Among the many fields of physics that I enjoyed touching in my scientific life, some with more depth, some others just for a little bit of fun, there is also fusion research, albeit from some distance. I was still with ENEA Materials Science Division in Rome around 2004, when some people from the Nuclear Fusion Laboratory in Frascati contacted me. At that time there was still an ongoing discussion about using SiC as a protection of ITER’s first wall and in some parts of the divertor. As it turns out, silicon carbide is a very complicate material to fabricate and maintain, and one of the things I was studying in those years was the microscopic fracture mechanics of polycrystalline SiC. In the Frascati team they had one person dedicated to this problem (which was later abandoned, I hope not because of me…) and he asked me to do some calculations of neutron irradiation cascades in SiC grain boundaries, to estimate fragilization from damage accumulation (grain boundaries are notorious defect getters in any material). Although quite remote from the main machine design, with all its problems of magnetic confinement, plasma stability and the like, it was also the occasion for me to get to know a little bit better that mysterious world.
It has been recently announced that the Joint European Torus (JET) nuclear-fusion experiment based in Oxfordshire, UK, has more than doubled the amount of fusion energy produced in a single “shot” – smashing a previous record that JET itself has held since 1997. Officials announced on Feb.9 this year that, during an experiment in late 2021, JET achieved 59 MJ of fusion energy over five seconds, beating the previous record of 22 MJ. The 11 MW of thermal power produced (with 40 MW of input electric power) is lower than that achieved in 1997, resulting in a ratio of fusion power to heating power of Q~0.3. A little more far south, the International Thermonuclear Experimental Reactor (ITER) still under construction in Cadarache, France, like its predecessor JET is still conceived as a test, not a true power reactor. With its 70m height and weighing 23,000 tons, ITER is often labeled the world’s largest science experiment, and the next step in the journey to fusion energy. Both JET and ITER are fusion reactors of the “tokamak” type, which use magnetic confinement to hold a 100-million K hot plasma inside a huge doughnut-shaped chamber.
The physics questions that these machines explore are super interesting, also from a fundamental point of view. The progress in plasma physics over the past 30 years has been impressive, and plentifully funded. However, that fusion physics works is no mystery: just look up in the sky, and you can see a beautiful yellow round reactor that has been working since 5 billion years already. The engineering side however, is wholly another story. With today’s writing I would like to spell out some points that represent serious limits for the scaling up from laboratory experiments to power generation from fusion. I am not a denialist, rather I am among those who believe that nuclear energy is the only viable solution, to both our energy and climate problems. Many of the technological limits will be overcome with further research and development, despite it’s already been 80 years that this effort is ongoing, making it also the world’s longest science experiment. However, I hate reading unjustified, hyped-up claims and bombastic announcements in the media, whose only result is to diminish the social credibility of science in case some such claims, as it often happens in science, may not be verified. Just one example, from the NYTimes, March 22, 1976: “Unexpectedly rapid progress in recent months in fusion research has virtually guaranteed breakeven power production for the reactor in Princeton… blah blah blah”.
It is often popularly explained that nuclear fusion works like in the Sun, and that fusion reactors such as JET, ITER and their future developments will harness a little star for us to live with endless and clean energy. One corollary to this very political (and only vaguely scientific) statement is that, since fusion burns hydrogen that can be found in seawater, this will be a practically infinite and cheap source of energy, available to everybody on the planet.
First, the nuclear fuel is not seawater but a mixture of deuterium and tritium. While deuterium is indeed relatively abundant in seawater, from which it can be extracted at about 500 dollars per kg, the tritium radioisotope is still produced in small quantities at a huge cost (about 30,000 dollars per gram) and is used mostly for hydrogen bombs. A future reactor is projected to require about half a kilo of tritium per day of operation. Any development of fusion reactors would require large-scale production of tritium with industrial methods that have yet to be invented.
Second, the Sun exploits the proton-proton reaction, which has a high energy threshold and very low cross section (that is, probability). Thanks to the enormous mass of hydrogen available, and the confinement provided by the enormous gravity of the Sun, the probability of a p-p reaction by tunnelling across the Coulomb barrier is still very high, so that our Sun burns about 600 million tons of hydrogen per second (ahem…) at the relatively low temperature of “just” 15 million degrees. In a fusion reactor that uses tiny amounts of fuel compared to a star, something must replace the gravity to compress the fuel. Be it strong magnetic fields or powerful laser beams, anyway, the much cleaner p-p reaction cannot be used (the first step is an endothermic weak transformation p+p→D+positron+neutrino, with a cross section so small that is practically impossible to measure). Hence the deuterium-tritium reaction, D+T→He+neutron, is preferred, since it starts having a decent cross section (=probability) at around 100 million degrees. The deuterium-deuterium reaction would be even more preferable (less neutrons, and at lower energy are produced) but it requires to get to at least 300 million degrees.
Third, D-T fusion also emits high-energy (14 MeV) neutrons that represent a problem, by heavily damaging and rendering radioactive the steels, tungsten and any other reactor materials, which need to be periodically replaced. Neutrons take up more than 80% of the energy produced in the fusion reaction: an energy source consisting of 80 percent neutron streams may be the perfect neutron source, but it’s at least weird to dub it as the ideal electrical energy source. Of course, the reactor will produce an awful lot of thermal energy, but this is mostly carried by energetic neutrons, and that is not something to be taken lightly. Besides, it is good luck that the Sun uses hydrogen and neither deuterium or tritium, otherwise every square inch of the Earth would be submerged by a tremendous flux of deadly neutrons.
In the future, a “breeder blanket” would be inserted between the plasma and the walls, with multiple objectives: to protect the outer walls, to host the heat exchangers, and to use the neutrons to regenerate tritium from nuclear reactions. This should be a circulating fluid containing lithium, to exploit the reaction n+Li→He+T, which regenerates the consumed tritium. Lithium is a relatively cheap material, at about 20 USD per kilogram; however, note that this price has exploded from less than 4 USD/kg in the last few years, because of the increasing demand for batteries, and it will likely keep going up. Anyway, the solutions considered are liquid PbLi blankets, or molten lithium salts like LiF-BeF2. We should remember that liquid metal coolant was the Achille’s heel of fission breeder reactors like Superphenix, and molten salts are the very same technology that is proposed in GEN-IV fission reactors, yet with much less severe technological requirements.
The complex reprocessing to extract “on the fly” tritium from the fluid would further reduce the energy output. Moreover, only about 10% of the precious injected tritium is burned, and the rest escapes quickly from the high-density region; this 90% must be recovered from the walls and reinjected several times, and experience with Princeton’s TFTR and the European JET shows that about 10% is lost anyway. This means that the breeder-blanket must be able to replenish a much larger supply of tritium than is consumed, with even steeper design constraints. Far from being a mere technicality, tritium breeding is probably the main reason why many still believe that the only viable solution in future commercial reactors will remain the D-D reaction, given the difficulty and cost of managing the tritium stock on very large scales.
Compared to the handful of semi-skilled personnel required to operate hydroelectric plants, natural-gas burning plants, wind turbines, solar power, and other conventional power sources, a fusion reactor will require at least double the personnel of a fission reactor, that is about 1,000 engineers, technicians and manipulators, with a dedicated training and experience. Similar to fission reactors, this will be an obvious limiting factor to export such energy sources to third-world country, at least in the beginning. Moreover, like in fission, the huge demand of water will make the plant unfeasible for countries that have geographical problems with water supply.
And once such powerful neutron sources (let’s not call them just “energy sources”, to be honest) would be worldwide available, another risk looms behind the door. The open, or clandestine production of plutonium 239 will be possible in a fusion reactor simply by placing natural or depleted uranium oxide at any location where neutrons of any energy are flying about. The ocean of rushing neutrons (more than 1000 times the fluence of a typical fission reactor) permeates every corner of the reactor interior. Slowed-down neutrons are readily soaked up by uranium 238, whose cross section for neutron absorption increases with decreasing neutron energy. Note that this risk is much higher for a fusion reactor than for its fission-based counterpart: the high-burnup rate of the fission fuels makes their plutonium a mixture of 239 and 240 isotopes, thus rendering the material unusable for making bombs. Weapon-grade plutonium must be 95% in the isotope 239, and this could be easily achieved with the short-time pulsed irradiations behind the D-T fusion plasma.
The biggest question of all, eventually, is when a fusion reactor will be a true source of energy? Until the output rate isn’t a lot higher than the input rate, it may be scientifically interesting, but practically useless. A fusion reactor will suffer parasitic power drain on a scale unknown to any other source of electrical power. About ~100 MWe are consumed continuously, partly also when the reactor is off, by liquid-helium refrigerators, water pumping, vacuum pumping, tritium processing. Other sources of parasitic drain are the power needed to control the fusion plasma and to pump the blanket coolant, typically estimated at 8-10% percent of output fusion (thermal) power. Such power drains determine lower bounds to reactor size. If the thermal power is 300 MW, the entire electric output of 120 MWe would barely cover on-site needs; at ~850 MW about half of the electrical power is “wasted”. In a nutshell, below a size of at least 1 GWe (that is, at least 2,5 GW fusion thermal power) parasitic power drain makes it uneconomic to run a fusion power plant.
ITER organization claims its 10-times power gain target, the already cited Q ratio. However, this is where the meaning of words has to be taken with care. The power ratio Q does not compare the rate of power-out vs. power-in. It only compares the ratio of the power that is used to heat the fuel versus the thermal fusion power that is produced by the fuel. This figure only speaks about what happens deep inside the reactor when fusion occurs, not the total amount of energy it takes to run the whole operation, or the actual usable electricity the fusion reaction could produce. I am pretty sure this is a standing confusion in the minds of journalists, people readers, and politicians who vote energy bills alike. If we look at the power-out/power-in ratio for the last reported, successful JET experiment, this value is rather close to 1%. Moreover, the power that is obtained is thermal power, that has to be passed through the heat exchangers to be converted into electrical power, with a thermodynamic yield of about 40%. From the point of view of operation, if the new ITER were to be hypothetically attached to the power grid, it would be a zero-power machine: taking in 50 MW to ignite the fuel requires about 200 MW of electricity, and when getting out 10 times the power as heat, these 500 MW thermal would turn into about 200 MWe: it would effectively be not more than a massive, expensive, complicated, scientifically remarkable extension of a simple electrical wire with zero gain. However, the Q-factor would be 10, as announced. Selling such 10-fold power claims to the general public can only fuel skepticism, summarized in the popular joke “we are constantly 30 years away from fusion energy”. I think we scientists should be more intellectually honest than the average social media, at least.
It must be admitted that, with all our hopes and earnest efforts, after 80 years we are still trying to demonstrate the scientific feasibility of the thing. Getting energy out if it, it’s a bit more far down the road.