Andreas PersboProducing Tritium in North Korea

Jeffrey has asked me to post a blog written by Hugh Chalmers, who works with me here in London. It covers possible ways in which North Korea could produce tritium for its nuclear weapons programme. It first appeared in Trust & Verify, which is our centre’s quarterly publication.

I’m not going to plug Trust & Verify too much, but if you’re interested in signing up for e-mail delivery, you can do so here. Our next edition, scheduled for July, will continue to feature lead articles on conventional arms control, with an emphasis on Open Skies.

Producing tritium in North Korea
By Hugh Chalmers

2016 has been a busy year for North Korea’s nuclear weapon programme. Having started it off with a bang on 6th January with a fourth nuclear weapon test, the programme has since rattled through a successful test of the Unha space launch vehicle, static tests of both liquid and solid rocket boosters, a static re-entry vehicle test, a few unsuccessful KN-08 flight tests, and a slightly more successful KN-11 flight test (at last count).

While Kim Jong Un has defied expectations and taken his finger off the test button during the Seventh Congress of the Workers’ Party of Korea, its not clear how long he will keep it off. It isn’t easy to predict when North Korea is going to test, and despite North Korea’s newfound love for revealing photographs it still isn’t easy to give precise assessments of North Korea’s current nuclear capabilities.

However, it is easier to guess where their programme is heading. Pyongyang’s claims to have tested a ‘miniaturised hydrogen bomb’ in January are a clear statement of intent. The regime aims to develop an arsenal of sophisticated nuclear weapons that draw upon nuclear fusion to some extent, even if they may have exaggerated their successes to date. To do this, it will need a reliable supply of tritium. Before Kim Jong Un presses another button, now seems like a useful moment of calm to post some (very) wonkish thoughts on North Korea’s tritium supplies (originally published in VERTIC’s Trust & Verify) for discussion here.

About tritium
Tritium is an isotope of hydrogen and a key fuel for nuclear fusion – the process underpinning all ‘hydrogen’ bombs. Like fission, fusion produces a large burst of energy distributed between electromagnetic radiation, subatomic particles, and product nuclei. However, unlike fission, nuclear fusion does not happen spontaneously: atomic nuclei contain positively charged protons that, like identical ends of a magnet, repel rather than attract each other. Light nuclei have to be forced into fusion by heating them to extreme temperatures, such as those found inside an exploding fission weapon.

While some light atomic nuclei can fuse in the correct conditions, one combination of nuclei seems particularly suitable for developing hydrogen bombs. If a deuterium nucleus (consisting of one proton and one neutron) can be made to fuse with a tritium nucleus (consisting of one proton and two neutrons), this ‘DT’ reaction generates a powerful burst of electromagnetic energy along with a highly energetic neutron and a helium-4 (He-4) nucleus.

Once a mixture of deuterium and tritium is brought above a certain threshold temperature, nuclei can fuse at such a rate that the energy released can keep the fuel hot enough to keep the fusion reaction burning. Importantly, this threshold temperature and density can be easily achieved in the core of a fission weapon.

Deuterium is relatively simple to acquire. It is safe to assume that North Korea can extract deuterium from seawater domestically by electrolysis or distillation. Tritium is a different matter. Natural tritium is almost impossible to come by: it is produced very rarely by spontaneous fission of uranium, or by the indirect interaction of cosmic rays and nitrogen. Tritium is also radioactive, decaying into helium-3 at such a rate that any stockpile is reduced by approximately 5.5% every year. North Korea cannot rely on any one-off acquisition of tritium: it needs a reliable and repeatable source of tritium to sustain a boosted arsenal of nuclear weapons.

Importing tritium
North Korea might try to import tritium, which has a number of civilian applications. Pakistan managed to get its hands on tritium and tritium handling facilities via a German company back in the late 1980s, but thankfully the international approach to nuclear trade control has changed significantly since then. The Nuclear Suppliers Group (NSG) have placed tritium and tritium-producing equipment or technologies on its trigger lists. Any transfers of tritium-producing equipment, or anything more than a few milligrams of tritium, should not be authorised without credible assurance that such transfers will not contribute to nuclear proliferation. Thanks to UNSCR 1718, this restriction equates to a blanket ban on all such exports to North Korea.

While North Korea would struggle to acquire tritium from the major commercial exporters in Canada, Switzerland, the US, and France, it might have better luck from noncommercial sources. While Pakistan was illegally importing tritium from Germany, it also reportedly received tritium directly from China. Israel also transferred tritium to South Africa in exchange for Pretoria looking the other way while their yellowcake was used to generate weapons-grade plutonium in Dimona.

(Aside: Nic von Wielligh’s fascinating book The Bomb: South Africa’s Nuclear Weapons Programme explains how South Africa smuggled four five-gram cylinders of tritium from Israel in a tea chest, packed away in hand luggage on a South African Airways commercial flight. What was left of this smuggled stockpile – which South Africa’s programme didn’t even need at the time – was eventually declared to the IAEA in the 1990s and used to make numbers for cinema seats that glow in the dark.)

I find it hard to imagine which country would be both willing and able to export tritium to North Korea. China is a member of the NSG, and seems to be cracking down on sanctioned exports to its neighbour. Any exports of tritium to North Korea are probably a thing of the past. The same could be said for Pakistan, whose bid for recognition with the NSG would be ruined if any past or present tritium exports to North Korea came to light. It is far more likely that North Korea is looking internally for a reliable source of tritium.

Breeding tritium in situ
The most direct method for injecting tritium into a fission weapon is to breed it within the nuclear weapon itself. Lithium-6 deuteride can easily be incorporated into the fissile material core of a nuclear weapon, where it would break down into tritium and helium-4 under the bombardment of neutrons generated by the weapon’s fission reaction. The tritium and deuterium can then fuse, releasing neutrons that can boost fission in the surrounding layers of fissile fuel while simultaneously breeding more tritium from the remaining pockets of lithium-6 deuteride.

This ‘layer cake’/’alarm clock’ approach to hydrogen bombs was explored by current nuclear weapon states as a simple and robust way of incorporating fusion reactions into a fission weapon. However, none of these states settled upon this design as a permanent solution to boosting nuclear weapons. Instead, lithium-6 is reserved only for the second reservoir of fusion fuel in two-stage nuclear weapons. This may be because each atom of tritium bred from lithium consumes a neutron that could otherwise go on to prompt fission. Speed is everything in the detonation of a nuclear weapon, and if neutron consumption by lithium-6 delays the cascading fission reaction, it may cancel out many of the benefits of boosting – such as the opportunity to shed heavy tampers and neutron reflectors from the core.

Given North Korea’s claim that it has ‘proudly joined the ranks of nuclear weapon states’ possessing hydrogen bombs, it seems unlikely that they would settle for an obsolete design eventually discarded by all nuclear weapons states. It seems more likely that North Korea will have to turn to its nuclear reactors to generate tritium.

Tritium breeding in nuclear reactors

Tritium production in the 5MWe gas-graphite reactor
The 5MWe gas-graphite reactor at Yongbyon is the source of nearly all of North Korea’s weapons-usable plutonium, and it may also be the primary source of its tritium too. The UK used a similar type of reactor (the Chapelcross MAGNOX reactor) to generate tritium for its nuclear weapon programme, and it is possible that North Korea is doing the same.

Lithium-6 again plays a central role here. The US currently maintains its tritium stockpile by loading lithium-filled ‘Tritium Producing Burnable Absorber’ (TPBAR) rods into the fuel channels of a commercial pressurised water reactor (PWR) at Watts Bar in Tennessee. TPBARs are removed after an 18-month irradiation cycle, and subsequently broken down in a dedicated handling facility to extract tritium produced by the Li-6 + n → T + He-4 reaction. According to an article in Science & Global Security, a PWR could conceivably generate between one and five kilograms of tritium each year per gigawatt of thermal power, if it was optimised purely for tritium production. If the lithium load were reduced so that the reactor can operate normally, this generation rate would drop to between 30 and 70 grams.

These figures probably don’t apply to gas-graphite reactors, and without more information on how North Korea might construct, load, irradiate, and process any tritium-producing fuel rod substitutes it is hard to estimate how much tritium they could generate. If one assumes that the gas-graphite reactor (with a thermal power of only 20 megawatts) can generate a few grams extractable of tritium per year without interrupting normal reactor operations, North Korea could have generated around ten grams of tritium between 2003 and the 2007 shutdown. 38 per cent of this would have decayed by now.

Tritium production at the IRT reactor
With this in mind, North Korea is likely to look to its pool-type research reactor to bolster its tritium stockpile. The Institute for Science and International Security (ISIS) recently reported that this reactor might be operational, fuelled by domestically enriched and fabricated uranium fuel. If this is correct, the IRT reactor will be an attractive source of tritium. After all, it is designed to irradiate material samples with moderated neutrons to generate medical isotopes.

The IAEA research reactor database suggests that if the IRT is refuelled at its former enrichment level (80% enriched uranium), it can generate a maximum flux of around 8 trillion (8×1012) neutrons per second, over one square centimetre. One cubic centimetre target of lithium-6 (with a density of 0.535 grams per cubic centimetre) could generate four nanograms of tritium for each second it is exposed to this maximum flux. According to ISIS, this reactor typically operates only 60-70% of the year. Over an eight-month operating cycle, this cubic centimetre target of lithium-6 could generate about 80 milligrams of tritium. To generate three grams of tritium (an approximate amount used in modern boosted weapons), North Korea would therefore have to irradiate approximately nineteen grams of lithium-6 over an eight-month operational cycle.

It is important to note here that these estimates do not take into account a number of important factors. First, it is not clear exactly how much lithium North Korea can access, and how much it could load into its reactors. Lithium reacts strongly with water, and any protective alloys (along with space in the channels to keep it cool) will limit the amount of lithium that can be irradiated.

Second, the available flux of neutrons will vary along the different experimental channels, and the reactor may not always be operated at full power. Not all of the IRT’s irradiation channels will pass through the core (where the neutron flux is highest). The IRT-2000 reactor in Bulgaria is similar to the one in North Korea, and only four of its twelve vertical experimental channels pass directly through the reactor core.

Finally, it is not easy to extract tritium once it has been generated. The US tritium production programme had trouble tackling the permeation of tritium from TPBARs, with each tritium-producing rod losing approximately 4.2% of its generated tritium into the reactor. Similar fractions may also be lost when targets are broken down for tritium extraction in hot cell facilities – that North Korea may or may not currently have.

Final thoughts
Nevertheless, the IRT reactor can conceivably generate a significant amount of tritium for North Korea’s nuclear programme. A single target containing nineteen grams of lithium-6 (equivalent to a slug roughly 3cm in diameter and 5cm in length) could potentially generate enough tritium for a single nuclear weapon in one yearly operational cycle.

Assuming that only four of the IRT’s experimental channels are located within the reactor’s core, North Korea would be able to generate enough tritium for twenty ‘DT’ boosted nuclear weapons per year by irradiating five such slugs in each channel. While fuelling the IRT reactor for tritium production would unavoidably divert enriched uranium that could otherwise contribute directly to North Korea’s arsenals, ISIS suggests that the IRT would only require 7.5 kilograms of highly enriched uranium each year to operate such a cycle.

All this is to say that North Korea is not lacking options for generating a sustainable domestic supply of tritium to support an arsenal of DT boosted weapons. While any fifth test might provide a more potent demonstration of their domestic supply of tritium, the rough calculations above suggest that the world might have to get used to hearing a lot more about North Korea’s ‘H-bomb of justice’.


  1. Steven Hayden (History)

    The successful nuclear and rocket scientists of DPRK are recognized as heroes and given lifetime job security and lifetime respect. With such intense motivation it is not surprising that they are successful. The benefits of the nuke program are permanent.

  2. tobiaspiechowiak (History)

    Very interesting topic! I was always missing a bit a discussion about this core element for the functionioning of modern weapons.
    It can’t be stacked like plutonium or HEU so the logistics is more difficult.
    An interesting question is also how other countries replenish their tritium supply.

  3. Pavel (History)

    Bob Alvarez had some interesting comments on the graphite reactor [here]( As for IRT, Russia last supplied fuel for it in 1991, so North Korea would have to produce its own 80% HEU. Not impossible, of course, but this may warrant a closer look.

    • HLC (History)

      Thanks Pavel for pointing me towards Bob’s comments. I’ve been trying to get some information on how the UK did it at Chapelcross, but so far no luck (unsurprisingly)!

    • rwendland (History)

      HLC, I’ve not come across technical details of Chapelcross tritium production. But Chapelcross could produce 173g of Tritium a year according to Nuclear Wastelands by Arjun Makhijani et al (ISBN 9780262632041) page 419. This info is cited to “Simpson 1986, pp 201, 202”, but I cannot work out the exact source from the fraction Google books shows, though almost certainly this is John Simpson:

      However there is a discrepancy in that the cite above says only one of Chapelcross’ four reactors was used for Tritium production, But the document below about decommissioning claims two reactors were used: “Tritium was also produced in the isotope cartridges irradiated in two of the Chapelcross reactors.” Maybe production was increased at some point:

      The post-processing of the cartridges was done in the Chapelcross Processing Plant (CXPP), which operated from 1980 to 2004. If you google for CXPP there are quite a lot of documents on the web mentioning decommissioning CXPP; maybe some of those would give some technical info.

  4. Sean Wain Dunlevy (History)

    I know that Lithium 7 also helps along the primary (eg the 40% lithium 6 miscalculation) so what % of natural, lithium 7 would work (or not work). I know it’s only a very minor point but could it just be a shortcut? The term ‘weapons usable’ is another term of interest. I’m guessing that prompt critical would require a higher mass unless far, far more neutron sources were used. Is this another shortcut? No, I’m not in the market to make a bomb, I’m just wondering that with asymmetrical warfare; IF they had at least one highly trained individual, what is the shortest path between 0 & small (yield) nuclear weapon. If it fits into a truck then that IS deliverable (presuming it isn’t searched for). Not trying to worry anyone, but we presume that only HEU works but how many zipple-type centrifuges are needed to make weapons usable material. I’m just looking at the bad guy point of view. Luckily, daesh appear to have lost or killed any experts in chemistry & biology, so the chances of such persons existing is VERY low.

  5. Allen Thomson (History)

    For the sake of completeness, it should be pointed out that tritium production based on a high-power proton accelerator has been studied in the US since the ’90s. See, e.g.,

  6. Gregory Jones (History)

    The 5 MWe gas-graphite reactor could only produce a few grams of tritium per year using natural uranium fuel. However, operations at Hanford showed that if the fuel enrichment is increased to just 0.95%, the reactor would be able to produce about 17 grams of tritium per year while still producing about 5 kilograms of plutonium per year. See my paper “The Implications of North Korea Testing a Boosted Nuclear Weapon”

  7. robgoldston (History)

    A simple way to look at this problem is that from the point of view of reactor neutronics, you can trade about 80g of weapons-grade plutonium production against 1g of T production. So if someone has a reactor that can produce 8 kg of WG Pu per year, it can instead make 100g of T per year, or 4 kg of Pu and 50g of T. A graphite or heavy-water reactor running on natural uranium can in principle make about 350g of WG Pu or 4.4g of T per MW(th)-year.

    • Gregory Jones (History)

      You can’t make the trade between plutonium and tritium using natural uranium fuel since with such fuel the U-238 is always there and the plutonium will be produced. You can make the trade IF you substitute HEU for the natural uranium fuel. Also your numbers seem to assume a 100% capacity factor so I might lower them to about 2/3s of what you give. But if N. Korea were to use HEU fuel in the reactor then it could produce about 70 grams of tritium per year (assuming 25 MWt).

    • robgoldston (History)

      My numbers are explicitly in units of MW(th)-years, but you are right that you would have to reduce the ratio of U-238 to U-235 to make T, and if you want all T, no Pu, you would need to go to HEU.

  8. Gregory Jones (History)

    Also the TPBARs at Watts Bar are not losing 4.2% of the tritium they are producing to permeation. It is more like 0.04%.

    • HLC (History)

      Hi Gregory – thanks for pointing out the slip in my maths. You’re correct, the GAO report on US tritium production says that the TPBARs were losing 4.2 curies out of every 10,000 generated.

      Incidentally, the report’s discussion of tritium permeation into the water coolant got me wondering whether any tritium that permeated into the coolant in the graphite reactor might be detectable somehow.. For instance, might permeated tritium react with CO2 to generate tritiated water, that would ultimately leak out into the Kuryong river where it could be detected? Tritium is used as a tracer material after all..

      I imagine most of the tritium would get caught up in the graphite moderator, and any leakage into the river would probably be too small to be detected. But if anyone has the Chemistry to tell me more I would be really interested!

    • HLC (History)

      This type of detection-via-permeation doesn’t apply to the IRT reactor (I’m guessing) where any permeated tritium would be confined to the coolant pool…

  9. Tobias Piechowiak (History)

    How much more difficult is the Tritium chemistry compared to Transuranic chemistry? It is a gas and could diffuse through an assembly?
    Is it save to assume that since they can handle PU chemistry they also can tritium chemistry?

  10. Gregory Jones (History)

    Tritium permeation is only an issue at Watts Bar because power reactors must operate at high temperatures. This is why they must use lithium aluminate instead of a lithium aluminum alloy. At low temperature plutonium production reactors or research reactors such as North Korea’s 5 MWe or IRT, tritium permeation will not be an issue. Also all reactors produce some tritium due to ternary fission, so the detection of a little tritium in North Korea will not tell you anything.

    • HLC (History)

      I guess my thinking was more whether the detection of a significant/raised level of tritiated water would tell us anything. As I understand it, somewhere between 2 and 4 fissions in every 1000 are ternary, and only about 7% of these produce tritium.

      Without having really run any numbers on this idea, I don’t know whether the ‘purposeful’/ternary ratio of tritium would be high enough to generate a detectable spike, or even whether that ratio would survive the journey out to the point of detection. But its always interesting to think about how the world might try to get a better picture of what is going on in Yongbyon 🙂

  11. E. Parris (History)

    This has been a very impressive and enlightening conversation. Well done!

    • Anon2 (History)

      “This has been a very impressive and enlightening conversation. Well done!”

      Agreed. Wonky — a good read for wonk nerds like me.