like a reverse geothermal system. The best case would be a mixed-use system that provides electricity, industrial process heat for thermochemistry and ice melting, and life support heat for maintaining livable habitat conditions. Using an array of greenhouses as your low-temperature radiator system would be ideal. The drawbacks of a system like this are complexity, need for available heat sinks and the fact that each part of the process relies on all other parts maintaining a certain pace. If you want to have electricity while your industrial processes are not running then you need an alternate heat sink to replace those processes.
Dealing with waste
Nuclear reactions produce radiation. Some of that radiation ends up activating parts of the reactor, which means those parts become radioactive themselves. Pumps, valves, pipes, pressure vessel walls, all of the structure in the core of a reactor will become radioactive over time. This material generally can't be reprocessed into a nonradioactive form. (It's possible but would be extremely expensive.) This is usually low to medium grade nuclear waste and the usual solution is to slag it, encase it in concrete and bury it. That probably works for surface bases on bodies with no 'weather' cycle, but it would be a no-go for active worlds like Titan / Io or icy worlds like Europa. Even then, there has to be some standardized way to indicate to future generations that there is something dangerous buried there. For craft and colonies that can't bury their waste, they would have to find some place to send it safely. This remains an unsolved problem on Earth; perhaps a waste repository and reprocessing center on the moon might some day be viable, provided shipments of waste are ever allowed to be launched.
The fuel itself produces radioactive byproducts as a result of fission. These are mostly actinides, but there are some radioactive gases like iodine as well. On Earth we generally store fuel elements indefinitely in cooling ponds or eventually in dry casks. Fuel elements can be reprocessed, meaning the component materials are separated, byproducts are filtered out and the repurified fuel is recast into new fuel elements. The actinide wastes can be burned in certain types of reactor (usually the same sort that can burn thorium, but some fast spectrum reactors are designed for waste destruction). The old liners or shells and any equipment used in fuel processing will generally be considered high-grade nuclear waste; this is treated much like other types of waste but will be radioactive for a much longer time due to contamination with radioactive isotopes. Fuel reprocessing facilities are a proliferation concern because they allow for the extraction of weapons-grade plutonium from spent uranium fuels. Thorium cycle reactors would be politically easier because it is far more difficult to get anything of military interest out of the fuel.
Radiation from an operational reactor is damaging to people, electronics and structures. Shielding must be provided to mitigate this damage. Earth reactors solve this problem using cheap, bulky, heavy material in abundance. Usually the reactor core is placed inside a containment building; the building is a thick stainless steel liner and several meters of concrete all around. Openings usually take sharp turns so there is no line of sight from the core to the outside world; radiation doesn't turn corners. (It does scatter, so it's still not simple.)
Free-flying reactor designs don't have to worry about contaminating a planet full of voters during a system failure. These usually have the reactor at one end of the ship on a long truss, with a small shield plug (a shadow shield) that protects the rest of the spacecraft. Ships like these are easy to see coming if you have gamma detectors. They are great for deep space exploration, but they make bad neighbors and are difficult to handle for docking maneuvers since a small misalignment could kill everyone on the other ship.
Possible scenarios - surface base
Let's look at the simplest case first. This is a manned surface colony with basic industry already online. Base metals (iron, nickel, aluminum) and bulk material (dirt) are available. First the coolant system is built (or installed) and tested. Next a containment pit is dug, then lined in concrete/sintered or pressed regolith/etc. Nickel-iron (simply iron from here on) blocks are piled up like bricks and welded together. A self-contained core unit is assembled on Earth and shipped in one piece, placed into the pit and connected to the radiator system. The pit is covered with iron sheets or beams with a layer of concrete/sinter/etc. then buried. The core unit is not activated until it is installed, so it is not radioactive and has no unusual handling restrictions. It would be designed to run for 20-30 years unattended, with no maintenance access possible; it would probably be limited to a few tons mass at most (~6-8t; 300-500kg fuel mass) and up to a few hundred kilowatts of electricity. New core assemblies would be shipped about every decade to maintain redundancy, more often if the colony's energy needs are growing. Cores would be in the few hundred million dollar range (plus shipping); comparable cores on Earth can be built for tens of millions but they don't need to survive a reentry accident and can be repaired on-site. Lifetime power generation (20 years, 95% availability) would be about 33 GWhr of electricity.
The whole assembly would be several meters underground, safe to stand above while operating. A coolant failure would leave the reactor hot but safe, which means the coolant system could be rebuilt or replaced without needing to do anything to the reactor core. In the event of a serious problem like a core meltdown, any released radioactive gases would escape into space or be diffused through the (already unbreathable) atmosphere. Particles could be a bigger problem; on Mars they would be swept away in the next dust storm but on the Moon they would likely stick around for a while unless they were small enough for electrostatic scattering. Still, no crops would be contaminated.
The next step would be an accessible reactor core that can be refueled. Fuel elements could be shipped from Earth or manufactured locally. The containment structure would not be much different, but the core could be bulkier; this would allow for things to be shipped in pieces and assembled on-site. Telerobotics would be ideal for this work, but the initial construction could be safely done in person. If the local industry is capable of building small superalloy pressure vessels then something like the CANDU approach can be used, where small tubes with fuel run through a large 'tub' of moderator+coolant at manageable pressure. Regardless, a gigawatt-sized pressure vessel is a tall order for local industry (many nations on Earth couldn't build a reactor pressure vessel today) and for in-space shipping; one way or another the approach will have to be modular and scalable. Perhaps an array of many reactor cores will feed a small number of high-power turbines. Core units will likely be in the range of a few hundred kW to about one MW each (5-25t including core coolant but not turbines).
This modular approach would allow the colony to transition into locally-manufactured fuel elements and other parts. These might initially be reprocessed fuel from earlier cores or they could start right away with locally mined material.
Beyond that, once the colony has the capacity to make high-performance turbines, pumps, pressure vessels, fuel assemblies, etc. then they will essentially be self-reliant.