When it comes to energy, nuclear fusion is often referred to as the holy grail as it would provide a source of abundant energy with no risk of meltdown, plentiful resources, and no greenhouse gas emissions. The main candidate fuel for future reactors is a mixture of deuterium (D) and tritium (T), which fuse to form energetic neutrons and helium. As neutrons can damage materials and limit their lifetime in the extreme environment of a fusion reactor, other potential fuels have been proposed that would strongly reduce or eliminate neutron formation. Among them, helium-3 (3He), one of the isotopes of helium, is often mentioned,1Mazzucato E. A First Generation Fusion Reactor Using the D-3He Cycle.Fus. Sci. Technol. 2021; 77: 173-179Crossref Scopus (2) Google Scholar motivated by the discovery that the moon contains significant amounts of this species, although it is virtually non-existent on Earth. As such, helium-3 is a reason often mentioned to go back to the moon and exploit its resources. Unfortunately, in many aspects, 3He fusion is much harder to achieve.2Stott P.E. The feasibility of using D– 3 He and D–D fusion fuels.Plasma Phys. Contr. Fusion. 2005; 47: 1305Crossref Scopus (29) Google Scholar We show that this technology, if demonstrated, is most likely to only emerge as a second or third generation of fusion power plants and given the current schedule for fusion demonstration and deployment, is unlikely to appear this century. Nuclear fusion requires bringing atoms sufficiently close together for the strong nuclear force to act and allow the heavier particle to form. However, since nuclei are positively charged there is a strong repulsive force, due to the Coulomb interaction, between the nuclei. Counteracting this repulsion requires very high temperatures of the order of hundreds of millions degrees kelvins. Several fusion reactions can in theory be considered. They need to be exothermic, involve low atomic number elements (as the electrostatic repulsion increases with the number of protons), involve two reactants, and conserve protons and neutrons. The list of the most interesting reactions is given below:D+T→He4+n+17.6MeV(1)D+D→H+T+4.03MeV(2a)50%D+D→n+He3+3.27MeV(2b)50%D+He3→H+He4+18.35MeV(3) Most reactions involve isotopes of hydrogen (H). One can distinguish two classes of reactions depending on whether they lead to neutron formation or not. The latter will be referred to as aneutronic fusion. Figure 1 shows the reaction rates for the five selected reactions as a function of the ion temperature. By far, the easiest reaction to achieve is (1) and involves D and T. Most (if not all) fusion research focuses on the D-T scheme. D is naturally present in water with a natural abundance of 1 over 6,500. Seawater provides a plentiful source of fuel. T is radioactive and has a half-life of 12.3 years (decaying in 3He). A very small quantity (0.2 kg/year) is produced through interaction between cosmic radiations and nitrogen-14. The natural T inventory on Earth is estimated at about 3.5 kg. Most of the tritium is produced in nuclear reactors using heavy water as a moderator, such as the CANDU reactors in Canada. The estimated world T inventory is about 30 kg while a 500 MW (thermal power) fusion reactor will burn about 28 kg of T per year (assuming operation at full capacity).3Kovari M. Coleman M. Cristescu I. Smith R. Tritium resources available for fusion reactors.Nucl. Fusion. 2018; 58: 026010Crossref Scopus (39) Google Scholar To solve the issue of the T scarcity, it is envisaged to produce it using lithium through the following reaction:4Rubel M. Fusion Neutrons : Tritium Breeding and Impact on Wall Materials and Components of Diagnostic Systems.J. Fusion Energy. 2019; 38: 315-329Crossref Scopus (21) Google ScholarLi6+n→He+T+4.78MeV The plasma will be surrounded by a tritium breeding blanket which will contain beryllium (for neutron multiplication) and lithium for T production. The produced T will then be extracted and used to fuel the reactor. A reactor will need to have a breeding ratio of 1.05,5Doerner R.P. Tynan G.R. Schmid K. Implications of PMI and wall material choice on fusion reactor tritium self-sufficiency.Nucleic Mater. Energy. 2019; 18: 56-61Crossref Scopus (18) Google Scholar i.e., produce 5% more T than it consumes, in order to be self-sufficient and provide T for future reactors. The concept of T breeding has not been demonstrated at scale, and ITER will test a number of concepts during its operations. It represents a significant technical complexity. The currently accessible resources of lithium could provide at least a few thousand years of fusion power, at today’s level of consumption,6Nicholas T.E.G. Davis T.P. Federici F. Leland J. Patel B.S. Vincent C. Ward S.H. Re-examining the role of nuclear fusion in a renewables-based energy mix.Energy Policy. 2021; 149: 112043Crossref Scopus (12) Google Scholar a number which also depends on the needs for other usages (batteries for example). Recovery from seawater would increase the known resources by several orders of magnitude but the environmental and energy costs of such extraction need to be studied in more details. Another challenge is the creation of a very energetic neutron (14.1 MeV) which will continuously bombard the material structures surrounding the plasma and can produce gaseous species (hydrogen and helium) in materials causing embrittlement and generation of transmutation products.7Knaster J. Moeslang A. Muroga T. Materials research for fusion.Nat. Phys. 2016; 12: 424-434Crossref Google Scholar The use of 3He as a fuel for fusion has long been considered as an alternative to the D-T scheme.8Post R.F. Nuclear fusion fuels.Proceedings of the BNES Fusion Reactor Conference. 1969; : 88Google Scholar 3He is produced on Earth from three processes: cosmic rays, beta decay of tritium, and lithium spallation. 3He leaks from the Earth mantle but at a rate of hardly a few kg per year. The natural abundance of 3He is about 300 atomic parts per million compared to 4He. The US currently possesses around 25 kg of 3He through its strategic reserve. The production through radioactive decay of tritium, both from military and civilian sources, amounts to an annual production of about 18 kg/year. To circumvent the resource issue, the idea has been repeatedly proposed to make use of the lunar resources. Having no magnetosphere, the moon surface is bombarded by the solar winds, which contain 3He created through fusion reactions on the sun. The 3He content in the Moon surface has been characterized by analyses of lunar samples brought by the Apollo and Luna missions. The 3He abundance is estimated at about 30 μg per g of regolith.9Wittenberg L.J. Santarius J.F. Kulcinski G.L. Lunar Source of 3He for Commercial Fusion Power.Fusion Technol. 1986; 10: 167-178Crossref Google Scholar Taking into account a depth of 3 m (3He is implanted from energetic particles and therefore concentrated near the surface), the lunar reserves would be of the order of 1 million tons, corresponding to 600 ZJ (1 Zettajoule = 1021 joules) i.e., about 1,000 times the annual world total primary energy consumption. Mining 3He on the moon is therefore often mentioned as a justification for new lunar missions and space resource exploitation, although the economic viability of lunar mining is not trivial but is outside the scope of this paper. It is important to stress that the D and lithium resources on Earth are so abundant that the case to mine on the moon will be difficult to make from an economic point of view. A few points need to be considered regarding the technical feasibility of D-3He fusion. Although the reaction itself does not produce neutrons, some D-D reactions can happen (reactions 2a and 2b), forming T and neutrons in the process. Since the D-T reaction has a much higher reactivity (Figure 1), D-T reactions will compete with the D-3He reaction and will result in significant neutron production (albeit strongly reduced compered to pure D-T). In practice, the D-3He scheme is therefore not aneutronic.2Stott P.E. The feasibility of using D– 3 He and D–D fusion fuels.Plasma Phys. Contr. Fusion. 2005; 47: 1305Crossref Scopus (29) Google Scholar,10Rider T.H. Fundamental limitations on plasma fusion systems not in thermodynamic equilibrium.Phys. Plasmas. 1997; 4: 1039Crossref Scopus (24) Google Scholar The use of 3He rich mixture (i.e., with a decreased D content) can still strongly reduce neutron production and increase the material lifetime. Besides the temperature requirements, the technical feasibility of this scheme will depend on the underlying plasma physics, which dictate the possible confinement schemes. In a magnetically confined plasma, an important parameter is β, the ratio of the plasma pressure p to the magnetic pressure pB, with B the magnetic field (kB is the Boltzmann constant, Ti the temperature):β=ppB=nkBTiB2/2μ0 In a tokamak, β is typically of the order of a few percent, while spherical tokamaks (with lower aspect ratios) can achieve values of about 40%. There exists a limit on the plasma pressure, which can be confined. Other confinement devices such as the field reversed configuration (FRC) could attain much higher values. For a given magnetic field and for temperatures maximizing the reaction rates (13 keV for D-T and 60 keV for D-3He), the product βτE would need to be 24 times higher for a D-3He-fueled reactor than for a D-T-fueled reactor for both to reach ignition. The ratio increases to 61 times for a 15% D-85% D-3He mixture. Another interesting parameter for a reactor is the volume averaged power density (Pf), which scales as the square of the plasma pressure p2 (or β2B4) and influences the possible cost of electricity. To get the same electricity producing power density, a 50% D-50% 3He fuel mixture would require a plasma pressure 9 times higher than a D-T reactor, while this ratio increases to 14 times for a 15% D-85% D-3He mixture. At the temperatures required for D-3He fusion, bremsstrahlung and synchrotron radiation will become important and will radiate a significant fraction of the plasma power.2Stott P.E. The feasibility of using D– 3 He and D–D fusion fuels.Plasma Phys. Contr. Fusion. 2005; 47: 1305Crossref Scopus (29) Google Scholar The plasma needs to be hot enough to not be limited by bremsstrahlung but not too hot that synchrotron radiation becomes too high. Such effects are not important in D-T fusion. Another advantage often mentioned for aneutronic fusion is the possibility to use direct energy conversion to convert the plasma energy into electricity, promising efficiencies up to 60%–70%. A D-T reactor would work with a classical steam cycle where neutrons heat a fluid that is used to generate steam and produce electricity through turbines, as in thermal power plants (coal, nuclear). This limits the conversion efficiency at around 30%–40% if steam is used, but advanced reactor designs with hot gases or salts have been proposed to raise the efficiency up to 60%. As a matter of comparison, typical nuclear power plants have conversion efficiencies in the range of 35%, whereas advanced gas power plants with combined cycle have efficiencies up to 50%–60%. While the first generation of fusion power plants will most likely not exhibit optimized efficiency and capacity factor,11Mulder R.A. Melese Y. Lopes Cardozo N.J. Plant efficiency: a sensitivity analysis of the capacity factor for fusion power plants with high recirculated power.Nucl. Fusion. 2021; 61: 046032Crossref Scopus (1) Google Scholar it is reasonable to expect that fusion reactors will tend to higher efficiencies to become economically more attractive. Direct energy conversion has not been demonstrated at scale yet and encompasses very significant technical challenges.10Rider T.H. Fundamental limitations on plasma fusion systems not in thermodynamic equilibrium.Phys. Plasmas. 1997; 4: 1039Crossref Scopus (24) Google Scholar More fundamentally, for D-3He fusion, bremsstrahlung and synchrotron radiation will become important and will radiate significant fraction of the plasma power,2Stott P.E. The feasibility of using D– 3 He and D–D fusion fuels.Plasma Phys. Contr. Fusion. 2005; 47: 1305Crossref Scopus (29) Google Scholar will increase the level of power deposition on the main chamber wall. This heat will be collected through a cooling system and reduce the amount of charged particle power available for direct energy conversion and thus the overall efficiency. Given all those difficulties, and since thermal energy conversion is a well-known technique, it is difficult at this point to see whether it will be able to deliver on its promises. An important note is that aneutronic fusion makes the power exhaust more difficult since all the energy is carried by the plasma and follows the magnetic field, leaning toward open magnetic configurations whose performances are still far from those of tokamaks. Mastering fusion for energy production is extremely difficult. Looking at the triple product (a measure of the plasma performance), the progress between the mid-1960s and the mid-1990s has been rather impressive with a doubling time every 18 months on average over that period—compared with Moore’s law and the doubling of the number of transistors on a chip every 24 months. The rate of progress has stalled since the mid-1990s mainly because of the inertia of ITER. The current planning for ITER foresees a first plasma around 2025 and the start of fusion operations a decade later. The European roadmap for fusion aims at a demonstration reactor (DEMO) producing electricity in the 2050s.12Donné A.J.H. The European roadmap towards fusion electricity.Philos. Trans. A Math. Phys. Eng. Sci. 2019; 377: 20170432PubMed Google Scholar If one follows that development plan and uses the same deployment rates for fusion as for other energy technologies13Kramer G.J. Haigh M. No quick switch to low-carbon energy.Nature. 2009; 462: 568-569Crossref PubMed Scopus (167) Google Scholar (Figure 2), fusion could produce a few percent of the world energy demand (WED) toward the end of the century. Several initiatives, some of them private, are trying to accelerate this timeline with a high-risk/high-payoff approach taking advantage of recent technology developments and have a much more aggressive timeline. This is an encouraging sign that fusion development is now taken seriously and that there is a will to try and make fusion happen as soon as possible. If one of these initiatives could demonstrate fusion in the period 2025 to 2030, fusion could represent 1% of the world energy demand around 2060 and could play a more significant role in the second half of the century. D-T has focused most of the attention and to date, very little research is done on D-3He. Active research is ongoing on materials development and T breeding as those challenges need to be solved for fusion to reach commercial exploitation. The much more demanding plasma performances and absence of viable supply chain for 3He make it a much more hypothetical technology, unlikely to see increased interest in the near future. The situation might change once a demonstration of fusion energy is made, be it by ITER or one of the private ventures, as fusion would then enter into the realm of feasible energy technologies and should see increased levels of funding. For all those reasons, D-3He fusion would only become a potential technology for a second or third generation of fusion reactors, the development of which would only really start when D-T fusion reactors get massively deployed. Several conditions need to be met if D-3He is to become a reality:•D-T fusion proves successful and is massively adopted•The material lifetime issue and T breeding requirements are too demanding•Suitable technologies exist for massive recovery of 3He from the moon•The price of resource import from the moon becomes competitive with those of terrestrial resources•Too-rapid depletion of vital resources for D-T Those conditions are unlikely to be met before at least the end of the 21st century. It is of course not possible to predict how things will develop in later centuries to come, but it is clear that D-T fusion, if mastered, would provide a very attractive source of energy with abundant resources. Would humanity benefit from the use of another scheme in the very long term? Perhaps, but at present the case is certainly not very strong in front of the technical challenges it encompasses.