Realistically, how small can a nuclear reactor get (fission or fusion)?

Truck sized? Table top? Mr. Fusion? AAA Battery?

Are there other physical limits when we include human safety issues?

Let's say the power range would be anything over 1kW. The higher the better.

I should also specify I'm looking for electrical power.

  • 12
    $\begingroup$ you have to state power production ranges for it. Because 0.1mG tritium is already nuclear reactor which converts its energy into light tritiumkeychains.com $\endgroup$ – MolbOrg Dec 31 '16 at 20:25
  • 1
    $\begingroup$ The power should be 1kW. If you mean kW.h, this mean 3600 J and is a unit of energy, not power. kW/h is a very strange unit ( 1/3600 m^2.kg/s^4) $\endgroup$ – Madlozoz Dec 31 '16 at 21:26
  • $\begingroup$ @Madlozoz Pedant. ;) Edited.. $\endgroup$ – Sam Washburn Dec 31 '16 at 21:38
  • 1
    $\begingroup$ You have IEC fusion reactors which can be desktop computer sized. $\endgroup$ – icc97 Jan 1 '17 at 21:06
  • 4
    $\begingroup$ I love how almost all the answers here beautifully ignore the question asked. The author wants to know how small a nuclear reactor can go, not what can be used in place of one. $\endgroup$ – user64742 Jan 1 '17 at 21:25

Currently there are Small Module Reactors which use fission to generate an electricity output of less than 300 MWe. One of these, the NuScale produces 50,000 kilowatts/hour and is 76' by 15'.


NASA is working on low-energy nuclear reaction(LENR) technology which they hope eventually to use to power cars, planes, and homes. https://www.extremetech.com/extreme/149090-nasas-cold-fusion-tech-could-put-a-nuclear-reactor-in-every-home-car-and-plane

MIT also claims to have made recent breakthroughs in an efficient fusion plant: http://news.mit.edu/2015/small-modular-efficient-fusion-plant-0810

Sadly the last two are theoretical. However as a reality check, it would not be unrealistic to expect nuclear reactors with a large output with the physical size of a home generator in the near future. And possibly the slightly further future may bring into reality AAA battery sized generators.

Human safety issues may be a concern, however NASAs LENR uses a nickel lattice and hydrogen ions and the reaction produces copper. Without more details this does not sound much more dangerous than the acids stored inside acid batteries.

  • 2
    $\begingroup$ using a nuclear reactor in a car would of course mean trusting your average joe with nuclear material. It would also mean trusting auto-mechanics with it. I'm skeptical that something horribly toxic wouldn't get accidentally released. Or even worse, that someone wouldn't buy up a bunch of old cars and make a fission bomb. $\endgroup$ – user4574 Jan 1 '17 at 22:00
  • 2
    $\begingroup$ The LENR uses low energy reactions apparently in the form of hydrogen ions. It does not appear to be horribly toxic and there are better, more powerful ways to build a dirty bomb. $\endgroup$ – LordMartlet Jan 1 '17 at 22:46
  • $\begingroup$ These "small nuclear reactors" are, however, neither particularly small nor do they make economical sense: The Forgotten History of Small Nuclear Reactors. We are a long way from "Foundation"-style mininukes that you can carry around the neck. $\endgroup$ – David Tonhofer Jan 1 '17 at 23:42
  • 1
    $\begingroup$ @user17082 accidental release can be prevented pretty much with making them self contained, well sealed units. Bolt in place, plug in some control cables, and that's all the auto mechanic can and has to do. No different from modern battery packs for say a Tesla or an advanced HEV. The deliberate release is a bigger risk, but you're not going to turn them into fission bombs. A dirty bomb (spreading contaminated dust through a conventional explosion) is what the risk would be, and while serious the contamination can be readily cleaned up. $\endgroup$ – jwenting Jan 2 '17 at 13:16

Use atomic battery instead

To reach 1 kW of power, you can use an atomic battery instead of a atomic reactor.

Atomic battery work by having any quantity of radioactive material (238Pt or 90Sr are popular choices ) in one point, then transform the radioactive energy into electricity (usually,radioactivity -> heat -> electricity )

You can make an atomic battery as small as you want by using Seebeck generator.

To reach 1kw, it's probably better to use a small Stirling engine. The contraption would be the size of a Mr.Fusion

Note that the amount of reacting material can be as small as you want. You only need to take something with a smaller half-life. This, of course, can easily become unpractical

Atomic battery has one major drawback compared to reactor: they get depleted at the same rate, whether you use it or not.

An atomic battery with a Seebeck generator has no moving part and therefore could operate for centuries (with a large enough amount of, say, 202Pb)

  • 3
    $\begingroup$ A Seebeck generator is a very reliable form of nuclear power generation. But you can't really make it as small as you want. The Seebeck effect relies on a temperature difference between a hot and cold side to generate power. The efficiency is also only a few percent. So for example to get 1kW of electrical energy one might need to move 30kW of heat. A radiator that can move 30kW of heat would not be very small. $\endgroup$ – user4574 Jan 1 '17 at 2:19
  • 3
    $\begingroup$ I will confirm that the cylinders for the natural gas powered 1kW sterling generators I have personally seen are about 1 cubic foot in volume. But the radiators and control systems were much larger. All together the whole rack was about 4 x 4 x 6 feet. $\endgroup$ – user4574 Jan 1 '17 at 2:24
  • 1
    $\begingroup$ I always like the Sterling approach, but NASA has decided it has too many moving parts for decades-long reliability in deep space. $\endgroup$ – SDsolar Jan 1 '17 at 6:48
  • 1
    $\begingroup$ @ SDsolar A free piston sterling engine using gas-bearings need only have two moving parts. Apparently two was too many for NASA. The first moving part is a gas-displacement cylinder with permanent magnets fixed on it. The other parts is a flat spring attached to the end of the cylinder. Everything else is just solid state electronics. But in a radiation environment, the electronics would probably be the weak point in the system. $\endgroup$ – user4574 Jan 1 '17 at 21:56
  • 2
    $\begingroup$ @Karl that doesn't mean viable alternatives wouldn't be of interest... $\endgroup$ – jwenting Jan 2 '17 at 13:18

There already are some nuclear batteries around the size of a AA or so, but you'd be hard pressed to power a pocket calculator with one, the trick is to pick a fuel that doesn't emit neutrons or gamma rays as you tend to need quite a bit of shielding (light elements for neutron shielding, heavy for gamma)

Your best bet is something that produced alpha rays as even a Christmas card will protect you. One of the most commonly used alpha sources for power applications is plutonium (which ironically shields you from much of its own radiation). The plutonium gets hot and you can use this heat to drive something like a stirling or thermoacoustic engine. You're never going to get 1kW from a AA sized device without serious heatsinking (budget for 30% electrical efficiency or so).

I see no reason why you couldn't get a kW of electrical power from something the size of say a beer mug (you'd just have to use a much more volatile alpha emitter than plutonium and accept a shorter lifespan)

  • 1
    $\begingroup$ Damned! You beat me by a few minutes! $\endgroup$ – Madlozoz Dec 31 '16 at 21:40
  • $\begingroup$ @Madlozoz Fear not! You're not alone. This happens to us all. Your turn will come when it's your fingers that are the faster. $\endgroup$ – a4android Jan 1 '17 at 3:16
  • $\begingroup$ +1 for mentioning the heatsink restrictions from a small battery producing large amounts of power! $\endgroup$ – Michael Yoo Jan 1 '17 at 6:07
  • $\begingroup$ I have seen them the size of D batteries, but that was 30 years ago and now they seem to be just a bit larger, and arranged in piles like a battery pack.larger. $\endgroup$ – SDsolar Jan 1 '17 at 6:49

ANSWER: Use an RTG - a Radiosotope Thermoelectric Generator. They have been around for years. They are small enough to fit in your garage or basement, and generate significant power.

I have run across them in remote parts of Alaska, presumably for weather stations. The structures are about 5ft by 4 ft by 6 ft tall, and you can hear fans whirring inside and see the satellite antenna outside, all inside a chain-link fence with a warning sign. And not a power line in sight.

Here are some links to relevant information about them:



It is also described in Wikipedia: https://en.wikipedia.org/wiki/Radioisotope_thermoelectric_generator


NASA also uses them in space probes, most notably Cassini. They design the GPHS hot cells to be able to survive reentry even if everything else burns. They are more the size of those square 6 Volt lantern batteries. The Plutonium is surrounded by solid ceramic.

They last for many years, and are perfect for space probes which get beyond the viable range of solar cells.

Here is an official NASA document that describes RTGs: http://saturn-archive.jpl.nasa.gov/files/power.pdf

  • 1
    $\begingroup$ An RTG is not what's usually considered a nuclear reactor, though. It generates heat from radioactive decay, not nuclear fission. $\endgroup$ – Philipp Jan 1 '17 at 2:59
  • 1
    $\begingroup$ You are absolutely correct, @Phillipp. The OP asked for a nuclear source of electrical power and wanted to know how large it would need to be. . This is the generally-accepted method of accomplishing what was asked, along with my personal recounting of finding them in remote parts of Alaska. They could fit in your basement easily.. $\endgroup$ – SDsolar Jan 1 '17 at 6:39
  • $\begingroup$ The OP asked about a reactor, not any kind of nuclear power source. $\endgroup$ – Karl Jan 1 '17 at 23:42
  • $\begingroup$ The last line of the OP is "I should also specify I'm looking for electrical power." - I think that is clear enough. If this is good enough for NASA it should be good enough for the OP. Only the Soviets were stupid enough to launch actual reactors. If you're old enough you might remember one malfunctioning and spreading radioactive waste across Canada. $\endgroup$ – SDsolar Jan 2 '17 at 5:22

The human safety issue for any fission reactor is the neutrons that are produced. Normally this is addressed by a few meters of concrete shielding, which also firms part of the reactor's last-resort containment. A low-power research reactor is often called a swimming-pool reactor because instead of concrete, several meters of water is used. When operational there is a beautiful blue glow from the water immediately adjacent to the core.

It would be possible to dispense with screening and simply keep one's distance. This makes sense for a nuclear-powered spacecraft (reactor at one end of a long pole, crew quarters at the other). Here on Earth it sounds like the sort of crazy that the Soviets might have countenanced, given a reason. (I cannot think of one).

Someone who knows more than me might be able to fill us in on the design of reactors in submarines where space for shielding is obviously a bit tight. (If this info is not classified). Is there a zone around the reactor where crew do not venture except for the shortest possible time in emergencies? Is the immediate outside of the hull an unhealthy place for a barnacle?

  • 1
    $\begingroup$ the main reason for the thick concrete is gamma containment, not neutron containment. A few centimeters of steel or preferably boron is enough for that. $\endgroup$ – jwenting Jan 2 '17 at 13:19

If more is better, and where it can get then I would say 10 tons per 1 GW of electrical power for thermonuclear reactors in space conditions.

By using CNT's as reinforcement/conductors etc and probably it may be less than that. This is for small reactors(few GW power), for bigger reactors it should be better than that with same technologies.

I would not say it is good numbers to use for you space ship if you make plans for it and will to build it in next 10 years, but for space ships, in this century or the next in hard-scifi it may be a reasonable assumption.

For portable devices, it is hard to tell. The answer to the question depends on the field of use of the energy source. Because sources are different and they can be classified/differs by massefficiency, the power they produce, an environment they should work in, should they be human-nearby safe, how long they have to work etc.

As funny energy source, I found recently, C14 based battery.

Half life 5730 years, so it will work few thousand years without a significant reduction in power production. Mass efficiency is not the best - 0.17 W per kg, so a 1kW source will weight about 6 tons. No moving parts. Excellent scalability, if you have enough C14.

For some purposes is an excellent source, for other, it is not so great. So proper answer really depends on where the source has to be used, and the purpose of it.


Polywell fusion scales in an amazing manner

The Polywell fusion reactor has a very curious scaling. Assuming Robert Bussard was right in his hypothesis, the energy output of Polywell scaled by the radius of the reactor to the seventh power, while the "Q-value", that is to say ratio of energy gained versus energy given to run the reactor scales to the fifth power of the radius.

This means that an almost arbitrarily powerful Polywell reactor will be in the order of 3 to 5 meters cubed.


Since the OQ started this question with "Realistically," I think some info on small, demonstrated nukes is on target. The https://en.wikipedia.org/wiki/W54 is the smallest nuclear weapon I know of (true nuclear reaction.)

"All four variants share the same basic core: a nuclear system which is 10.75 inches (273 mm) diameter, about 15.7 inches (400 mm) long, and weighs around or slightly over 50 pounds (23 kg)."

As is, this doesn't fit your requirements, but it does give us an idea of what the minimum amount of fissile material is. A reactor for power needs two key things the weapon lacks:

  1. Moderation to control it, so it produces useful power when desired, rather than exploding.

  2. Something(s) to take the output energy and convert it to electricity.

Both 1 and 2 will add mass and volume, but for small power outputs (by which a mean a few kilowatts), the moderation and power extraction/conversion subsystems ought to be much smaller than in other examples already cited. I assert that the core could be on the close order of a few cubic meters, exclusive of shielding. Small size may hurt efficiency, but efficient operation wasn't specified.

Speculation: My guess is that people who control fissile materials (and what gets built therewith) are more into higher power systems -- megaWatts.

I'm surprised nobody's mentioned reactors (not simply RTGs) that have been put into orbit, mostly by the Russians. Those are small, though larger than the W54: https://en.wikipedia.org/wiki/Nuclear_power_in_space#Fission_systems


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.