 Developing a working, controlled and economically viable fusion reactor is a major challenge. We have seen how even a moderately sized such reactor must produce the energy equivalent to 100kg of TNT every second. Once the plasma physics issues are solved, let's speculatively look at the practicalities of building a useful fusion reactor, mindful of the fact that one has not yet been realized. A controlled fusion reaction requires the fuel mix, such as Geterium Tritium, to be extremely pure. As a result, the reaction chamber must be pumped down to a very good vacuum, 10 million times less than a standard atmosphere or lower. Getting chamber pressures down to this level requires special cryo pumps. Cooled internally by liquid nitrogen, they operate at such low temperatures that gases condense into a liquid on the moving parts and are thereafter ejected from the chamber. However, going from the tenuous gas inside the reactor directly to the atmosphere would create too large a pressure differential, even for a cryo pump. A second backing pump, often with a more simple design such as a turbine, is used to aid the cryo pump. In reality, large fusion experiments are serviced by a labyrinth of pipes, pumps, valves and sensors large enough to fill an entire building, and so will reactors in the future. Even so, pumping out all the gas takes a long time. After confinement experiments where impurities are particularly detrimental, first perform something called glow discharge cleaning before main operation begins. This involves passing a current through gas to form a relatively cold plasma, which can then get the last stubborn particles which remain adsorbed onto the walls. Large scale experiments such as Eter and W7X have or will first run a pure helium plasma to condition the vacuum vessel and to get everything ready before any thermonuclear reactions take place. Once everything is up and running, fusion reactions produce stable helium-4 nuclei which do not react any further. Helium-4 is sometimes called ash, and the fuel is said to have burned up. The rate of thermonuclear fusion in a given volume of plasma is given by the reactivity multiplied by the densities of each of the reactants. As the fuel reacts, there is proportionately less and less of it in the plasma, so the reaction rate drops. In other words, the more of the fuel that has already burned up, the slower it will continue to burn. In the case of any sort of pulsed scheme, such as laser or X-ray based inertial confinement or magnetized target fusion, the reaction lasts for only a brief moment. This short time is enough to use up only a certain proportion of the total fuel, referred to as the burn fraction, which is realistically no more than about 30-50%. This is problematic for achieving energy gain because the remaining 50-70% of the fuel must still be heated and confined. After the pulse is over, the rest of the fuel and ash fills the chamber and must somehow be emptied out, or at least made to not interfere with the subsequent pulse. If the ash does get mixed in, it will dilute the fuel, lower the reaction rate and cool the plasma by emitting Bramstrahlung radiation. The danger of this happening depends on the confinement scheme. In the case of magnetic confinement, the aim is to have a long term steady configuration. Plasma in a Tochimac or Stellarator will flow out to a limiter or diverter at the edge and recombine back into a gas. At this point it is pumped out. The density of plasma must be maintained by puffing in fresh gas, firing in fuel pellets or particles via neutral beam injection. When for example a Tochimac is started up, it will be full of pure fuel and this is what will be coming out of the pumps. As fusion reactions take place, the concentration of helium ash will rise, and so too the amount of it being pumped out. Eventually an equilibrium will be reached where the rate of helium production in the core will equal the rate at which it is pumped out at the edge. In both cases, the gas that has been pumped out will be a mixture of fuel, ash and other impurities. Both deuterium and tritium are chemically identical to normal hydrogen and therefore both explosive and leak-prone in their pure forms. Typically they are forced to react with oxygen to make water and then the two isotopes can be separated and stored as necessary. Special care must be taken, since if radioactive tritium were to leak or be stolen it would be disastrous. It would mix with watering the environment and be impossible to isolate. So far, three magnetic and one inertial fusion experiments have used tritium. At Jet and Eta, the water detritiation plants are sizable, even compared to the nearby enormous torus holes. This indicates that while these problems are not insurmountable, they will be expensive to solve both in terms of energy, making break-even just a little bit harder, and simply in terms of money. To maintain the requisite vacuum, there must be a wall made of solid material around the reaction chamber. This appropriately named first wall must maintain a firm barrier between gases inside the chamber and various contaminants outside, while also resisting damage from the fusion plasma, neutrons, and any other radiation which is given off. In the case of pulsed schemes such as inertial fusion, the first wall will experience cyclic loads which are especially damaging. Sputtering is the process by which ions from a plasma erode atoms from a target material. The surface remains overall solid, but as the hot plasma ions come in, atoms of the material are plinked off one by one. Sputtering is a problem for any component with line of sight to the plasma, because even strong magnetic fields are not perfect at confining the plasma. Under the influence of incoming plasma, materials such as tungsten turn to fuzz, this is a technical term as shown here in images from an electron microscope. Ions originally from a plasma can penetrate some ways into a metal and recombine back into a gas. This gas will coalesce some distance below the surface and create blisters to further damage the first wall. One area which is particularly vulnerable is the diverter or limiter on any magnetic confined machine. For a tokamak with a conventional diverter, a lot of energy will make its way to a pair of thin, circular strike points. Tungsten is a better choice than most for a diverter material because it's tough and has a high melting point. Even such a hard metal will eventually be churned into fuzz by the plasma as we have seen. One proposal is to shape the magnetic field so that it splits into multiple directions, referred to as the snowflake configuration due to the plasma's intense political opinions about public health care, or perhaps just due to its shape. The energy coming out of the plasma will then be spread over multiple strike points and some of it can also be radiated away. Most of the energy will come out of the plasma as heat in one way or another. In thermodynamic terms, heat is the useless random motion of particles, while work is the sort of useful energy which moves vehicles forward, generates electricity and so on. Heat can only be converted into work when going from a hot to a cold medium. The conversion process will have some efficiency. A fraction of the thermal energy gets converted into work and the rest is wasted, raising the temperature of the cold medium. Theoretically, the most efficient possible device capable of converting heat to work is called a Carnot engine. Its efficiency is given by one minus the cold temperature divided by the hot temperature, where both of these have to be relative to absolute zero, so using something like units of Kelvin. For greater efficiency, make one end as hot as possible and the other end as cold as possible. On Earth, the temperature is limited to that of cooling water at about 300 Kelvin, while the hot temperature is limited to about 1000 Kelvin, above which the pipes and turbines would melt, giving a theoretical upper limit of 70%. From the point of view of the Carnot efficiency, modern steam engines are pretty close to what is theoretically possible and are therefore widely used. A combined gas turbine might get up to 60% efficiency, while fission plants realistically achieve 35%. Unsurprisingly, current plans to realise fusion take the larger of these two values. In a fusion power plant, somewhere close behind the first wall, there will be a blanket, containing various pipes to take away the heat. The blanket should not be cooled directly by the same water which will drive the turbines, because this could allow radioactive material like tritium to leak out. There should be an intermediate coolant going from the blanket to a heat exchanger and only then into the turbines. Two good choices for this coolant are water and helium. Both are easy to pump, are chemically inert and resistant to radiation. Blanket designs are still being finalised. Recently, some have been proposing direct energy conversion from the plasma. In principle, a hot plasma exerts a high pressure. If it is magnetised, as the plasma expands it will push magnetic field lines apart. When moving magnetic field lines cut through a conducting coil, they induce a voltage, as in any electrical generator. The expanding plasma can therefore generate electricity directly. From a thermodynamics point of view, the efficiency can be nearly 100% since the starting temperature is so large. However, this approach has not been practically demonstrated for a thermonuclear plasma. It would probably require a fundamentally pulsed or transient approach as opposed to a stellarate or a tokamak, which is intended to run continuously for hours, because having expanding field lines will almost certainly destabilise the plasma eventually. Importantly, the direct conversion method will also lose 100% of the energy carried by neutrons, bramstrahlung or any other electromagnetic radiation, all of which are not charged and therefore will not interact with the magnetic field. For deuterium tritium, the neutrons alone are two thirds of the energy. Speaking of neutrons, as we saw in episode 1, the fusion reaction with by far the largest cross-section is the deuterium tritium one which gives off energetic neutrons. Beyond that, the next few fuels, ranked by reactivity, all contain deuterium, which reacts with itself to also give off neutrons and it produces tritium which then gets you back to the deuterium tritium reaction. The only truly anutronic reaction is proton boron, but it requires enormous temperatures and even then has a low reactivity, factors preventing it from being practical anytime soon. Therefore, any fusion reactor in the near future will generate a lot of neutrons with energies of 14.1 and 2.45 million electron volts. Because the energy that they carry is quite large, these are referred to as fast neutrons in the context of proven fission reactors. By themselves, neutrons are deadly to living beings and destructive to inanimate objects, particularly the fast ones. Consider that deuterium tritium fusion makes about 22.4 million electron volts of energy involving just over one neutron. This is if we include the tritium breeding component I will get to in a moment. Meanwhile, common uranium 235 fission produces about 2.5 neutrons for an average energy release of 180 million electron volts. This means that for any unit of energy released, joules, kilowatt hours and so on, fusion energy would produce three times more neutrons than a conventional fission reactor, all of them fast. Deuterono will give some cases where this high flux of neutrons may actually be desirable. In general though, the neutrons pose a major challenge which must be overcome before fusion power becomes a reality. Neutrons do not really interact with the electron cloud which takes up most of the space of an atom, but only with the nucleus. Just like with fusion reactions we saw previously, a cross section depending upon the energy of the neutron determines how likely it is to undergo a reaction with a given nucleus. Such a reaction is quantum mechanical and therefore a probabilistic process. A neutron has some likelihood to interact with a nucleus, but it is not guaranteed. Therefore, you cannot say that so many centimeters of material will categorically stop all the neutrons, just that statistically it will reduce the number by a certain amount. For example, in my favorite paper from Princeton, they have calculated that shielding in a realistic power plant would reduce the neutrons by a factor of 10 for about every 13.5 cm, meaning a factor of 100 for 27 cm and so on. If the absorbing material had a higher neutron absorption cross section, this thickness would be smaller and vice versa. There are two important results of neutron interactions. Firstly, the neutron may be absorbed into the target nucleus, which fundamentally changes it, possibly into another element. Secondly, the neutron may transfer an enormous amount of energy to the nucleus, causing it to be displaced. In the latter case, the neutron may be absorbed, but may also continue moving independently and strike other nuclei later. When a neutron is absorbed by a nucleus, the mass number changes by one, leading to several potential consequences. For some nuclei, there is not much change. The most common and stable isotope of carbon is carbon 12, 6 protons, 6 neutrons. By absorbing another neutron, it becomes carbon 13, which is also stable and the two isotopes are chemically indistinguishable. No problem. There are quite a number of similar elements where the most common isotope can quite happily absorb a fusion neutron and be basically unchanged. And a good thing too, because these are precisely the elements from which a fusion reactor should be built. In other cases, when a neutron is absorbed, it makes the daughter nucleus unstable and therefore radioactive. This process is called neutron activation and is clearly undesirable. One neutron further up, the carbon 14 nucleus is unstable, decaying by beta emission. Radiocarbon dating works by looking at the ratio of radiation from an old sample of carbon as compared to a young one. For example, the carbon 14 and the wooden beams of a sunken Byzantine dromon would have had over a thousand years to decay relative to identical wood produced today and is therefore less radioactive ground for ground. That's all well and good for archaeology, but creating significant amounts of carbon 14 or any other radioactive isotope through neutron activation is bad for fusion power. Any activated material must be carefully handled during the lifetime of the reactor. It must thereafter be disposed of safely and any noticeable radioactivity is almost certain to be difficult politically. Suppose that you somehow built a fusion plant out of pure carbon. Even also that the amount of neutrons produced during the operating lifetime was enough that on average one in every 100 carbon nuclei absorbed a neutron. If the neutron absorption cross sections of both carbon isotopes were the same, this would mean that statistically about one in every 10,000 nuclei would absorb two neutrons and become radioactive. Taking the ratio of cross sections it would probably be about one in every 30,000 nuclei. This is a far, far higher proportion than occurs naturally. One would then need to consider whether this is a dangerously large amount and what the cost would be to dispose of it. Carbon is actually a pretty good element in terms of activation because it usually takes two neutrons to do so. Realistically, a reactor will be built using many more elements than just pure carbon though. Iron, the other major component of steel is also fairly safe. What's more concerning in terms of activation are nickel, cobalt, and copper, all of which are required for strong heat resistant alloys to be used for the first wall. A fusion power plant would probably also require some other components to function like coils, optics, sensors, and so on, which would be made out of a whole range of chemical elements. All of the individual isotopes would have different neutron absorption cross sections which also depend on the energy of the incoming neutrons. To predict the amount of activation, calculations would have to be performed while keeping track of all that stuff and then accounting for the half-lives of the daughter nuclei. This is being done by very involved computer models which try to account for everything I've just mentioned. The overall target for a fusion plant is to have everything decay to below the levels of background radiation within 100 years of shutting down. Simulations show that this is more or less achievable if the plant is designed correctly. In practice, however, to fully determine the exact amount of activation a given material or sets of materials will undergo, you have to perform realistic experiments. It's sort of like when Sheldon Cooper tried to learn to swim by reading a website. This is something you can't approach purely conceptually. You have to go out and do it. This problem is being taken increasingly seriously by fusion researchers. Small samples have been placed into fission reactors to experience a similar, though smaller, flux of neutrons. There are special facilities to inspect activated materials under electron microscopes and so on. Once a fusion reactor comes online, produces neutrons and becomes activated, it must be left to sit for decades to decay to safe levels. This might be a hard sell, but should be doable. The problem is that while it remains radioactive, it is very hard to perform maintenance on it. If anything is simple as a bolt or as complicated as a cry of pump breaks in the activated area, humans could not go in to fix it. Human machines can struggle in a radiation environment and must be rad-hardened. Special robotic tools exist to work on the jet experiment and are being upscaled and made more resistant to work on ETA, which will likely be the first fusion experiment to seriously deal with neutron activation. Laser inertial fusion concepts and any credible private fusion start-ups would do well to bridge this technology gap. The radioactivity aside, activation is a problem simply because it changes one chemical element into another, a process called transmutation. Once carbon-14 decays, it becomes nitrogen, which is different chemically. If a certain proportion of the atoms which make up a material change to another chemical element, this could have an effect on the material's tensile strength, density and so on. A load-bearing beam might become weaker and break, an important component might stop conducting electricity and therefore fail. Again, the diverter is one particularly vulnerable area, but so too are any diagnostic or fueling ports, mirrors in a laser inertial scheme and the like. Under neutron irradiation, tungsten transmutes to rhenium and dosmium. Atoms of these two newly generated metals seem to clump together within the original structure and change the mechanical properties. Tritium does not occur naturally and must be generated or bred within a future power plant. Fortunately, lithium-6 can be activated by neutrons to produce tritium and a little extra energy. One tritium nucleus undergoes a fusion reaction to produce one neutron, which in turn produces one tritium again. The cycle works assuming perfect efficiency, but as we have seen, some neutrons will be absorbed elsewhere, some tritium will be lost or decay away radioactively. Fortunately, beryllium can release two neutrons when struck by a single one. Therefore, for some of the original neutrons the tritium production is doubled and the losses can be made good. The plan then would be to have beryllium just inside the first wall, probably in solid form as it is on jet, and some compound of lithium. The lithium could be formed into removable blocks or round pebbles. One promising idea is to use a lithium-led eutectic, a mixture with a lower melting point than either of the metals on their own. With a low melting point, the liquid metals could be pumped through the blanket to simultaneously breed the tritium and take away some of the heat. Another good possibility is flyb, a mixture of lithium and beryllium salts with a relatively low melting point, which could also be pumped through the blanket. All of these approaches have their advantages and disadvantages, but none have yet been built or tested in reality. One way or another the deuterium tritium fuel cycle has to be solved. In order for deuterium tritium fusion to be commercially practical, which as I have explained is by far the most favourable from a plasma physics perspective, Tritium breeding must be made to work. There is no economical pathway to obtain tritium any other way than from a fusion power plant itself. Indeed, for fusion to provide any significant fraction of humanity's energy, hundreds of plants would need to be built. The early power plants would not only have to produce enough tritium for themselves, but for subsequent ones as well. Commercial fusion does have a small amount of room for manoeuvre, initially, with regards to tritium. If the entire world inventory were to be used up before a self-sufficient power plant could be established, the plant could run with deuterium alone at an energy loss. The deuterium-deuterium reaction produces tritium directly half the time, and neutrons capable of tritium breeding the other half. Over the period of perhaps a few months, it would be able to transition to the nominal 50-50 deuterium-tritium mix. Going back to neutron reactions, some very heavy nuclei such as uranium and thorium fission when absorbing neutrons. It is straightforward to avoid unintentionally having fissile material present, but it does open up the possibility of a combined fission-fusion plant. The fusion reactions could even be run at an energy loss as long as they produce enough neutrons for the fissile component to be worth it. Such a scheme could be safer than a conventional fission plant if the fissile material were kept subcritical, meaning that it alone could not sustain a chain reaction. In that case, as soon as confinement to the plasma was switched off or failed in an accident, all the fission would immediately stop dead. Besides being absorbed, the other major effect of fast neutrons is simply how hard they smack into any nuclei which happen to be in their way. Ignoring for a moment any activation, a collision with a fast neutron would transfer an energy in the range of millions of electron volts to the target atom, far more than it would get from any chemical reaction. With such a large energy, the atom would then collide repeatedly with its neighbors in a cascade like a bunch of billiard balls. In a fission reactor, it's possible to immerse the fuel in a liquid such as heavy water or a molten salt. This is indeed how most fission reactors operate. For a liquid, the constituent molecules are not bound together, they are constantly moving and colliding. In this arrangement, being struck by a neutron is not a big deal. A fusion reactor, on the other hand, must have a solid first wall as I discussed previously. Atoms in such a material are formed into a lattice or similar arrangement, which does not take kindly to being treated as a pool table. If a given lattice atom receives anything more than a small threshold amount of energy, it will settle away from its starting location. Gaps will open up in the material, it will become brittle and swell in size. Gaps builds up in the cracks and voids, further stressing the material. Neutron damage is quantified by a metric called displacements per atom, or DPA for short. For the first wall and similar components, this number may become greater than one, and even in the tens and hundreds. Needless to say, this many displacements is bad news for an otherwise ordered solid material. To give an example, here is a study where mirrors were exposed to neutrons up to the levels of 0.001, 0.01, and 0.1 DPA, compared to a control without any exposure. As the neutron damage is increased, the mirrors become pitch black, and at the highest dose in the study the material has shrunk by over 2%. Granted, mirrors are generally very finicky and susceptible to damage. For most other components it doesn't matter if they end up looking scuffed as long as they do their job, but this gives an indication as to how much damage neutrons can do. On the bright side, the study also shows clear evidence that the damage can be repaired somewhat by annealing or heating the material. Indeed, this is something many other experiments have reproduced. The damage scales up with the number and energies of the neutrons. In many ways, this is uncharted territory for science. For the reasons I've mentioned, even fission plants do not reach these levels of neutrons. One particular concern is that microscopic gaps will open up in the material's surface, trapping tritium among other substances. This would make the reactor components more radioactive on top of whatever activation is happening. The damage, particularly to the first wall, from neutrons and plasma in conjunction is likely to be severe in any fusion reactor. One family of materials being developed worldwide to deal with it are reduced activation ferritic martensitic steels, alloys of mostly iron, just under 10% chromium, and up to a couple of percent other metals. The name informs us that this type of alloy consists of elements which do not easily become radioactive from neutron exposure, and that they are rapidly cooled during manufacture in order to harden them. If you want to know how good they actually are at doing their job, then so would I. Quite a few tests have been done on small samples of these steels, but rarely if ever to the level of a fusion power plant, and never on a fully structural component. The exact composition of these steels is still being optimised. We will know the full story when all the development and testing is complete, but until then these types of steel remain, on paper, good all-round options from the points of view of structural strength, neutron damage and activation. A closely related concept is that of oxide dispersion strengthened steels. This involves microscopic clumps of metal oxides, yttrium oxide for example, being implanted throughout a given steel. In very simple terms, any crack which forms as a result of neutron bombardment stops growing when it hits one of these clusters. Since the cluster is small and evenly distributed inside the steel, the cracks are also kept small and as a result the damage is limited. Again, these alloys show promise, but research is still ongoing as to how to actually mix the oxide clumps into the steel in industrial quantities. There is one exception to the all-solid first wall currently being trialled. It's possible to coat the wall in a thin layer of liquid lithium. Not enough to stop most of the neutrons or breed all the tritium required, but just enough to take some of the edge off the neutron and plasma damage. The layer does have to be very thin because otherwise the lithium would be peeled off the walls due to the magnetic field. The Lithium-Tochemac experiment at Princeton is having some success with this concept. Naturally, for an inertial fusion configuration without a magnetic field, the liquid lithium layer could be made much thicker. Questions about material damage have been on the minds of fusion researchers for decades. IFMIF, the International Fusion Materials Irradiation Facility, was originally slated for construction as early as the 1990s to address these problems. Japan was granted the right to construct IFMIF as a consolation for losing out on hosting ETA. Currently, a prototype is just about finished in a building at Rokasho Nuclear Reprocessing Facility. Eventually, it will be able to produce 14 megaelectron volt neutrons relevant to fusion in enough quantities to produce tens of displacements per atom. This would just be for a sample of a few centimeters, however, not a full-scale wall or blanket. Beyond the immediate engineering concerns of building a working fusion power plant, something to keep in mind is the global supply of materials. Deuterium is safe as a fuel source, but fusion must compete for lithium with battery manufacturers. No matter how batteries are made and disposed of, atoms of lithium stay lithium, whereas in fusion they are permanently used up. On the other hand, fusion will generate helium which might increase in demand due to its many applications. Relatively rare metals such as tungsten and chromium will also be transmuted away permanently so this could be a concern. Fusion could find economic applications beyond power generation. The enormous amount of neutrons from fusion might possibly be turned into an advantage. In the near term, fusion reactors could and should be used to study the effects of neutrons and materials. Fundamental science experiments might well lead to the development of new alloys for space or high energy industries. Longer term, existing nuclear waste could be transmuted into stable isotopes. Eventually, transmutation could create valuable elements in industrial quantities even when running at an overall energy loss. As I mentioned previously, this is the pitch of NASA's lattice fusion concept, which can otherwise not hope to achieve a net energy gain. So to sum up, I have talked in broad terms about the practical aspects of engineering a fusion power plant. A vacuum must be maintained, fuel injected, and the helium ash separated out. The energy of neutrons, which is the lion's share of the fusion power output, can only be absorbed as heat. This heat must be transferred out of the reactor to drive turbines. The energetic neutrons cause several types of damage. Transmutation is when a nucleus changes from one chemical element to another. Activation is when a nucleus becomes radioactive in the mid to long term. Collisions with neutrons cause each atom, on average, to be displaced many times within a solid material. This and exposure to the plasma itself can greatly damage a fusion reactor. Fusion power plants must also breed enough tritium to be self-sufficient. A problem which, despite the cowboy attitude of certain startups, has never demonstrably been solved. There's definitely been a lot of information, so thanks for watching.