 At thermonuclear temperatures, bonds between molecules break down and electrons are stripped from atoms. The resulting plasma has all the properties of a fluid such as flowing and taking the shape of its container, as well as responding strongly to electromagnetic fields and electromagnetic waves. In a previous video I explained why plasma must be confined to stop it from coming into contact with its containment vessel. Now let's look in detail at how magnetic fields can do so. At school you may have performed an experiment where you looked at iron filings around a permanent magnet, which may have looked something like this. The filings line up along so-called magnetic field lines. These are the lines you would trace out if you took a magnetic compass and followed the arrow from the north pole to the south. In other words the magnetic field points along the field line. You may also remember a version of the right hand rule. Keeping the fingers at right angles to each other, line up the index finger with the direction of the magnetic field, the thumb with the direction of motion of an ion, and the middle finger will show the direction of magnetic force. For a negatively charged electron, do the same thing with your left hand. These simple memory aids illustrate that the magnetic force is always perpendicular to the direction of motion. This means that a charged particle moving at a right angle to the magnetic field will orbit around it in a circle. If the particle is moving parallel to, or in other words along the magnetic field line, it feels no force whatsoever and is free to move however it likes. In general a particle will have a combination of these two motions which adds up to a spiral around a given magnetic field line. So far things look pretty peachy for controlled fusion. Just keep hot deuterium and tritium nuclei orbiting a field line until they collide many times and eventually fuse. As an added bonus the fusion energy is split. 80% is carried by a neutron which is free to escape the magnetic field and be captured, while 20% goes to a helium 4 nucleus which will remain confined and give up its energy to heat the fuel even more. There are a few potential problems which must be carefully avoided however. One thing that magnetic field lines do not tell you at a glance is how strong the field is at any point and therefore how much force the particle will feel. Suppose that the magnetic field is pointing out of the screen but is stronger on the left side than the right. A particle will make a tight semicircle while it's on the left because of the large amount of force it feels there but a very wide semicircle while it's moving on the right. Overall then after the particle completes an orbit it has shifted its position. Ions move downwards, electrons move up. This type of effect is called a drift and it occurs either due to variations in magnetic field strength or other forces such as gravity. If plasma particles drift to the walls of a reactor this would be very bad for confinement. Every charged particle moving in a circle is itself a small magnetic dipole just like a very tiny bar magnet with its own field. The field of a single electron on its own is insignificant however the additional magnetic field coming from a plasma is important when the density of electrons and ions becomes quite large which is advantageous for fusion. The plasma feels the influence of a magnetic field and in turn affects it back. This collective action is called magnetohydrodynamics. One consequence is that the magnetic field lines are what's called frozen end to the plasma and must move along with it. It is as if a blob of plasma is a bead and the magnetic field line going through it is a string. The bead can move along the string freely but it cannot move sideways unless it drags the field line with it or vice versa. This means for example that a high pressure can cause plasma to expand and alter a given magnetic field. Any design for a fusion reactor must be careful that magnetohydrodynamic effects or instabilities do not degrade the plasma confinement. For example a moving blob of plasma could drag a certain field line out of the reaction region allowing the rest of the plasma on that field line to leak out. If the field geometry changes drifts could start to occur. Hot plasma regardless of whether it's a magnetic field will experience turbulence just like any other fluid. An aeroplane might get a bumpy ride due to the way different air masses mix chaotically. Similarly a fusion plasma will experience eddies and whirls which churn it up, tangle up the field lines and allow particles and the energy they carry to escape. Turbulence and similar instabilities occur on the scale between individual particle orbits and the magnetohydrodynamic behavior of the entire plasma. They are not fully understood theoretically so it is not possible to precisely predict how a future experiment will perform. So to summarize a magnetic field spins particles right round baby right round like a record right round. Plasma is trapped mostly attached to magnetic field lines and therefore quite well confined. Problems arise when individual particles drift out, the collective action of the plasma alters the applied magnetic field or random turbulence begins to mix it up. As a final note the stronger the magnetic field the better the confinement is in general. Particles orbit more tightly, drifts are less pronounced, magnetohydrodynamic and turbulent instabilities are less problematic. One of the simplest approaches to magnetic confinement is to pass a very large current through some material. It does not even have to be a plasma initially because if the current is large enough the temperature will quickly rise to plasma temperatures. The current creates ring-like magnetic field lines which exert an inward force which compresses the plasma. Thus the current raises the temperature and the resulted magnetic field raises the density, both the necessary ingredients for fusion power output. The whole thing is bound to last only for a short amount of time but that's worth it if the fusion energy released is much greater than the electrical energy input. This type of approach is called the Z-pinch because the current is flowing along what is usually the Z-coordinate of a cylinder. However the Z-pinch is never really able to reach its full potential because of a pair of instabilities. The narrower the plasma column the more compacted the current becomes, the greater the magnetic pressure. The more compressed the pinch already is the faster it will continue compressing. If by pure chance the plasma cylinder begins slightly narrower at a few points then those points will pinch off much more quickly. What should have been a perfect cylinder gets broken up into isolated chunks giving this the name of the sausage instability. This eventually cuts off the current and ruins the entire plasma column. If the Z-pinch cylinder has a small bend or kink as another instability is called then the resultant magnetic force acts from the inside to the outside of the bend. The resulting force makes the bend even bigger. This means that a small and unavoidable initial imperfection grows until the plasma is twisted out of shape. The kink instability remains a danger in some form or another not just for the Z-pinch but for most other magnetic confinement techniques. Here the kink instability has been photographed inside a toroidal Pyrex vacuum chamber. Instabilities rule out the possibility of getting a fusion gain directly with a Z-pinch because the energy confinement time is always far too low. Z-pinchers are nonetheless used to do fundamental science experiments. Arrays of wires each of which an individual Z-pinch can also be used to generate X-rays and compress a capsule inertially. The magnetic mirror is another early approach to fusion. A pair of magnetic coils some distance apart create a cylindrical magnetic field which bulges out in the middle. Particles which are moving perfectly parallel to the field lines will feel no force and escape. Particles moving at right angles to the field will orbit in circles. The idea of the mirror is that the magnetic field is strongest near the coils and weaker between them so that even some of the particles moving along the magnetic field become trapped in the middle. I won't go into the full mathematics but depending on how closely the motion of a particle is aligned with the magnetic field there is a cutoff at which it will follow the field line and escape. The problem is that collisions between trapped ions and electrons will constantly be knocking more and more of them onto trajectories parallel to the magnetic field lines and consequently out of the mirror. The end effects can be mitigated by making the mirror machine very long perhaps with multiple coils so that proportionately only a small fraction of plasma is affected. Another improvement is to have a pair of baseball shaped coils at both ends of the mirror. Regardless of what happens at either end the magnetic mirror is not a successful concept because it is magneto hydrodynamically unstable. In the weak field region midway between two of the coils the plasma will tend to balloon outwards dragging the field lines with it and therefore degrade confinement. As a result the confinement time of mirror machines is poor and they are lucky to get up to 10 million degrees let alone the few hundred million needed for a high fusion output. Early on in fusion research it was recognized that curving the mirror machine back in on itself in a torus would completely negate the end losses. A configuration with a purely toroidal magnetic field so that all the field lines go around in circles is still vulnerable to the sausage and kink instabilities. Historically machines such as zeta in the united kingdom showed great promise early on but then failed to reach high temperatures. A concept called the bumpy torus involves a number of magnetic mirrors chained together in a ring but suffers from all the same issues. Another problem with the torus is that the strength of the magnetic field is no longer constant. In simple terms the larger the circumference of a circular line the lower the field strength. The fact that the field is therefore weaker on the outside than the inside means that ions and electrons will begin to drift in opposite directions and the whole thing becomes unstable. What must happen to counteract this problem is that magnetic field lines must somehow be twisted in such a way that part of the field line is on the outside of the torus and another part on the inside. This would mean that particles would drift one way half the time and drift the other way the other half of the time so that the drifts cancel out. Two approaches to solve this problem are the two most mature and successful types of magnetic confinement machines the tokamak and the stellarator. Tokamak is a Russian acronym which stands for toroidal chamber in a magnetic coil. In it the plasma retains a toroidal shape but a current is passed through the middle of it. The effect of the current is like the z-pinch it creates an additional component to the magnetic field in circles around the torus. In other words the magnetic field making the donut shape is provided by the coils in the machine and a field wrapped around the donut is generated by current through the plasma. Overall the two components combine so that the magnetic field is twisted up. Each of the field lines goes from the outside of the torus to the inside and back just as required. The stellarator which was invented by American Lyman Spitzer to be an artificial star is more like a mobius strip. The plasma takes the form of a ribbon twisted on itself. The magnetic field is shaped by the coils of the machine to wrap back in on itself. Each of the twisted field lines comes closer and then further from the axis of the torus at the cost of much greater engineering complexity. In short the defining feature of each of these machines is their greatest challenge. A tokamak must always maintain a current running through the plasma and if the current stops so does any hope of thermonuclear reactions. The stellarator must be precisely engineered to maintain its characteristic magnetic field and if any of its complicated coils deviates from the design during construction or afterwards the machine too will fail. Once a plasma is established usually by an electrical discharge resembling a large spark it must be heated to thermonuclear temperatures. The current which must be passed through a tokamak plasma is a good first step but generally not enough. Additional thermal energy is provided by beams of particles and electromagnetic waves. Charged particles orbit around magnetic field lines at the so-called cyclotron frequency which depends on the magnetic field strength and their mass. Electromagnetic waves such as those used in a microwave oven are absorbed if they are of a matching frequency. In such a case they resonate with the orbiting particles. Energy can be deposited in the plasma by targeting a specific type of particle in a location with a particular magnetic field where the resonance occurs. Ion cyclotron resonance heating is done by radio frequency waves of tens of megahertz. Electron cyclotron by microwaves in the hundreds of gigahertz range. The so-called lower hybrid resonance between these two is also often used. For comparison most wi-fi and microwave ovens use 2.4 gigahertz waves. The main challenge with wave heating is successfully delivering the electromagnetic waves to where they are needed. Just like light microwaves and radio waves can be reflected or refracted by the plasma, potentially all the way out of it depending on changing conditions. In that case the energy is wasted. Beams of particles are created by an accelerator and injected into the plasma so that they can deposit energy and heat the plasma up. Since the particles are usually deuterium atoms they also add some fresh fuel. To be able to pass into the magnetic field undeflected the particles must have a net charge of zero so this method of heating is called neutral beam injection. However particle accelerators can only accelerate charged particles so the deuterium atoms must have an electron stripped off or added then be accelerated and then be neutralized again when moving at high speed. Both approaches have their trade-offs but the version where an electron is added to make the deuterium atom negative is more promising for future tokamaks where each one has to be accelerated to a higher speed. Generating neutral beams is less energy efficient than microwaves but the advantage is that the beams will surely go where they are pointed. If deuterium tritium fusion is taking place sufficiently quickly and the confinement of heat is sufficiently good the fusion energy carried by helium-4 nuclei will begin to be the dominant heating mechanism. Mathematically this will happen above a fusion gain of 5 which is a prequisite for any successful power plant. How such a large number of very energetic helium-4 nuclei will behave and interact with plasma instabilities is still a major unknown for magnetic fusion because this regime has never yet been reached. A tokamak must have a large current flowing through the plasma. The simplest and earliest approach to do this is by electromagnetic induction. Outside the vacuum chamber are coils which act as the primary coil of a transformer and the plasma torus is just like a giant secondary coil. As a consequence of Faraday's law the current induced in the secondary is proportional to the rate of change of current in the primary. In an ordinary transformer there is alternating current in the primary and as a result also the secondary coil. But inside the tokamak the current must be constant and therefore in the primary coil it must rise constantly. If it is rising steadily amp by amp eventually it will reach a thousand a million or whatever the engineering limit is on the coil. The current in the primary will then plateau or fall and the tokamak plasma will fail one way or another. Fortunately the other methods of heating the plasma can also be used to drive a current through it. If the neutral beams come in at an angle to the torus there will be a resulting flow of charge, an electrical current. This is the typical setup on high performance tokamaks. Electromagnetic waves can also drive currents. The lower hybrid resonance is particularly good at making electrons flow around the torus, so many tokamaks have microwave systems for lower hybrid current drive. There have been some proposals to enable the helium for fusion products to also induce current but they are as yet totally unproven. I have been talking about a tokamak plasma as if it is the same shape as a Homer Simpson donut where the cross section is circular. The size of the plasma is limited by wherever a magnetic field line intersects a solid wall. Any plasma which makes its way out of the center of the donut to such a field line will quickly cool and recombine into a gas. On the flip side the solid wall heats up and erodes. A better idea is to add another current through a coil below the plasma so that the magnetic field lines make something more like a figure of eight. The plasma will then flow down the legs until it strikes the walls of the reactor. The vacuum vessel can be armoured at those strike points to best resist this hot plasma. This is called a diverter, a particularly tough device at the bottom of most tokamaks. On advanced tokamaks the diverter is or will be made of tungsten, a dense metal with a high melting point. Nonetheless this is a very challenging area for magnetic fusion because in a full power reactor a huge amount of power will be dumped on the diverter. It has been likened to a spacecraft re-entering the earth's atmosphere. Many ways to tackle this problem have been considered such as purposefully inducing the plasma to radiate away all its energy by Bremstrahlung just before it reaches anything solid. This is still a very active area of research particularly on smaller tokamaks which might never be able to get high power output and therefore irrelevant to fusion research. The diverter region is also a logical place to put vacuum pumps to pump out plasma after it recombines back into a gas. In a fully fledged power plant this would allow helium for waste gas to be removed and separated out. Other coils are added around tokamaks to shape the magnetic field because it turns out that the confinement improves if the plasma is extended vertically. This is how most modern tokamaks operate. The plasma cross section is also somewhat triangular usually pointing outwards although the idea of so-called negative triangularity pointing inwards has recently had some successes. Taken to the extreme this results in the spherical tokamak concept which as the name suggests has an almost spherical plasma pierced through the middle by a metal column. Spherical tokamaks still have a diverter at the bottom and perhaps a second one directly above the plasma. This approach is good from the point of view of scaling down tokamaks but it makes the central column very vulnerable. Deuterium tritium reactions which any future power plant must have produce high energy neutrons which would make swiss cheese out of that central column. More on this in a subsequent video. Overall no matter what the cross section of the plasma looks like the temperature and density peak in the middle and decrease as you go outwards. When the input power is high enough the plasma transitions to a high confinement mode of operation usually referred to simply as H mode. This means that the plasma density begins to rise much faster at the edge of the plasma and as a result reaches a much higher peak in the middle. This is very advantageous because a larger density means a larger fusion power output. The benefit to performance is seen as practically essential for any future reactor. The drawback to H mode is that several times every second an instability causes a fairly large expulsion of plasma often not just to the diverter but to the walls in general. Fittingly this type of instability is called an edge localized mode or elm. Elms are not so bad from a confinement point of view but they are very unfavorable technologically. If heat and particles are to leak out of the plasma to the walls of a reactor better that they do so steadily rather than in a single violent eruption. H mode therefore becomes a trade-off between getting a large power output and minimizing damage to the reactor vessel. There are several promising solutions at various stages of maturity for how to get the best of both worlds. Depending on plasma conditions there are different types of elms with some less destructive than others. Other modes of operation without elms such as the quiescent H mode and the improved confinement or I mode have been observed occasionally. It is hoped that future reactors will make use of these modes. This is why upcoming fusion machines are experiments and there is still a lot of work to be done before a commercial reactor can be built. I have mentioned tiny instabilities which are happening continuously throughout the plasma and intermediate scale ones like elms which happen more occasionally. There is another type of instability which is large and dangerous. Occasionally Tokamax experience a disruption where the entire plasma wax itself into the walls and dumps a huge amount of thermal energy there. For years experiments have attempted to predict when a disruption might happen and prevent it by quickly filling the reaction vessel with cold gas or with shards of frozen material to take up the thermal energy instead. In principle most disruptions can be caught and mitigated so that they don't do any damage. However in a future reactor running 24-7 disruptions would be very uneconomical because they mean that all the effort of heating up the plasma has been temporarily wasted. Reports of artificial intelligence being used in magnetic fusion experiments usually refer to the detection of disruptions. It is of course very useful to be able to catch such disruptions early but understand that AI is not a silver bullet to solving every possible problem in magnetic fusion. Speaking of the walls of a Tokamax, early versions were covered with carbon tiles. Small numbers of carbon atoms would become eroded away and circulate through the plasma where each one is 36 times better at radiating away energy by Bremstrahlung than hydrogen. When carbon atoms cool at the edge of the plasma they chemically bond to the hydrogen isotope fuel to form things like hydrocarbons. For a fully fledged fusion reactor this would be a major problem because relatively large amounts of the radioactive tritium fuel would become locked up and eventually decay away. Any future magnetic fusion power plant must therefore have walls made of metals which do not absorb hydrogen isotopes nearly as much. Two existing Tokamax have made a conversion in the mid-2000s. Jet in the UK to beryllium tiles and Azdex upgrade in Germany to tungsten. Paradoxically it turned out that the small amounts of carbon were actually beneficial to confinement. After switching to metal walls the plasma temperature actually became lower inside those machines. During the decade and a half since then experiments have been working to restore fusion performance to the level it was with carbon tiles. Most of what I've mentioned so far, heating, diverters and H-mode of sorts, applies to cellarators as well as Tokamax. The difference is that a stellarator does not require a plasma current to twist its field lines. Indeed, currents are undesirable for stellarators. This means that a fusion reaction could sustain itself indefinitely without any external input power in the stellarator. The flip side is that the magnetic coils are very complicated geometrically. Most advanced magnetic fusion experiments now and in the future have superconducting coils to keep down the required power. That's not a problem for the simple coils of a Tokamax, but it makes it very challenging to engineer stellarator coils and keep them running. Overall, Tokamax are the most ubiquitous and so far the most successful magnetic fusion machines. Due to their simplicity and relative resistance to many of the most debilitating instabilities encountered early on, Tokamax achieved early success and historically gained momentum. There are other magnetic fusion concepts such as the field reversed configuration and reversed field pinch among others I have not mentioned here, but none have yet approached or surpassed Tokamax in terms of triple product and therefore fusion performance. One thing I hope you take away from this video is the kind of work being done to advance the field. Progress in magnetic fusion is not a case of building a bigger machine, switching it on and getting it to work on day one. A lot of optimization has to be done to finally perfect fusion power.