Power Conversion Systems

Nuclear reactors don’t produce electricity directly, rather they produce heat, which then is used to power a system which produces electricity through either mechanical work (the heat engine) or chemical work (through various materials properties). All of these systems exploit the temperature difference between the reactor and the heat rejection system of a fission power supply in order to produce power needed for the mission. 

How much power is produced, and how efficiently, with a given amount of heat and system mass depends on which of a large number of systems you choose to select during the system design phase of an astronuclear power supply program, and these choices will affect core temperature, system life, and system cost.

Let’s look at the options briefly, with more information on each system available as research into this fascinating subject continues.

Heat Engines: The Workhorses of Modern Life

The most familiar method of electric power generation is the heat engine, where heat is turned into mechanical work, and this in turn spins a generator. The technical name for these types of devices is a heat engine, which was extensively studied by a 19th century military engineer, Nicholas Leonard Sadi Carnot, who worked out the theoretical limits of the efficiency of these types of engines in 1824. For this, he’s often called the “grandfather of thermodynamics.” These engines, no matter what specific type, have a hot side and a cold side, and use the temperature difference to create mechanical work. The bigger the difference, the more efficient; within certain limits. This temperature difference is why large cooling towers or ponds are used in most power plants: the amount of heat coming out is a fixed quantity, but the amount of cold available could be improved (down to ambient temperature at least) to maximize the efficiency of the system. In greatly simplified terms, the theoretical maximum efficiency of this type of engine is: [hot end temperature]-[cold end temperature])/[hot end temperature] (somewhere around 80%, but this is a matter of some debate). In practical terms, a free-piston Stirling engine can achieve 50% efficiency, while a supercritical-CO2 Brayton convertor can achieve similar numbers – but with many more moving parts to break.

There are three major types of mechanical heat engines in use today, each with their own advantages and drawbacks. Each also has many different variations on working fluid, operating temperature, efficiency, and more.

They are (in order of commonality)

Rankine (Steam) Turbines

The most common type of heat engine for electricity production in the world today, the Rankine (or steam) turbine cycle uses the energy required for a working fluid to shift from a liquid to a gas to extract energy from the heat source and carry it through a turbine through the expansion of the working fluid during boiling. 

On Earth, the most common working fluid is water, but while water is good at carrying huge amounts of heat, at high temperatures it requires incredibly heavy plumbing to prevent ruptured pipes.

In astronuclear design, the water is often replaced with a metal, the two most common being mercury and potassium, although an alloy of  potassium and sodium – NaK – has also been proposed. 

Mercury Rankine System

For the US SNAP-2 and SNAP-8 power systems, a mercury Rankine system was developed, called the CRU. This system worked much like a water-based steam engine, with a boiler, turbine (seen above), and condenser connected to a radiator, but was able to transfer much more energy per unit mass than water could for the same system mass.
Development difficulties delayed the final design and production of a flight-ready prototype until after reactor development had been cancelled, and both the materials and environmental challenges prevented the further development of this technology. More information on this system can be found on the SNAP-2 reactor page, although there's much more to come in the future!

Potassium Rankine System

Potassium has been proposed as the working fluid for several reactors, possibly starting with the SNAP-50 reactor in the 1970s. Another system, the Pegasus, used a combination of a K Rankine system with a Ljungstrom turbine (also known as a disc turbine), a smaller version of what can be seen above. K isn't as efficient as Hg, but is far more environmentally friendly and mildly less corrosive to components. More information will be coming on these systems in the future!

Brayton (Gas) Turbines

Brayton, or gas, generating cycles use a turbogenerator to extract rotational energy from a heated gas. The working fluid remains gaseous for the full cycle, which simplifies the associated plumbing – and can achieve higher theoretical efficiency. The main downsides to these systems are their far higher rotational speed and the larger volume of working fluid needed to extract the same amount of power (due to lower thermal uptake in the working fluid). 

These have been proposed on many missions with high power requirements, with a variety of working fluids from a fairly standard Argon-Neon mix to helium and hydrogen to supercritical (i.e. in between liquid and gas) CO2, but like the Rankine cycle have never flown, and are routinely far behind schedule and over budget at the time of program cancellation.

Stirling Engines

A far older concept than the last two, but one that requires far higher materials and machining tolerances for decent efficiency, the Stirling engine has seen a renaissance in astronuclear power conversion system proposals in the last decades. While it has been proposed since the 1960s, it hasn’t been considered mature until testing through the Advanced Stirling Radioisotope Generator program (a more efficient replacement for current radioisotope thermoelectric generators), and more recently in Kilopower (one of eight pistons can be seen to the left). Fewer moving parts and a higher theoretical efficiency make this a very attractive option for spaceflight, especially for lower-powered systems.

Materials-Based Power Conversion

So, at this point, an interesting question rears its’ head: what if we could convert heat to electricity without moving parts? There is in fact a way, in fact a number of them, but they come with a big trade-off: single-digit power conversion efficiency. This tradeoff was still worth it to every nuclear electric power source designer that has had hardware fly, though. No moving parts, combined with rigorous quality control, means that whatever’s going to break is not going to be your power conversion system. There are two main options for directly converting heat to electricity, with similar names: the thermoelectric effect and the thermionic effect.

The thermoelectric effect exploits how different metals hold onto electrons at different strengths, and also react to heat differently, by running heat along a join between two different metals, and along the seam having one side be heated and the other cooled. This creates an electrical current (how much depends on the temperature difference and the combination of metals used). The main downsides (other than low efficiency) to this process is that for any two given metals, the sweet spot for system efficiency is very small, so unless the reactor is at JUST the right temperature, and unless your radiators are working properly, no power is going to be produced.  

These have been used extensively in Radioisotope Thermoelectric Generators, and come in a variety of material combinations (many more than are listed below, which will expand over time).


A high temperature thermoelectric option, Silicon-Germanium (SiGe) thermocouples were used on most early RTGs in the US, including the SNAP power sources used in Apollo, Voyager, and Galileo missions (among others). While incredibly durable and efficient at high temperatures, they fell out of favor in the 1980s in favor of the potentially more flexible TAGS-85 thermocouple used in later missions. Radioisotope Thermoelectric generators that have used SiGe include:


Powering many notable NASA missions since the 1970s, TAGS-85 (a lead-selenium, and tellurium alloy) is the current workhorse thermoelectric material for NASA. It offers better low-temperature efficiencies than SiGe, and has since replaced it as the thermocouple of choice in most modern missions - and all that use the Multi-Mission RTG. Able to work both within and outside an atmosphere, it has been used to explore everywhere from Mars to Pluto. Some missions include:


Common in commercial applications for energy recapture, BiTe is one of the leading contenders for the European Space Agency's upcoming Americium-fueled RTG. This is a lower-temperature thermocouple than either SiGe or TAGS-85, which not only more closely matches the power output of the slower-decaying Am, but is also more easily commercially available, reducing system cost.


A potentially highly (for thermocouples) efficient option for an advanced thermoelectric generator, skutterudites have been investigated by the Jet Propulsion Laboratory since at least the 1990s as a replacement for TAGS-85, which has shown lifetime issues as the material decays. While it has not been approved for any advanced RTG design, it continues to offer promise for a new generation of RTGs.

Thermionic Power Conversion

Designing and developing any new system is a long, complex process, but with the challenges astronuclear engineering designers face, the problem grows far larger. We explore the processes, facilities and difficulties designers face extensively as part of our ongoing efforts to encourage the development and deployment of these fascinating, mission-enabling technologiesThe next option is the thermionic effect. This is an incredibly ancient concept; known since the days of ancient Greece, and described in detail by Alexander Graham Bell, before the electron was ever discovered. Certain materials, when heated, emit electrons, which then build up a static charge on another material. The tendency for a coal on the end of a stick to attract ash was observed and described in ancient Greece, and this is that principle in action. 

Bell’s experiments were more efficient, since he used a vacuum: an incandescent bulb’s filament will emit electrons and become positively charged, depositing them on the glass of the bulb rather more efficiently than the coal, because there’s no air to capture the electrons and become mildly ionized. Once again, a static charge builds up on the glass, and this then can be drawn off to be used (assuming the positive side of the circuit connects back to the filament). In this case, the thermal difference being used is the filament (which gets quite hot) and the glass (which has a lot of surface area, and rejects heat by radiation of convection). 

This option is the one that has been used in space nuclear power sources, although with somewhat more efficient design of the power conversion system. These power conversion systems use an interesting property found in cesium (which we aren’t going to explore in this post), where it turns into a very unusual state of matter: Cesium-Ryberg vapor. This vapor collects the negative charge from the hot end of the power convertor, and then condenses on the cold end, transferring the electricity as it does so. Usually, a wick is then used to return the cooled cesium to the hot end of the convertor, and the process begins again.

Theoretically, thermionics can achieve greater efficiency for a couple reasons: first, they can run much hotter; and, just as in all power conversion system options, the larger the thermal difference, the more efficient the system is. Secondly, the Rydberg matter state allows for greater conductivity than the metals of the thermoelectric effect, so less energy is lost in conduction (as heat, which makes the system less efficient the less conductive it is, in a couple different ways).

These systems also have another property that makes them interesting to in-space fission power system designers: they can be installed in the core of the reactor itself. This was actually done in the TOPAZ series of nuclear reactors, which flew 30 times for the USSR and Russia, on the KOSMOS RORSAT program. Individual fuel elements were effectively wrapped in thermionic power conversion systems, in a design often called a “flashlight core,” since they look something like the batteries in an old-style flashlight handle.

However, these systems are still fundamentally limited, in that they are so low efficiency, usually in the single or low double digits. This means that if you’ve got a 1MWt nuclear reactor on your spacecraft, you’re going to have to get rid of that 1 MW of heat in order to keep your spacecraft running, but you’ll only get a few kilowatts of power. This is fine in many cases, and can be designed around (a fission reactor’s core doesn’t have to get much bigger to put out a lot more power, and fission and fusion scale up much better than they scale down), but that doesn’t mean that this is an ideal circumstance.

I’ve covered the Soviet astronuclear program (which used these systems extensively) in some depth in a blog post, which you can find below.

Advanced Power Conversion Systems

As we saw from both of the previous concepts, there are theoretical fundamental limitations to both heat engines and material power conversion options; primarily, that the heat is needed to induce another phenomenon, and the difference between the hot and cold temperatures of the system or material define how efficient it can be. Is there a way to get around this thermal limitation? Are there other options available? The answer to the first question is, unfortunately, no, in most cases; but the answer to the second is a resounding yes! While there are many options, there are three common ones:

Magnetohydrodynamic Generator

Magnetohydrodynamics (MHD) is often discussed in spaceflight, but as a thruster, not a power generator. This works in largely the same way that an electric generator and an electric motor are effectively the same thing, in reverse; it just matters if you’re putting in electricity to make the shaft spin, or spinning the shaft to make electricity. In this case, charged particles are sent through a series of magnetic coils, which slow down the charged particles, changing their kinetic energy into a charge in the coils, which is then used as a power source. This, obviously, requires a source of charged particles, though, which leads reactor designers that want to use this power conversion system to use some very different reactor core geometries, including one that we’ll look at more when we cover this subject in more depth: the vapor core reactor, where the fuel is a liquid that is sprayed in a very fine mist within a reaction chamber, allowing the particles to have less physical interaction while undergoing fission. These fission products are highly charged, and moving at very high speeds, so theoretically this can be a very efficient form of power production (and could make a good NTR as well, so we’ll be coming back to this a couple different ways). Work on a vapor core MHD propulsion system has been explored since the early 90s at the Institute for Nuclear Space Propulsion Innovation at the University of Florida, and shows promise; although only a handful of tests examining the criticality requirements of this reactor have been done, so it remains in advanced early stages of design (other experimentation has been done, and continues to be done, though).

Alkali Metal Thermoelectric Convertor (AMTEC)

The second option is an oddity among power conversion systems: AMTEC, or the Alkali Metal Thermal to Electric Convertor. Here, the capacity for alkali metals to carry high negative charge is exploited by running a liquid metal (usually sodium, but potassium has been used as well, at lower operating temperatures) in a closed loop, through convection between a high pressure hot end and a low pressure cold end. Due to the high energy potential difference, this process doesn’t require mechanical pumps, merely a difference in heat. The theoretical efficiency of this system approaches 40%. The fact that this approaches the efficiency of a heat engine, but without moving parts, makes this a very attractive option for space nuclear power designers, but the details get technical rather quickly, so sadly (for the moment) we'll have to wait for more time to research this fascinating concept.