I always hear that NASA has only a limited amount of radioisotope thermoelectric generators (RTGs) remaining. I did some superficial research on how they work and what is needed to produce them and couldn't really understand what is the problem with just manufacturing new RTGs.

So why can't NASA's stock of RTGs simply be resupplied?


3 Answers 3


The Plutonium isotope 238 used in RTGs is highly specialized. It's not produced in large quantities routinely. Not very many radioisotope applications need that much of a highly radioactive isotope, and it's only produced in certain reactors. In fact, there was only one reactor in the USA that produced it. Nuclear stuff is expensive in general and, now that the 1950s period of nuclear optimism is behind us, it's very burdened by regulation.

The Plutonium isotope 238 may be used for RTGs, but not for nuclear weapons. Nuclear weapons mainly use the isotope 239 which can not be used for RTGs.

See Wikipedia for 238Pu and 239Pu.

Most of the USA's supply until recently was a byproduct of the nuclear weapons program in the Cold War. We are (mercifully) no longer building large quantities of nuclear weapons, so RTG plutonium now must be produced as a special job.

For typically political reasons, no strong effort to start making plutonium for RTGs was established until we were nearly out. However, it's now in progress. Ramping up has been slow. The extreme radioactivity and the small concentration of useful product make things harder.

  • 16
    $\begingroup$ I agree this is correct, but some references would make it superb. $\endgroup$ Commented Jul 16, 2020 at 12:06

Any isotope used as the basis for a radioisotope thermoelectric generator (RTG) has to have a short but not too short half life. A half life of several decades is ideal. Such isotopes effectively do not exist in nature. (Tritium, with a half life of 12.32 years, does exist in nature in trace amounts due to generation by cosmic rays, but the half life is a bit too short.)

Plutonium-238 ($^{238}\text{Pu}$), with a half life of 87.7 years, is ideal. While extremely trace amounts of $^{238}\text{Pu}$ do exist naturally due to extremely rare decays of primordial elements, these are atypical decays. The amount of naturally occurring $^{238}\text{Pu}$ is so small that it is essentially non-existent. The only practical source of $^{238}\text{Pu}$ is a breeder reactor.

Breeder reactors are typically used to produce products that can be used in nuclear bombs or in nuclear reactors. $^{238}\text{Pu}$ is not good for either of those purposes. The only practical use of $^{238}\text{Pu}$ is in RTGs and their equivalent. With an apparent surplus back in the 1980s, the US stopped production of $^{238}\text{Pu}$ in 1988. This surplus is now gone, and so is Russia's. Production has started again as of late, but the production rate is very low.

  • 6
    $\begingroup$ Now that you mention the half-life, I guess it also doesn't help that since 1988 about 20% of the Pu-238 produced then will already be gone due to natural decay. $\endgroup$
    – mlk
    Commented Jul 17, 2020 at 7:11
  • $\begingroup$ @mlk and because of that it probably has to be re-refined to be usable today. $\endgroup$ Commented Jul 17, 2020 at 17:57
  • 3
    $\begingroup$ As well as the half-life, you want an element which is an alpha-emitter (and whose decay producets are either fairly stable or also alpha emitters) so that the radiation doesn't impact the rest of your probe, and is easily converted into heat in the RTG. $\endgroup$ Commented Jul 17, 2020 at 18:01
  • 1
    $\begingroup$ Pu-238 actually is used in triggers for nukes. $\endgroup$
    – ikrase
    Commented Jul 20, 2020 at 4:53
  • 1
    $\begingroup$ @CuteKItty_pleaseStopBArking RTGs unlike reactors have no moving parts so are much more reliable $\endgroup$ Commented Nov 15, 2021 at 15:56

It's not just the half-life of the material (short enough to make heat, long enough to have a slow changing energy curve) It's also the type of radiation.

For instance Cs137 is plentiful and easy to separate, and it's a beta emitter, which is shieldable. But its decay products are gamma emitters, which is not shieldable. So nope.

Pu238 is an alpha emitter, which is the easiest stuff to shield... and its decay products are (only) alpha emitters, or stable. It's the right stuff.

The problem is, you have to make it by exposing Neptunium-237 to neutron bombardment inside a reactor. Same thing you do to U238 to make Pu239 for weapons. And you get great gobs of Neptunium-237 when you run almost any reactor, and it's easy to separate if you're already running a PUREX line to extract plutonium-239 for nuclear weapons.

But there'd be no earthly reason to do that, unless you are breeding plutonium on purpose for weapons, and that requires the output of a special reactor capable of 90-day fuel changes. That is a short list consisting of Hanford style purpose-built reactors, CANDU, or RBMK.

Aha... suddenly you understand the importance of the rather oddball, silly and dangerous RBMK in Soviet nuclear strategy.

By the way, loading neptunium into a CANDU reactor for short term neutron exposure is exactly what Canada is doing for us, to help us manufacture some RTG fuel. The Russians, our friends in space exploration, could do the same with their RBMK reactors... and again I suspect that’s the reason Russia still keeps them. Not for Pu238 but Pu239.

I suppose you could also drop a Neptunium rod in a pure civilian BWR or PWR type, and stop it every 30-90 days to change fuel. But it's a huge production to change fuel: you must cool the reactor and remove the top of the pressure vessel, control rod drives and all. The question, is if you leave the Neptunium rod in there for the normal refueling interval of several years, will that overexpose it, have the lovely Pu238 capture another neutron or three, and yield the useless-to-an-RTG Pu239 or the makes-an-RTG-dangerous Pu240 or 241?

I don't know the answer to that question, but I can tell you it doesn't work for weapons Pu239 breeding. Pu240/241-contaminated Pu239 is useless for bomb making, because its spontaneous fission will create neutrons at inopportune times shall we say. That is why we can let countries like Iran have BWR/PWR type reactors; they would have to change fuel every 30-90 days to prevent accumulation of Pu240/241, and everyone with an IR satellite watches their reactors' cooling systems to make sure they are not doing that.

I’ve been asked for a sidebar on what’s bad about Pu240 and Pu241? 240 after all has a decay chain that is all alpha, so what’s the prob? The spontaneous fission releases gamma, that you must then shield, and also creates 2 daughter isotopes out of at least 8 possibilities, and those have their own decay chains, often releasing gamma or beta. It’s out of control at that point. Other than that, Mrs. Lincoln, if Pu240 chooses to alpha-decay instead of split, the subsequent decays are all alpha, which would be alright... but Pu241 can’t say the same. If you’re leaving the Neptunium-237 in the reactor long enough to get Pu240, you’re also getting Pu241.

Aside from the other way Pu240/1 affect bombs, the gamma emitters are problematic for both bombs and RTGs. Because humans have to handle those bombs, sailors have to sleep right next to them, and spacecraft need RTG shielding to be liftable. In fact the Navy uses special Pu239 in their bombs, for the sake of crews. Their U238 “ore” spent far less time in the reactor, so has far less Pu240/1 at the expense of far less Pu239 also, meaning lower yield “ore”.


And then, you have to set up a PUREX line to separate out both the neptunium and plutonium, and IAEA is just gonna love that.

By the way, the reason it's so important to make the correct plutonium isotope in the reactor is that separating plutonium chemically from other stuff is a straightforward chemical task; but separating plutonium isotopes from each other is effectively impossible. It's only possible with natural uranium because it's 3 units apart instead of 1 (235 vs 238), and even then it's ridiculously hard.

  • 2
    $\begingroup$ +1 for the excellent explanation of the importance of chemical separation instead of isotope separation. An explanation (or a link to Wikipedia) of the abbreviations like CANDU or RBMK would be nice. $\endgroup$
    – Uwe
    Commented Jul 18, 2020 at 14:15
  • 2
    $\begingroup$ why are Pu240/241 poison? $\endgroup$ Commented Jul 18, 2020 at 18:05
  • 3
    $\begingroup$ @NooneAtAll Second this. Pu240 is poison to a bomb because it's a neutron emitter that's prone to causing a predetonation (not that it explodes sitting there, but that it explodes after the detonator is fired but before full assembly is reached--this robs the bomb of most of it's power.) Other than being a neutron emitter I don't see how it's a problem for an RTG, though. $\endgroup$ Commented Jul 19, 2020 at 2:15
  • $\begingroup$ @LorenPechtel My first paragraph. Pu238 decays alpha into 2 other things with long half-lives that also decay alpha. Pu240, granted, does the same thing. However Pu241 is a mess; it decays beta into something which decays gamma and sometimes spontaneously fissions, releasing more gamma. If you left your Neptunium in the reactor long enough to make Pu240, you’re also making Pu241. $\endgroup$ Commented Jul 19, 2020 at 4:13
  • $\begingroup$ @Harper-ReinstateMonica That makes the RTG hazardous to deal with, it doesn't poison it. It will still produce power. $\endgroup$ Commented Jul 19, 2020 at 4:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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