Thanks to everybody for giving up some of your time on a Saturday to come out and say a few words about fusion. So whenever we were first organising this event a couple of months ago, it was just at the time that GPT was exploding on to the scene. And I, like I'm guessing many of the people in this room spent some small amount of my time playing around with it and seeing what it was capable of. And that all culminated in the title for our talk today.
So this is one of the things that I asked GPT three. I said, well, you know, give me a nice interesting title for a public lecture on fusion and, and this was the result. Although it wasn't the only one that almost made it in. Here are a few of the honourable mentions that I came up with, so quite a few of these I was quite happy with.
So, you know, whether or not the future of AI is to become the salvation of humanity, its downfall, or just a footnote in history, at least we'll have this moment. Right? So getting on to Fusion, whether or not you're aware of it, most of the energy on Earth already comes from fusion. Right. And it's a bit sneaky, but that's because most of the fusion energy on Earth comes from the sun.
And the sun delivers about. What is it? 44. Well, at the time it's 4.4 times ten to the 16 watts of fusion on average to the Earth's surface. That's a lot of zeroes. And so to put things in context that something like 44 million of our gigawatt power plants, and so we harvest a very small amount of that energy over time. So some of it directly by a solar energy, some of it indirectly by other means.
So, for instance, the uneven heating of the atmosphere leads to wind that powers our wind turbines. We also have biomass, which is fuelled by the sun's energy, and this ultimately eats the fossils and the fossil fuels. So fusion works already. What we're trying to do is cut out the middleman and get fusion to work here in the labs on earth. And so we can hear two different approaches to that today.
One that I'll talk about in Georgia we'll talk about and one that [INAUDIBLE] talk about at the end. But the common theme is here is that to get the fusion energy out, we need the fusion reaction. This is probably set a lot of you know, but the idea behind fusion is that were going to take two light elements. If you bring those light elements close enough together, then they're going to fuse.
We're going to focus on one particular fusion reaction here today. Turns out to be the easiest for us to achieve in a lab. That's between two different ions of hydrogen called deuterium and tritium. When you bring them together and they fuse, you get these two products out, you get a helium ion and a neutron.
And as is well known now from Einstein's famous equivalence of mass and energy, if, as it turns out, these products have less mass than what goes into it, an energy must have been moved into some other form, in this case into kinetic energy. So the leftover energy is about 17.6 metres of energy. And this reaction, which is about 10 million times as much energy that comes out of gas combustion.
So there's a huge amount of energy here. And it turns out that the products aren't too difficult to find. The deuterium is readily harvested from seawater. So that's not the big issue. The tritium is not naturally occurring, at least not in large amounts. And so we have to breed the tritium by bombarding enriched lithium with a neutron that it gives our tritium and some energy out.
Though this process is limited by the amount of lithium that's going to be available and the different ways they estimate it. But they come up to something like 20,000 years. So the hope is if we can get fusion to work, then it's going to give us a huge amount of energy and there's a lot of fuels that's going to last us for a long time. And so there are two main approaches to doing this. One of them called the Inertial Confinement Fusion Art he's going to talk about later.
George and I are going to talk about a version of thermonuclear fusion, which is the way the sun works. Right. And so the problem that we have is if we're going to take two of these hydrogen ions and we want to bring them very close together for fusion to happen. But to do that, you have to overcome the natural tendency of these light charges to repel one another due to the Coulomb force.
And so the we the way that we do this in thermonuclear fusion is we basically just heat the gas up really, really hot. So if you heated up hot enough, then the random thermal motion of these particles become sufficiently fast that the particles occasionally can come very close to one another, overcome the Coulomb barrier and fuse. Now it turns out the temperature you need to do that for this deuterium tritium reaction is about 100 million degrees.
So that's hotter than the hottest parts of the sun. So how are we going to get this to work? There are clearly a couple of issues that we have to overcome. One of them is the fact that, you know, we eat something up hot in the sun. How are we going to do that? So how do we put the energy in to begin with? And the second challenge that we definitely have to overcome is how do we keep this thing insulated long enough, keep the energy in enough fusion reactions occur to harvest the energy.
I should point out as a side note that at this temperature, way below this temperature at about 10,000 degrees, this gas becomes ionised and turns into a plasma. So everything going to be looking at here is an ionised gas called the plasma. So how are we going to do this? Well, the basic idea behind magnetic confinement, fusion, which is what I'll be describing, is to take magnetic field lines and wrap them around our plasma and keep all the energy in.
And this way. So what I'm going to try to convince you of here in this slide is that our aim is roughly, very roughly speaking, to get one gram of this hydrogen ions at about 100 million degrees for about a second. And the physics behind it is pretty simple. What we're saying is that we have some fusion power which is being created in our plasma. And what we need for this to be self-sustaining is that to at least balance the rate at which the energy is leaving the plasma.
And so we have some thermal energy density here divided by the time it takes for that thermal energy to leave our plasma. And that's got to balance the rate at which the energy is going in. So the power delivered by fusion is going to be the energy of one of these fusion reactions. Times to the fusion cross-section shown here.
I'm the density, the product of the density of our reactants, the deuterium and the tritium, the simplicity here, I'm assuming those two densities are going to be the same. So there's like the square, the density, and we first get our constraint here. Now, by staring at this, what you find is that if you look at this fusion cross-section for the nuclear physics, it has a temperature dependence and you find a peak in this thing at around 100 million degrees.
So that's what sets our temperature to be as high as it is to maximise our fusion cross-section. What about this one, Graham business that comes from considering the stability of our plasma? So we know both from theory and from empirical observations, that there are limits on how much plasma you can stuff in your reactor before things start to go microscopically unstable. And in particular, if you try to put in densities much in excess of 10 to 20 particles per cubic metre,
then we find that you have these macroscopic stability limits. At the same time, even if you say below that and everything's macroscopic, be stable, then hidden underneath what you're kind of looking at from the outside, you see little microscopic of instabilities that start to occur. You get little small scale turbulence. This mixes hot and cold regions. I'll talk about this a lot more throughout the talk.
And the point is this limits the temperature gradients that you can sustain in your plasmas. And since you want to be hot in the middle, I did million degrees cold at the edge. So you don't melt walls. This limits roughly the size of the device. You're going to have to have to make this work. So it gives you a volume at something like one cubic metre. And so if you combine these two things, you find that you need about a gram of your fuel in the device at any one time.
And so finally taking these all these numbers, plugging them back in up here and solving for my energy confinement time. Yet something which is on the order of a second. So this is really what we're aiming at in the fusion program. And so the first challenge I'm going to discuss is just how do we get the energy into the plasma at the outset? Because it seems like a pretty big problem, like heating something up to 100 million degrees.
It turns out this is actually one of the best understood parts of the problem. And it turns out that the ways in which we heat it up are fairly standard ways that we see in other aspects of our lives. So the first thing that we tend to do is we use element heating. And that's just saying that we we work in the same way that a light bulb filament would work. And so you run current through this plasma, which is not a perfect conductor, has some resistivity.
So when you run current through it, it heats up and it turns out that's pretty efficient, up to some tens of billions of degrees, maybe a third of the way. We need to get to our our fusion temperatures. And at some point, when the temperature gets very high, this resistivity of our plasma decreases and this becomes efficient way to find other ways to heat up plasma. And there are a variety of ways we do it.
One of those common ways is to shoot in radio frequency waves and have these things resonate with the motion of our charge particles. And this accelerates our charged particles and heats them up. Okay. And so this is very effective, as is evidenced by scene such as this one, where you can see a bridge which is built such that its resonant frequency would resonate with the wind blowing past.
Right. So this can be a very powerful phenomenon. It can get us the rest of the way up to 109 degrees if we can keep the energy in long enough. And so the basic idea of how are we going to keep that energy in with magnetic confinement, fusion is illustrated here. And so it's basically casting your mind back to your first year undergraduate physics. We simply take a charged particle. Shown here, I put it into a magnetic field which is coming out of this board.
And you have some Lorentz force which is acting on that particle in a direction which is perpendicular to its motion and perpendicular to the magnetic field. So all it does is it take our charge particle and it makes it go in a circle about our magnetic field line. And now along the magnetic field is now this thing doesn't act on our particle in the direction along the field, so it's free to string along the field however it likes.
And so in the end, you get these sort of sort of helical orbits of our charged particles that stream along these lines and gyrate about them. So that's the idea. And. The stronger your magnetic field is, the smaller this radius becomes. So if you put it really strong magnetic field, then into a pretty good approximation, the particles just stream along magnetic field lines.
So the trick then is to find a way to take these magnetic field lines and somehow confine them to some surface or some volume. So that the particles can't leave. That's right. It turns out the only way to do that is to make them look like doughnuts. And so this is basically saying we're taking our magnetic field. It's some kind of vector field. We want to confine it to some surface. And this, Harry both tells us the only way to do that is with something topologically equivalent to a tourist.
It's called the Harry Ball Theorem. Because of this diagram showing you on the left, the idea is if you try to take some ball or sphere with little hairs all over it and you try to comb those hairs, so they all lay in the surface, much like your vector field all needs to lie inside the surface. You find you could never make it lie flat everywhere. There was always be some cusps appearing somewhere which appear which correspond to zeros in your vector field, which is an acceptable way.
This is why everything looks like these these doughnuts whenever you look at any fusion devices. So what I would like to do, if you'd indulge me, is look at the simplest possible doughnut and see if it works. And spoiler alert, it's not going to work. But in finding out that it's not going to work, we're going to see some of the physics we need to make it work. So here's a simple possible device we can kind of think of. We have just a current carrying wire.
Parent carrying wire makes you feel, which gives you circles which encircle that wire and whose strength drops off like one over the distance from the wire. So what we're going to do is going to put a charge particle, give it like a proton into this magnetic field and see what happens. So naively, based on the picture I've just drawn for you earlier, we expect these things just filed around the seed line and go round and round and round.
That's not going to happen. It's not going to happen because any time you have a curve magnetic field like this, it's going to have an in homogeneity, in the magnetic field strength. And that magnetic field string is going to stop these nice close orbits that we've been talking about.
So that's illustrated here. So if I have my charged particle here and I zoom in on it and naively I expect it to generate around the field line and if I have a stronger field on one side than the other, then locally each gyro radius over here is going to be smaller than it is over here. And so your circle's not going to close. So instead it does something like this and it drifts vertically upwards.
In this case. And so if you repeat the same process now for negatively charged particles for electrons, then they're going to drift downwards. So positive charges go up, negative charges go down. You get an electric field which is generated by this process. And so what we expect to happen presumably is something like this. So we have an electric field particles going to stream long electric field, right. Well, that's not right in the presence of a magnetic field.
The Magnetised plasma gets something which looks kind of like the picture I showed you before in the case of, you know, what you need in the magnetic field. Namely, if I take any force and I put it in a direction perpendicular to my magnetic field, then you get these drifts and they drift for the same reason I argue, before particles are accelerated over this part of its orbit.
And so it's local gyrating. It gets big. But at some point, the Lorentz force turns it around, its decelerated small dry radius. And so it it drifts the direction that's perpendicular, both to the force and to the magnetic field. So if we apply that now in our sort of simple doughnut device, we're going to see that you have drifts which are to the left in this case, maybe to the right out here, just radially outwards. So particle is a stream radially out of your device so it doesn't work.
The solution. Solution, unfortunately, is to make life more complicated. And so instead of having these simple circles, you really do have to use the full set of stories. And so we can take a magnetic field, which before just went a long way around this doughnut. And now we're going to add a component which is the short way around the doughnut. And so my apologies, my lapse into jargon at some point.
This is the toroidal direction, the long way around. And this is a little direction, the short way around. And so here it's illustrated how you generate such a set of fields. The idea is you would take some current carrying coils or magnets and put them encapsulating your plasma volume that will generate this, you know, the long way around. And what you can do is run some kind of currents up through the centre of your device.
That changes in time. I don't want to use a current in your plasma and that generates these little magnetic fields around this way. And so I don't want to belabour the point, but this sort of common approach of generating this so called potent field is not a steady state way of generating your magnetic fields, right? Because to do this, you have to increase the current through this centre column in time.
And at some point you can reach your current limit and you got to turn it off and start over again. So Georgia might talk a bit more about ways around this Hertog. Okay. So why does this help? I've said it's going to solve the problem if I put this twist. Now, let me try to convince you. So again, here now you can do this dot dash line as a line going vertically through the middle of this doughnut.
And so if I took this circle and I rotate it around, that makes my tours. So again, let's put a charge particle here. It has an inhomogeneous magnetic field that's going to start drifting down. And now as it drifts down, it's also tied to a magnetic field which is moving around trying to make it stay on the circle. And so as it drifts down, it drifts off the steroidal surface. Now down here, it's drifting down, back on to the surface.
And so if the device is symmetric with respect to this vertical line, then on average when the particle drifts, it's going to come exactly back to its starting location. So there's no net magnetic drift in this case. And so there's no radial transport of our particle. So what this means is this device, which has so-called axis symmetry called a tokamak, can confine individual charged particles. So this is the picture that I'm trying to sell you based on this single particle picture.
Basically, if I take a bunch of these toroidal surfaces now and I put them inside one another like this, then we're going to find that energy in the particles attached to a given surface. Stay on that surface. So they're like metal insulators. And so you can imagine having a hot plasma in the middle, which is perversely blue here, and a cold plasma at the edge.
Now, in reality, this is what happens in our plasma. So this is visible light which is emitted by the relatively cool plasma at the edge of an experiment just down the road at Culham Centre for Fusion Energy called Mast. And what you're seeing here, turbulent fluctuations of the surface of this toroidal surface in the plasma. And this is due to all these small scale instabilities that I mentioned and the resultant turbulence,
which is going to makes hot and cold stuff in the plasma. And as I'll tell you a bit more about later, this is what limits are confinement time in the devices you just saw a minute ago that this sort of turbulence fluctuations largely went away here. And you get this nice, clean picture of our plasma now interspersed by these unfortunate violent outbursts of the plasma.
And so what we can see something hopeful and something horrible in this picture, I hope will thing is that there's a way to take this turbulence which was mixing the hot and cold and to reduce it substantially. So you can barely see it any more in this image, but at the same time, it's replaced by these arguably much worse outbursts of the plasma. So these things on current devices are a nuisance.
If they're bad enough in certain devices, they can damage some of the surrounding wall and components in a fusion reactor. If these things happen, there's so much energy that's going to be ejected out, they'll just melt the wall. So these are one of the things that community is very worried about, making sure that they can sort out.
There's been a lot of progress on this, but most of the time I'm going to be talking about this turbulence problem in part because it does limit the confinement device and in part because that's what we do here at Oxford. So we're one of the world leading groups in trying to understand this turbulence and how we can reduce it. So the pretty picture I showed you before, these nice concentric, confined surfaces underneath lurking is this turbulence.
So this is something which you can think of as triggering density fluctuations taken from a numerical simulation of one of these tokamaks out in California. And you can see some interesting features of this turbulence that hopefully by the end of the lecture today, you'll understand. So one of them that's quite striking is the fact that this treatment is highly isotropic. So if you follow one of these eddies, say this blue, and here you can see it's very long weighted in this direction.
But by the time it comes over here, you get these little circles which are much smaller than the device. This is quite easy to understand. Without giving you any more physics, giving you it's in isotropic because you put a strong magnetic field in the plasma that introduces anisotropy because the particles are free to move along these lines, but not across these blue lines. Basically map out magnetic field lines in our device.
What I haven't explained is why what actually sets the scale of these eddies in this sort of poetic cut? I'll talk a bit about that later. The other thing that's not quite clear yet is why is the turbulence seeming so much stronger out here on the outside and it is on the inside? That's another feature that will explain today. And so as soon as turbulence is mentioned, then perhaps your heart drops. I don't mind us sometimes late at night when I can't figure out a problem.
And there's good reason because for a very long time now, people have been working on turbulence, using the neutral fluid context. And some very clever people are some examples of which are shown here have despaired about solving the problem. So here's a quote which is variously attributed to these gentlemen. And I guess what I can say is that it has been a big problem. This has been one of the reasons why Fusion has taken much longer than people thought originally.
When people built these devices, they weren't counting on turbulence, mixing everything up. They were thinking collisions would just move particles out and slowly. And that's why you have these ideas of tabletop experiments or fusion might occur several decades ago now. And as it became more and more clear that turbulence was a problem, these devices got bigger and bigger. But what we are going to take today is say turbulence is complicated. Let's not try to approach the turbulence directly.
Let's instead think about what drives that turbulence and see if there's a way we can shut off its drive. So linear physics is easier than nonlinear physics. So let's start there. So what is it that drives our turbulence? So asking the illustrated by this cartoon here you're looking at a cut now a piece of our tokamak. So that would be the Tokamak. We gyrated around in another board and we're going to focus on a little patch of plasma near the edge.
And as I said, we want to seem to be hot in the middle for fusion to occur, cold at the edge. We don't melt their walls. So I expect the plasma on this little side of the patch, the hotter than it is over there. The only other piece of information we need is again, we have an inhomogeneous magnetic field, so we're going to have our ions drifting downwards in this diagram and the hotter ions have more random thermal motion. And so they're going to react down faster in the colder ions.
So what happens if I take this nice picture where everything is perfectly well confined in return, but just a little bit. Then what I see is the following. So my hot particles, my hot irons approached this surface faster than they're leaving that surface. And so you can end up with a net excessive charge here. Up here, they leave this interface faster than the replenish.
So you end up with a deficit of positive charges. And so you end up with this horrible thing we saw earlier that the problem, which is charge separation, positive and negative charge is alternating, which. Give rise to alternating electric fields. You have an electric field and a magnetic field, they give rise to drifts. And these drifts just so happened to reinforce the initial perturbation that we gain.
So this is our mechanism for instability. We need a temperature gradient and we need magnetic field in 1980, and that's basically it. So if this were everything, then we'd be we would be in trouble. But if you repeat the analysis on another little patch of plasma on the inside of our device, then you find the opposite is true. Almost everything's the same. The only thing that's changed is that now the cold part of the plasma is on the left and the heart's on the right.
All the other analysis is identical. And so what you find is that the drifts now actually work to stabilise the initial perturbation. So it's unstable on the outside, stable on the outside, inside. What's going to happen? Well, the competition between these two things gives rise to a critical temperature gradient. If you stay below that critical temperature gradient, then you have none of these micro instabilities.
And if you go beyond that critical temp to gradient, then you have this instability, which is exciting. And so roughly speaking, you can see that here. If we took this horrible device, which is just circles, so purely toroidal field and if you looked at the plasma out here, it would just go unstable and it would grow and grow and grow. That nonlinear turbulence, that stuff happens.
If you instead twist the magnetic field like we're proposing, then you take this instability, the amplitude start to grow. But at the same time, it's being swept along the field line to this stable region where these perturbations now decay. And so the analogy here is with this honey dipper, the idea being that if you take your honey deferred and leave it there stationary with honey on it and gravity pulls it off as gravity in this case is like the instability in your system.
But if you rotate it fast enough compared to the rate at which gravity's pulling it off in the honey to stage on the dipper, because at some point gravity is doing a work for you and pulling the honey back onto the difference. It's the same thing you with the twisting magnetic fields. So what does this mean? It means that sort of the boring but reliable solution to the turbulence problem is to make a really big device.
Right? Because if you make your temperature gradient sufficiently shallow, then as long as I make my device big enough, I'll get the high temperature I need in the core. And everything's fine, at least from the standpoint of turbulence. The other reason you don't want to do this, which I'll touch on later. But it's worth considering. Then what happens if we do inside these instabilities? Let's imagine I don't want an enormous device.
I want to get a smaller device. And so how bad is my turbulence going to be if I cross the threshold? And to estimate this, what I'm gonna try to do is give you some argument for what the size of our energy should be in this turbulent system. And basically, the bigger the eddy, is it worse? The mixing is going to be right? You can make an eddy the size of your device. You're going to mix hot and cold stuff immediately.
If you make tiny eddies, you can take a long time for the energy to diffuse out. So we're going to find that the Eddie Eddie's must be roughly the size of this gyro orbit particles around the lines. And you can think of that in a fairly crude way by saying that if the gyro radii were really, really small and effectively, they wouldn't see the magnetic field and homogeneity at all. There'd be no magnetic drifts and maybe be no instability wouldn't be a problem.
And the other limit with it, a radius is very big compared to these perturbations that I've drawn here. Then the particle doesn't see all these little complicated physics of what's happening in the cold region. The hot region. It just averages over all this stuff. And so the instability goes away then as well. So the only perturbations are going to be a variety of instabilities or those which are the size of our gyrations.
And so here's a pretty movie. Basically, I'm just showing you the simulation where you start out with the money instability. It's already gone away. Things have gone non-linear. Turbulent starts to fill out volume. And so you can see exactly all the stuff we've just been describing, turbulence, which is mostly unstable on the outside and the inside thing to stabilise.
You also see something kind of interesting, which I'll come back to in a moment, which is you see all of these kind of differential flows appearing in the plasma and in regions. We have strong gradients, these flows, it turns out you can have the turbulence which largely goes away. Okay, so I'll come back to that in a few slides. But if we know that the any size of the gyro radius, how does that help us figure out how much confinement time we're going to get?
Well, we can do that by coupling this to a random walk kind of estimate. So random, which are used in lots of different areas of physics, if you cast your mind back to undergraduate degree and your kinetic theory, you might realise the kinetic that these random what we use to estimate heat transport and neutral gases in there is collision. But you're moving things around here. It's a turbulent eddies which are moving things around.
And so the idea is that a particle might start out on some eddy, this blue eddy here, and start moving along it in this black trajectory. At some point that eddy decays and is replaced by this green eddy. So the particle has taken one step. Now it takes another step along a different eddy and then other eddies. And this continues on and on and on. And over the course of this random work, what you can show is the time it takes to move some given distance.
LS can be the time for step in the random walk and the ratio of how far you're looking to go. How big a step is it? And so if we put in all the quantities you just described to us, the system size, it's how long it's going to take for energy to take off in the middle to the edge. These the any size is our radius and the time for step is how long it takes a particle to move along.
Our Eddy, you come out with this estimate the confinement time is about a second right at the edge of what we need to make fusion devices work. If this were way in excess of a second, then we wouldn't be here right now. Fusion would have been working much sooner than it currently is. Instead, we're right close to where we need to be going through here. So one of the big headlines that came out last year was this record fusion energy yield shot taken from the Jet Tokamak.
Again, this is a column in the Centre for Fusion Energy just outside Oxford and it got 59 mega joules of energy over five or 6 seconds, which was sort of more than double the previous record also on Jet in a previous campaign. And if you look at the amount of energy that you get out of your plasma for this experiment. And compared to the amount of energy that actually hit the plasma, then what you find is, roughly speaking, that about half of the energy came out that you put in.
So it's a net loss of energy. I say roughly because there are different ways to measure this, but it's roughly half what came up when. And so we need to go a bit further. I can confinement time to be at least a few times bigger than it currently is. And so the question is, how are we going to do that? And there are different approaches. The first one is the one I've already mentioned to you. Right. Let's just make our device really big.
That way we can have small temperature gradients and still get the fusion images that we need. But this is not a uniform good idea, because, for instance, the bigger you make something, the more expensive it gets. Roughly, the cost tends to scale, like the volume of your plasma or the volume of your reactor.
So bigger is going to be worse economically. Also, there's some technological issues with doing this because imagine you have a certain amount of energy in your plasma and now I make the plasma volume bigger. The energy goes up like the volume, but the walls that surround it through which all this energy is not to be taken out at some point only go up like the area, not like the volume.
And so the heat hitting the wall per unit area goes up. And so at some point you have materials problems which are already on the edge of now, how are you going to take this heat out safely without damaging your device? But this is the approach roughly taken by etre. This is a schematic showing this ITER experiment which is being built currently in the south of France. And there's a person for comparison. So this is, of course, a very expensive device.
It's ballooned in cost to something like €25 billion or something. So this might demonstrate that fusion is going to work in a scientific point of view, but it's not going to demonstrate that it's going to work economically. I give. It is the biggest experiment ever built coming online in a few years time. This is an old graphic, so it's over 80% complete now. But the idea behind this is that we hope to get out ten times the amount of energy that we actually put into the plasma.
So this will be our scientific demonstration that you can make this work. Isn't the only approach to make things better. You can also take a technological sort of hack and try to make the magnetic field we're using stronger. As I told you, the stronger the magnetic field, the smaller the radius of these particles and gyration is, the smaller the eddies become, the better you can find guns. That's the idea. And so hope we can get away with a smaller device by doing this.
And so this is the approach that's being taken by a number of private companies these days. Commonwealth Fusion Systems in Cambridge, Massachusetts. Tokamak Energy just outside Oxford. The idea is they're going to use high temperature superconducting magnets. You're still creating this technology when you get ten times as much magnetic field in these devices than we're currently using. So that's the big selling point. It is still in development, this technology.
And you have a problem, which is that if you put such huge magnetic fields over such a small area, the stresses on all of your material components become enormous. So you have another sort of technological problem, and how do you actually keep everything from tearing itself apart? And finally, the approach which I guess all of the things this approach may be is to reduce the aspect ratio of your device.
So we started out most we've been showing you look what these sort of tokamak which look like this, they look more like doughnuts or bicycle tires. But one idea is to use what's called a spherical tokamak or spherical Taurus, which is more like a cold apple. And the benefit behind this is that, again, remember that the magnetic field strings, the toroidal magnetic field string drops off like the distance from the centre of this apple.
So if you bring your plasma in really close, then you get the magnetic field for free it simply by doing that. That's one benefit. The second benefit is that the rate of filtering is going to drop off really rapidly as you move across this volume because it goes like one over the distance. And so as you get very close to it, it drops off rapidly. So and these devices, the field is mostly toroidal on the inside, but it has quite a significant political component on the outside.
You might ask, why is that good? Well, if you remember, the plasma in here is stable and the plasma out here is unstable. And what this means is that plasma spends a long time where things are stable and very little time where it's unstable. So this is going to improve our confinement and our micro stability in these. There are issues with this as well. There's no free lunch. One of the big issues is what do we get?
If you have such a small space here in the middle of your device, how are you actually going to put like shielding to stop neutrons from kind of destroying sort of your central solenoid in this case? You also have the same problem with heat loads the inverse to the big reactor. So here, Matt, you say you take a different approach. You say I want a gigawatt coming out of this thing. I want a gigawatt power plant. And so that sets how much energy is coming out of your device.
If I now take my walls and I make them really, really, really small, which is great from the point of view, you still have a huge amount of energy coming out of a very small area. So there are also some technological challenges, this type of approach, and this is the approach largely being championed in the UK. So based that video I showed you earlier is one of the UK fusion experiments here. It's a vehicle for this step shown here, which is being funded by the UK Government.
Is a reactor like design using this vehicle tourists concept and also Tokamak Energy is pursuing these simple tours, right. So the UK is kind of championing this approach.
So I've given you some different ideas for how we're going to do this. What I'd like to say now in the very limited amount of time I have left, is that some of the things we're trying to work on as theorists is understanding ways in which we can improve the confinement time independent of any of these approaches that I've laid out for you. If we can suppress the turbulence, then that gives us a lot more leeway in the design of our fusion reactors.
And so here I'm going to give you one possible way that you can try to suppress turbulence in these devices showing this cartoon. So, again, I want you to imagine a magnetic field coming out of the board here. And I have my little particle which is gyrating in that magnetic field shown here by this black circle. And I have some turbulent eddies here, which, if you stay in one of these eddies, is going to take you across the plasma and makes hot and cold.
So as I mentioned, these particles gyrate with a radius roughly the size of one of these eddies. So any given particle is going to stay within its eddy and makes hot and cold. Now what I'm going to do is I'm going to say let's take a flow which is going up on the right and down on the left. So here's my shear flow. What does that do? It takes my eddies and it stretches and tilts them in a way that I've shown here.
So the now a particle which is gyrating around this field line actually samples multiple these eddies. So the first eddy, it might start to be taking heart outwards and then it starts to come inward and then outwards. And so again, as you average over many, many of these eddies, it effectively reduces the efficacy of these eddies to and. So this is one way that people are trying to suppress turbulence.
Now I know I said I'm mostly focussed on turbulence and that's what I've done, but I don't want to completely neglect other advances in the field. So I'm going to quickly show a couple of slides here and some other exciting developments. So this is one development that hit the news. Also, I came into this this year, last year, now using machine learning to help us shape our plasma and stop macroscopic instabilities. And so what you're seeing here are this this outer sort of shape.
Here is the vessel wall for an experiment in Switzerland TV. And this is the plasma that you're seeing inside that, doing some set of experiments. And what they've done is they've tried very hard to achieve certain plasma shapes. Turned out by shaping the plasma, you can change the properties of both the turbulence and the macroscopic stability.
But this is a difficult challenge. They come up with these shapes because if you have no plasma, then sure, I can design some magnetic fields which you can to map out whatever shape I'm trying to get. But as soon as you put the plasma in, it generates its own magnetic fields which interact with these ones, which change the plasma, which change these. And so it's a complicated feedback system.
And so what they've done is they've used machine learning in collaboration with DeepMind to have some target shapes they wanted, which are given by these blue circles. And they've used real time feedback, control this machine, learning to give it the shape they wanted over time in the device. So AI is not just good for making dark tiles, it's actually starting to do something useful for us. I don't want to say much about this. George is going to talk about celebrities in a minute.
But I do want to say that one of the big problems that we're still going to have overcome at some point in the future is how to make a device, a steady state and not inherently pulsed, as I've discussed before. And sort of the leading contender for doing this, I would say, are stellar ideas, which are ways of making these magnetic surfaces that don't have currents running through the plasma.
Okay. So this is my last slide. Basically it's showing some measure of progress, confusion over the years. So on this axis, it's a measure performance is this thing called the triple product. That's basically the pressure times, the confinement time versus year. And we're comparing progress in fusion with that in other areas. So you have Moore's Law shown here in red.
The energy of particle accelerator shown in green and Z fusion is actually very respectable on this plot blot because, you know, we might say we started out very low, but anyway, we've we've moved quite far. And I mean, the obvious elephant in the room here at this ends at 2000, but that was the last time we pushed things forward. Higher performance, really? And that's just because neither has been on the horizon forever. Right. And so it should put us up here.
It's going to be a pretty big leap. And so, obviously, you know, we're all waiting anxiously for this to happen. But one of the things that has really changed in recent years is that this is not the only thing on the horizon anymore. Right. We now have a range of private fusion companies who, whatever you think about them, are promising to give you results somewhere between these two on this line in a faster timescale.
And so the nice thing about this is, you know, putting all of your eggs in one basket. We actually have a lot of different ideas coming out at the same time, and I'm going to be testing them in real time. There's a lot of excitement in the program that now we're actually building all these different things, all these different approaches all at once, which you really need to do, I think, to test out, you know, what's going to happen next.
So I'm going to leave you with that and if you have any questions. Yeah, thanks.