The supply of tritium fuel is perhaps the biggest problem for mainstream fusion reactors, perhaps even a show-stopper. Surprisingly, fusion advocates say that the fuel is abundant and an advantage of fusion, e. g. the ITER website says

Fusion fuels are widely available and nearly inexhaustible

In reality the opposite is the case for tritium.

A single fusion reactor would consume about 50kg/year of tritium but, currently, only about 0.5kg/year is produced worldwide. In all the oceans of the world there is about 30kg of tritium which is produced naturally but decays with a half-life of 12 years (5% per year).

Generation and Recuperation of Tritium

Due to the impossibility of external supply, designs for fusion reactors involve breeder systems to generate tritium from the neutrons emitted in the D-T fusion reactions, in components called Breeder Blanket . Further, because only a few percent of the injected tritium is burnt in the reactor, the un-burnt fuel must be recuperated, purified and re-injected.

This subject of tritium self-sufficiency in a fusion reactor is studied in detail in:  Abdou et al. Nuclear Fusion 61 (2021). The fuel cycle of a fusion reactor includes two sub-cycles: the inner fuel cycle (IFC), i.e. plasma exhaust (vacuum pump), fuel clean-up, isotope separation, water detritiation, storage and management, and fueling systems, and the outer fuel cycle (OFC), i.e. first wall, divertor, breeding zone, coolant processing and tritium extraction system. A schematic of a typical fusion fuel cycle and tritium flow is shown in the block diagram below.

The tritium which enters the reactor exits in three different ways:

  1. Part is burnt 
  2. Part is re-cycled for rapid reuse
  3. Part is absorbed by the inner walls of the reactor and must be re-cycled later

The list below gives the complete set of parameters considered in analysis of the paper.

Designs for breeder blankets contain a mixture of elements with different functionality. One is lithium which when hit by a neutron makes tritium and another is beryllium, to multiply the neutrons so more than 1 tritium atom is produced per incident neutron. For existing breeder designs, this number, the Tritium Breeding Ratio (TBR) is between 1.05 and 1.15  .

The figure below compares the required TBR with the possible available TBR is the green bands. Parameter values which are needed for tritium self-sufficiency are not possible with the ITER "state-of-the-art".

The conclusion of the Abdou et al paper, is that the physics and technology state-of-the-art will not enable DEMO and future power plants to satisfy the principal requirements for tritium supply.

Start-up Tritium Supply

A further massive problem with the tritium supply is the initial start-up inventory, equally addressed in Abdou et al. The current world stock of tritium is about 18kg which will increase until the CANDU fission reactors reach the end of their lives. From then on, the radioactive tritium will decay away at a rate of 5%/year. When the ITER project starts operation it will take a major fraction of the world stock. The figure below shows an estimated evolution of the world stock of tritium.

According to this estimate, around 2050 when the construction of a DEMO fusion reactor might start, world tritium supplies could be less than 5kg. The Abdou et al paper gives estimates of required start-up inventories for a reactor.

It seems difficult for multiple fusion reactors to start operations without sufficient start-up supplies of tritium and this might require specific new fission reactors to be built to provide part of the tritium supply.


This is such an essential issue that several more references are given.

An excellent treatment of the supply issue of tritium, as well as lithium and other rare minerals for use in a fusion reactor, is given in the book of L.J. Reinders: Economic and Sustainability Aspects of Nuclear Fusion .

Presentations of the tritium supply problem can be found in a series of article on the website of Steven Krivit at:

An article by Daniel Clery in Science :  

Opinion of EUROfusion managers Federici and Donné:

"Achieving tritium self-sufficiency will be an unescapable requirement for any next-step fusion nuclear facility beyond ITER. However, no fusion blanket has ever been built or tested. Hence, its crucial integrated functions and reliability in DEMO and future power plants are by no means assured. However, the program in Europe benefits from many years of design and R&D, primarily carried out in European Fusion Laboratories. In addition, ITER presents a first and unique opportunity to test the response of representative component mock-ups, specifically called Test Blanket Modules (TBMs) at relevant operating conditions, in an actual fusion environment, albeit at very low neutron fluences.

As an example, a 2GW fusion power DEMO is expected to consume around 110 kg of tritium per full power year. The large majority of this tritium must be produced by the reactor itself, and this clearly underscores the indispensable requirement for the breeding blanket to produce and enable extraction of the bred tritium to achieve tritium self-sufficiency (i.e., it must produce its own fuel). However, there is the need to start-up the reactor at the very beginning of operation with a sufficient amount of tritium provided by external sources (5–10 kg). This raises a need to better understand and monitor the future availability of tritium and understand the impact of limited resources on the timeline of DEMO. However, there is essentially very little that the fusion community can do to exert an effect on the supply side, as tritium is a by-product of the operation of some specific fission reactors and not the primary economic incentive. Defence stockpiles of tritium are unlikely ever to be shared, and commercial CANDU operators will not alter their plans just to sell more tritium for the start-up of the first fusion power plants. In the short-term it is recommended to monitor the production of tritium in Heavy Water Reactors and estimate the commercially available supply. If, at some point in the future, it looks as though the demand for DEMO, alongside the other tritium consuming devices foreseen globally, will exceed the supply from CANDUs, then action would have to be taken. It is likely that production of significant amounts of tritium from a dedicated source would be very expensive and take a long time. The “tritium window” as it was once defined by Paul Rutherford is not open indefinitely. Based on current estimates, we believe it would be open until around 2050–60, after which it closes quite rapidly, unless the future of the CANDU reactor program turns out much more favourably than could presently be expected. Any program strategy that substantially delays DEMO places fusion at risk, by allowing the unique and effectively irreplaceable tritium resource to decay to levels which may be insufficient to complete fusion's technological development."

(G. Federici, Fusion Program Manager. and A. J. H Donné, Program manager in Concluding Remarks, DEMO Pre-conceptual Design Phase ).