 X, F, E, L, X-ray, free electron lasers, what's that? Yeah, something like that. It's just a little bit of a shorter wavelength and it doesn't definitely fit the size of a pen. Well, Thorsten Hellert is a physicist working at the Deutsche Elektronen Synchrotron in Hamburg. And he knows the answers to that. And he's going to introduce us to his world of free electron lasers and their applications. Warm welcome to Thorsten. Yeah, thank you. I have to admit I'm quite nervous not only because of this thing but also because my computer broke and I got this about one hour ago and I don't know if the presentation works on that as well, but let's hope for the best. Despite that, I'm very happy to see that so many of you are interested in particle accelerators and I have to say that this is not a one-directional relation. I talked to many colleagues at DZ and also in the US and all of them literally knew about this Congress and most of them even knew that it was going to happen in Leipzig this year. So I think I can say that at least every particle accelerator physicist I know likes the CCC as well and is interested in this Congress. Okay, but maybe enough small talk for now. Let's get to some science. So while you're watching this talk, your neurons firing incessantly, sending electrical impulses to the neighboring neurons. But how does this work? I mean, what are the neurons made of? Well, this artist's view from Harvard University lets us have a look inside. So inside each neuron, there's an enormous variety of proteins, big macromolecules, each consisting out of hundreds of thousands of atoms. Out of up to 40% of the volume of each cell is occupied by these proteins. And while the DNA serves as a blueprint, the proteins are manufactured somewhere inside the cell and then have to be transported to the destination where they are needed. For example, membrane proteins have to be transported to the outer shell of the cell, right? And this is done by the so-called vesicles like the blue fellow you see over there. So the proteins stick on them, and then motor proteins like this kinesine here pulls the vesicle along these long molecular strands which spawn through the cell, here, the green one. I don't know if you have seen such an animation before. When I saw this movie for the first time and when I realized about the enormous complexity of the molecular basis of life, this was literally breathtaking to me. But have you ever wondered how do we know about this? I mean, how do we know about the structure of this kinesine protein? And the answer is synchrotron light sources. So the vast majority of these proteins have been resolved in the latest third generation synchrotron light sources. And in this talk, I will show you what it takes to build such a machine and how to make a picture. But then the next question is, how do we know about the dynamics? So how do we know how these proteins fold? And to be honest, we have no fucking clue. So don't get fooled by the name Harvard University. This is just an artist's view and we have no idea how a protein folds. No one has ever seen something like this or witnessed a chemical reaction. But by the end of this talk, I will have shown you that just by now, we have a machine at hand, the X-ray free electron laser, which might be able to resolve these proteins within their natural time scale of a couple of femtoseconds. But okay, to start and bring everyone on the same page, I have to recall the electromagnetic spectrum. We are surrounded by a variety of electromagnetic waves which can be categorized according to their wavelength into different ways. For example, radio waves with a couple of meters or more in wavelengths, then we have the microwaves with some centimeters, and then the infrared and the visible light with a couple of hundred of nanometers in wavelengths. If we decrease the wavelength further, we will get to the ultraviolet and then finally at 0.1 nanometer or one angstrom, we have X-rays, okay? And there exists a fundamental limitation if you want to observe something with electromagnetic radiation, namely the diffraction limit. So it says basically that if you want to resolve two objects with a distance d, then you have to use a wavelength which is in the order of that distance or smaller. So if you want to study an ant or a bacteria, you can use visible light because visible light has a wavelength smaller than the size of these objects. But if we want to study viruses or the proteins we just saw or even smaller molecules, we have to use X-rays. But actually our ways of making a picture from something that small is quite different from what you are used to with your eyes or your camera. So we do X-ray diffraction images. And before I can show you how that works, I have to tell you something about coherence. So we start with an ordinary light source which emits light in different wavelengths, which is here indicated by the different colors. And the origin of these wave fronts is spread out, okay? So we have no fixed face relation at one point in space. And this we call incoherent light. That's the kind of light we are all surrounded by. You may know from high school physics that if I place an aperture in here, then the wave fronts propagate as they would be emitted by a point source in the whole of this aperture. And now we have a fixed face relation at one point in space and we call this spatially coherent light. And the next step towards coherence is then to put a filter in, which lets through only one particular wavelength, okay? So now this is coherent light. And if we consider ourselves to be very far away from the source, we can consider these ways to be plane waves. And then if I place something, for example, a double slit in here, I will get an interference pattern downstream. And on a screen, I would be able to detect a diffraction pattern. And the clue is now that mathematically, there exists a relation between the diffraction pattern and the physical arrangement of these objects. So I am able, if I know the diffraction pattern and I know the distance between the screen and the object, I'm able to calculate from the diffraction pattern the physical arrangement of these objects. In our case, we're doing x-ray diffraction. So we don't have double slits, but we have electrons on which the photons get scattered. And to give you an example, this is a microscopic image from a sample which was hit by x-ray laser pulse. And this is the diffraction pattern you record at the screen at your detector. So it's a bit more difficult than the previous example. But the key is, this is the reconstructed image. So from this, you are able to calculate this one. These two, although this is not very intuitive, are mathematically equivalent, okay? We can calculate this from the diffraction pattern without knowing the original sample. And these kind of x-ray diffraction images have been carried out for many decades. Just to give you one example, the discovery of the DNA structure was only possible because Rosalind Franklin made these diffraction images of a DNA crystal. And just a side story, guess who got the Nobel Prize for this? Of course, the two white men. But this is another nasty story I recommend you to look up afterwards. The thing is, about these x-ray tubes, they are very limited in brightness. And this becomes a problem if you want to study which moves, right? You all know if you want to make a picture from something which moves, you have to decrease the shutter speed. So for a running horse, for example, it's sufficient to have a shutter speed of one millisecond. But if you want to watch a bullet smashing a melon, you have to use like 1,000 frames per second more. And finally, if you want to go to chemical reactions, the shutter speed is orders beyond that. And you may have seen how such a movie is made. So you need big lamps in order to get enough light hitting your object in the very short amount of time while the shutter is open, right? So the figure of merit for a normal lamp is the luminous intensity, which is defined as the photons per time per solid angle. So basically the amount of light which are directed to your target. But we want to make x-ray diffraction images, so we need coherent light. And our figure of merit looks a bit different. It is called the brilliance of a light source. And what we want is basically, we want a lot of photons per time. We want them emitted in a small spot size with a small angular divergence and basically only one wavelength, okay? So this is the brilliance. This is our key. But before I want to show you what it takes for the brilliance to get from here to here, I want to sensitize you a bit more about the scales of what we are talking. So this is an example of some objects which I sorted according to their length on a logarithmic scale. So we start with a fingertip of some centimeters over a human hair down to molecules and atoms. And we are able to produce plenty of technology basically on the whole scale so we can produce a microgear with a diameter of some micrometer and even nanotubes. And although this is rather something academic yet, but in principle we are able to arrange matter on an atomic scale. The corresponding plot in time domain could look something like this, so we start with an eye blink of a couple of hundred microseconds down over the time it takes a shockwave to propagate by one atom in a crystal finally to chemical reactions or the Bohr period. And it takes a one gigahertz CPU about one nanosecond to make one computational step. And optical networks, which is even a bit faster, but basically we are not really able to produce technology on that time scale. I mean we are able to produce a laser pulse in the visible light which is as short as one femtosecond which is really amazing, but keep in mind the diffraction limit. So with this we can watch macroscopic objects like for example the microgear and we can watch the microgear within one femtosecond and see how it changes, but macroscopic objects don't change within femtoseconds. Things which change in femtoseconds are proteins or molecules and we are literally blind at these objects within their natural time scale. And to give you a better feeling of the scaling here, a fingertip is to an atom is about two times, two times 10 to the eight times bigger than an atom. And that's about the scaling of the walking distance from here to Tel Aviv to a fingertip. And in time domain an eye blink is to a chemical reaction like one year is to an eye blink. And now keep in mind when you go to a hospital and you want to make an X-ray image with a modern X-ray tube from a finger, you have to stand still for let's say a second, right? And if you want to scale that to an atom and to such a time scale, you immediately see that X-ray tubes are nowhere near what is needed for resolving proteins on their natural time scale. And I want to relate the development of our brilliance with something you knew, so you know. So this is the computer speed and you all know Moore's law and you have kind of a feeling what it means if a figure of merit increases in 12 orders of magnitude in six decades, right? The X-ray brilliance increased by 18 orders of magnitude in five decades. And this was not possible by small innovations, but very different steps. So we have different generations of synchrotron light sources. And finally, the fourth generation, we call X-ray free electron lasers. And in this talk, I will go through the steps what it takes to build these machines. But before I can tell you how we can build such a particle accelerator, I have to tell you why these particles actually radiate. And for doing that, I have to tell you something about relativity. Maybe you have been at Stiney's talk yesterday. I will try to summarize it on one slide. So we call our machines particle accelerators, right? But I guess your intuitive understanding of acceleration is an increase of velocity. But that's not really the case, but okay, step by step. So maybe you're familiar with Newton's law of kinematics telling you that the kinetic energy is one over two times the mass of a particle times the velocity squared. But as Einstein revealed, the speed of light is a constant which can't be exceeded by any particle with a finite mass. So it turned out that Newton's law of mechanics is only a borderline case for very low velocities for Einstein's more general equation of kinematics. And in here, you have this relativistic factor gamma. Gamma is one over this square root. And it basically relates the energy of a particle with its rest mass. And it's quite an important parameter for us. And it will pop up a couple of times in this talk. So let me give you an example. Let's assume that we accelerate an electron and a proton with five million volts, okay? So five mega volts. And then the kinetic energy of both particles is five mega electron volts. The rest mass for an electron is about 500 keV, kilo electron volts, while it is about 2,000 times more for a proton. And this means you can now plug the numbers in that the gamma factor is about 10 for electrons while it is about one for protons. And if you calculate the speed now from this, you will see that electrons, while being accelerated with five million volts, travel with 99.5% of the speed of light, while protons only travel at 10%. So electrons and protons, or in general light and heavy particles, have a very different relation between the energy and the velocity. And in our cases in synchrotron light sources, we are always interested in a very high gamma. So it's obvious that we are only using electrons, okay? So the next step is why do they radiate? Now this is an electron and here I plotted the electric field lines. And you may be familiar with a relativistic effect called length contraction or Lorentz contraction. A very neat example is a ruler which travels close to the speed of light and it gets compressed with respect to an observer and rest. And if we apply this length contraction to the electric field lines, we will see that while the speed of the particle is increased, the electric field lines are compressed into a very narrow cone, perpendicular to the speed of the particle. Okay? And now consider you want to change the velocity from here to here, so you accelerate the particle. And the electric field configuration has to change from that set up to this one. But this can not happen infinitely fast, but only with the speed of light. So you have a time-varying electric field and this is basically radiation. Maybe come a bit more clear on this slide. I made this simulation, you can download the simulator from the Shintake lab. This is again a point charge and I drag it now with the mouse and increase its velocity. And you can see as I increase the velocity, the field lines are compressed into this very narrow cone. And the radiation pattern gets more obvious if I change the direction of motion. For example, by running it on a circle. And if you think you sit here and watch the electron, you will get hit by narrow flashes of electromagnetic radiation. And this is basically a synchrotron light source. But I would like to look a bit more detailly on the radiation properties. So here again, this is our electron and I calculated the radiation pattern for this motion and I plotted the angular distribution here with this surface plot. So you see that most of the radiation is directed in the forward direction. And the opening angle here of this radiation cone scales with one over gamma and the overall power which is emitted scales with gamma to the four. So gamma is directly proportional to the energy. So if we have a very high energy, basically all the radiation is emitted into a very narrow cone in forward direction. And in our cases, gamma is something like 10,000. So it's really small. And a nice property of this radiation is that it covers a relatively wide range in frequency domain and you can easily tune it by changing the gamma or the energy of your particle. And this kind of radiation was first observed in a particle accelerator called synchrotron and that's why we call this synchrotron radiation. So coming back to that picture, synchrotron radiation is very suited to study small things like proteins or molecules. Now the question is how can we put this into technology? So how can we make use out of it? And of course, you know it's particle accelerators. So what are the principles of a light source? Well, first of all, we have to generate our electrons. So we need a device which serves as an electron source. Then we need something to increase the energy. And finally, we need a device to make them radiate. And with this radiation, we can then perform our x-ray experiments. It's as simple as that. And it's not a too ambitious analogy to think of such a light source as a radio station. There also you have your input signal, then you have a high power amplification and then you put this high power signal through a device which is designed to produce electromagnetic radiation of which only a tiny fraction hits your receiver. Okay, so in the following, I want to go through these different devices starting with the acceleration. You may know that if I connect a capacitor with a DC voltage source, I will get an electric field between the plates. And if I place a negatively charged electron in here, it will get accelerated, right? And we have these kind of accelerators. We call them fundagraph accelerators. And modern ones like this one is up to 10 meter long and reach, or can accelerate the particles by six million volts, which is not bad. But the problem is we can't really put them in series. And we can't increase the voltage either because we would simply get a discharge between the two plates. So the problem with this technology is it doesn't scale. So what we do is we replace our capacitor with an empty metallic resonator called cavity. And we connect this cavity with a wave guide to an AC voltage source. And this voltage source is operated usually in the radio frequency domain. So some gigahertz, that's why we call this RF. And the nice thing about such a resonator is that a relatively small RF field will start to resonate inside. So we will get a relatively high oscillating electric field. And we can easily put these in series. And if we put the face or set up the face relation between adjacent cells correctly, we will get an alternating oscillating electric field. And the real cool thing is now that we can put holes in here without really changing the geometry. And now the cells are coupled so we can remove all the power sources except of one. And if you put the beam pipe in here, an electron there. And if we synchronize everything correctly, you will see that we get an acceleration in each cell of the cavity. And of course, I mean, the devil is in the details, but this is the basic principle of an RF cavity. And there was no joke intended. And basically, every particle accelerator on Earth is operated with these kind of devices. Just to give you one example, this is a Tesla cavity. We have an hour linear accelerators at Daisy. Here we have these nine cells. And it's a superconducting technology. So everything has to be assembled in the clean room, which is very challenging. And then we put eight of them into one of these cryo vessels with a lot of support. And then we plug it in these yellow things here and put it down in the tunnel. Then we cool it down with liquid helium to two kelvin. And in these cavities, we can reach something like 30 million volts within one meter. So this is 50 times more what you can get in a fundagraph accelerator. And I mean, think of this 30 million volts between these two hands. To me, this is really an impressive technology. Seriously. OK. So the next step is the electron source. This is a movie made from the photo-injector test center in Sweden. But the electron source, we have at Daisy, look basically the same. So you see it's a very complicated device. And their whole laboratory is only building these electron sources. But this movie shows basic principles. So on the inside, you have a copper cavity, which with its connection to the wave guide. And from the inside, you have a photo cathode sitting here. And this photo cathode is impinged by a UV laser pulse. And while the UV laser hits this photo cathode, there are electrons emitted due to the photo effect. So each of these red things are about 1 billion or 10 billion electrons. And we call this an electron bunch. And then again, we have two cells of a RF cavity. And everything is synchronized in a way that accelerates the electrons immediately as they are generated. OK. Finally, we need a device to make them radiate. And I already told you, we just have to bend them around the circle. And we can do this most easily in dipole magnets. You may know from high school physics or whatever the left-hand rule, if we have an electron with speed v and the magnetic field perpendicular to it, we'll get the Lorentz force in the third direction. So the whole thing is bent around a circle. So now we have everything together to build our storage ring. We have an electron source. We need an RF cavity and then a big dipole magnet. So the particle will move from the circle and continuously emit synchrotron radiation. But it's not that easy because we have energy conservation. And while the particle emits power, it will lose kinetic energy. So it will finally spiral in and get lost. So we have to replace that and insert straight sections where we can place an RF cavity to compensate for the power loss in the dipole magnets. And then we have to put some focusing elements in here. We use quadrupole magnets to stabilize this system. And this particle accelerator is called synchrotron. And originally, these kind of devices were built for high-energy physics applications, like for example LHC, the Large Hadron Collider in CERN. There's nothing more but this, of course. But the basic principle is synchrotron. And this could be your Atlas detector. And in the early 50s, when they started to build these kind of accelerators, the synchrotron radiation was found to be nothing more but a nuisance, which made everything more complicated. But in the 60s, x-ray diffraction became a thing. And scientists started to realize about the capabilities of this radiation. So they placed some x-ray optics in here, which guided the synchrotron radiation to the experiments. And these kind of devices are considered as the first generation synchrotron light sources. And as an example, this is the Tantalus 1 accelerator in the late 60s. Here is the accelerator. So this is the RF cavity. And there are some dipole magnets. You see it's a fairly small device. Very soon, scientists started to want to have more power in their radiation. So in a bending magnet, each electron radiates. So the intensity or the brilliance scales with the number of electrons. Double the electrons, double the power. And starting from that, if you want to increase the power, the first obvious step is to put more dipole magnets in. So this is an insertion device called Wiggler. And it's basically nothing else but a series of dipole magnets with alternating polarity. So the electrons will move on a slalom trajectory. And in each curve, you will get synchrotron radiation as from a single dipole magnet. And by doing that, you will increase the brilliance by a factor of the number of magnets. So nothing more but that. Then the next generation or the next step towards brighter synchrotron light sources was the invention of an undulator. And an undulator is a very similar device than a Wiggler. The only difference is that now the bending radius is so small that the radiation cone basically always points in the direction of the experiment. And the mathematical details of this radiation are a bit complicated. But the idea is that now you have interference of the light emitted in each of these curves. And by doing this, you compress the overall power here from a Wiggler into very narrow spikes and frequency domain. And this is great because remember, we want to make X-ray diffraction images. So we need coherent light. So we need only one wavelength anyway. So we put a filter in somewhere. If you place a filter at that frequency, we'll gain a huge amount of brilliance. And these kind of devices are considered as third generation synchrotrons. So facilities which are dedicatedly built to produce as much synchrotron radiation as possible with many beam lines and many experiments. And as you can see here, there are plenty of them operated in the industrialized countries around the world right now. And as an example, I want to show you the Petra 3 accelerator we have at Daisy in Hamburg. But let me drink something. OK. So this is the Daisy campus. And here, this ring is Petra 3. It has a circumference of about 2.3 kilometers. So it's a fairly large device, including this 300-meter-long experimental hall, of which the schematic sketch you can see here. And each of these lines is an X-ray beam line with dedicated experiments. From the inside, it looks like this. So you can't really see the accelerator because everything has to be shielded with these concrete walls because of the radiation. But the accelerator is here on the inner ring. This is a picture from the inside. And here you have the beam lines with the experimental chambers at the end. OK. As I said, this is a picture from the inside. So these are the quadrupole magnets. And we have some steering magnets. And the yellow devices here, these are the undulators, which produce the radiation. And beam line at these facilities is very expensive. So most of the beam lines have to be optimized. For example, at this one here, we have a robot arm which takes the crystal samples out of the viewer here and then mounts it on the sample holder. And the accuracy here is very impressive. I mean, we have X-ray sample crystals as small as 100 nanometer. And then they are rotated around the axis inside a photon beam, which is as small as 100 nanometer as well. But why do we need crystals at all? The reason for this is that the cross section between our X-rays and matter is very low. So statistically, we need 1 million atoms in a row to get one single diffracted photon. And you can imagine we need much more than one single photon to get an image on our detector, which we can calculate something from. So what we do is we have to increase the amount of photons. But this is limited by some constraints of our particle accelerators. So we have to increase the amount of atoms in our sample. And we do this by growing crystals. So this is a protein. And we have to find proteins which we can form unit cells out of and then grow a crystal. So we need many of these. And then we can put the crystal in our X-ray beam and get some diffraction spots. And while turning the crystal around its axis, we'll get a 3D diffraction pattern. And from this, we can then calculate the 3D electron density map of our sample. And if we know the electron density map, we know the structure. Here you can see the cumulative number of structures which are available in the protein database. And you can see that within the last 20 years, there's been an astonishing increase, most importantly made possible by X-ray diffraction images in these modern third generation synchrotron light sources. And right now, we are not only able to make pictures of small proteins like the myoglobin, but even very big ones like the ribosome. But this is not really by far not trivial. So for example, the ribosome, the first X-ray diffraction pattern of ribosome was made in 1980. But it took scientists 20 years to calculate the structure from this. And although this number here seems very high today, less than 2% of the human proton is known. So 89%, 98% of the proteins of our body are unknown. And the reason for this, the bottleneck is the crystal growth. So it's very hard to get most of the proteins to form big crystals. Some of them are even not, it's just not possible to crystallize them at all. For example, membrane proteins. But for others, it's very difficult to grow large crystals. So what we ideally want is to be able to make a picture from a very small crystal or even a single molecule. But in order to do that, we have to increase the amount of photons by something like 100 million. This is really not easy. But let's consider for now that we would be able to make a storage ring bright enough. So 100 million times brighter than it is and to make a diffraction image of a single lysosyn. What would happen? Well, this. This is a simulation published a couple of years ago. And what you see is the Coulomb explosion of lysosyn. So as the x-ray beam hits the sample, it immediately blows away all the electrons in the molecule. And what is left are the positively charged nuclei which repel each other. So the whole molecule blows apart. And the problem is now that because of fundamental particle beam dynamics, it's not possible to make an x-ray storage ring pulse smaller or shorter than about a picosecond. So even if we would be able to make a storage ring pulse bright enough to watch a single molecule, we would just be able to see a blurry picture of an explosion. And this is where free electron laser came into play. Because in a linear accelerator, it is fundamentally possible to produce an x-ray pulse as short as some femtosecond. But as I told you, we have to put 100 million times more photons in that short pulse. So this is not easy. And what we do is, well, first of all, let me rescale this picture. We replace the undulator by a very long undulator. And by doing that, now comes the point. Because if we set up everything correctly, we will get on top of this radiation pattern from a long undulator, we will get narrow spikes of coherent radiation. And this is what makes a free electron laser so important. So mathematically, the radiation scales now with a number of electrons squared. And the number of electrons in our bunches is something like 100 million or a billion. So this is really a huge number. But let's have a look inside what's happening in the undulator. So this is an electron bunch. And all the red circles are supposed to be electrons. And the whole bunch moves in the undulator. And there exists a resonance condition between the undulator period and the period of the emitted light. So here you have the undulator period, the emitted light, the relativistic gamma factor, and this k-value, which cooperates some information about the magnetic field. But it's not important for now. So I'm only looking at the wavelength of the emitted light, which satisfies this condition. Now let's have a look. So this is the electromagnetic wave, which is emitted by that electron. And the whole bunch is moving up and down in that picture. So some of the electrons move in the direction of the electric field. Sorry, this is the electric field line, which I plotted here. So some of the electrons move in the direction of the electric field, while some of them move in the opposite direction. So some of them will gain transverse momentum, while others will lose it. And if we hit this resonance condition, both the direction of the motion of the electrons and the electromagnetic waves flip sign at the same time. So this process repeats itself. And while all of this is happening, we are in a magnetic chicane, meaning that there exists dispersion. And dispersion means that the bending radius depends on the energy. So if you have a larger energy, then the bending radius is bigger. If you have a smaller energy, the bending radius is smaller. So some of these particles have larger transverse momentum, so larger transverse energy, so to say. And they will move, they will fall back, and others will overtake the bunch. So we have a self-ordering effect, which repeats itself. Now coming back to the big picture. So from at the beginning, we start with an incoherent radiation. So all the electrons, while they are bent around the circle, radiate. But there is no fixed phase relation between them. So this is incoherent radiation. And the intensity of such kind of radiation scales with the number of emitters in this example, number of electrons. And now as the bunch moves through the undulator, the self-ordering effect leads to a micro bunching on exactly that length scale of that radiation. So for the wavelength satisfying this condition, we will get a coherent radiation and coherent radiation scales with the number of electrons squared. But it's not easy to get from incoherent to coherent radiation, especially if you want to have x-rays here. Sorry. So what we need is a small beam. This is just to give you an idea of the order. So I don't take these values too serious. There can be a factor two or three between them. But we need a small beam, something like 10 micrometer in transverse size. We have to have it as short as 10 micrometer. And we have to get it on the high energy, something like 10 billion electron walls. And we need a very long undulator, some 100 meters. And within that undulator, we have to align the electrons to better than 10 micrometer in order to have an overlap between the electrons and their light. So this is really challenging. This is a sketch of such a free electron laser. So usually, we have different acceleration stages. And in between, we have magnetic chicanes. We call them bunch compressors. And in them, we are able to produce the very short bunches. And we have a long undulator. And finally, we dump the electrons. And our light gets to our experiments. As you can see here, there are just now five of them in operation. And at least five of them operating in the hard x-ray regime. And just as an example, I would like to show you the European XFL, which is the largest free electron laser we have on Earth. This is a map from Hamburg. You can see with its overall length of about three kilometer, it reaches out the Dezicampus and reaches the adjacent federal state of Schleswig-Holstein where the experimental hall is built. You can't see much from above because everything is beneath the Earth. I would like to show you that movie which was made while the accelerator was still under construction. So right now, it would not be possible to walk down there. You would just die. But there it was possible. And I think it was really impressive to be down there and see all this high tech next to you. And it just never stops. But anyway, you can see this is now the main accelerator. It goes on for another kilometer. You see where we are. And this goes on for two minutes. I think it's a bit boring. But you can watch this movie if you want at home. I think I doubled the speed anyway. But I want to give you some numbers. So in average, we take about 9.5 megawatts from the grid. This is about the energy consumption of a small city. And from that, due to the utilization of a superconducting RF technology, we are able to put 10% into our beam. So we have an average beam power of 900 kilowatts, which is really impressive for a linear accelerator. And from this, we get 0.1% into our x-ray beam. But finally, only less than 1% hitting or get in the diffraction spots. So you could argue that the overall efficiency of this machine is terrible. And I would agree. And also, 900 watts of x-ray beam power seems not very impressive. But what makes this machine worth a billion euro is its ability to compress that power into very narrow spikes. So what is interesting is the peak power. In average, we have a repetition rate of 27 kilohertz. So 27,000 x-ray pulses per second are produced with a wavelength of about 0.5 angstrom, pulse energy of 1 millijoule, and a pulse duration down to 3 femtosecond. I mean, this is the time it takes light to travel 1 micrometer. This is really short. And we can focus this x-ray beam down to a very narrow spot. And in this spot, in the focus point, we have a power density of about 10 to 17 watts per square centimeter. I guess you don't know what 10 to the 17 watts per square centimeter is. But let me give you an example. This is about the power density, as if you would take the total sunlight hitting the earth on 1 square centimeter. So this is really intense. And you have to be careful, because if you accidentally hit something, another thing I would like to show you is that it's really not easy to build or to operate such a machine. Like just for the European XFAL, we have 9 million control system variables. This is a picture I made from the control room at Daisy. And you see that there are a lot of screens. And you have access to all of them. So it's really not easy to design a control system, which can be operated by many people and give you access to all of these. And I made an animation or screen recording, because once I had a measurement shift at flash, which is another free electron laser we have at Daisy. And I had to dig out a toroid signal, which was not on the top layer of the control system. It took me quite some time to find it. So this is the top panel of the control system. And as you can see, as you press some of these buttons, they will open up a panel with a lot of other buttons. And if you press one of these buttons, another panel opens. And please note the sub-panels over here and here. But finally, so really, we need a lot of experts working together because no one is able to keep all of that in mind. Another interesting number I find is the data production rate. So now I'm not talking about the machine. I'm just talking about the x-ray detector. And in there, we have about one megapixel with a resolution of 16-bit. And we want to record this 27,000 times per second. And this means we have 60 gigabytes per second. Just to give you a number, the LHC, after filtering, has about 600 megabytes per second. So you can imagine that we also need very sophisticated trigger levels in order to deal with this amount of data. Because no one is able to record or manage 60 gigabytes per second. And as an example, this is the amount of stored data in the first weeks of operation of the European XFAL. So you see, we are hundreds of terabytes. And keep in mind, within that period, the machine was working with less than 10% of its full capacity. So we are talking about petabytes here. So this is also not that easy to control. But finally, I would like to close this talk with a unique application which you can only do at these free electron lasers. And it's about molecular movies. So for example, this iron complex in et cetera nitrile solution, if you hit it with a UV laser or UV light in general, then it will perform a chemical reaction and will lead to an acid legend and the bending of such a solvent molecule. This is chemistry. We know this for many decades. But the problem is that basically all of our knowledge of chemistry is equilibrium science. So we know the reactants and we know the reaction products. But we don't know what's happening in between. And usually, there's not only one reaction path, but there are many with different probabilities. And you can imagine if we don't know anything in between, it's very hard for us to design a drug or a catalyst or something like this. This is basically nothing more than applied science alchemy. I mean, we just try an error. So we would really benefit from knowing what's happening in between. And with the XFEL, we can do this. This is a picture of the experimental hall in Schoenafeld. Here we have these five beam lines. And now we are watching one of them. So here come our x-ray beams. This is a photon diagnostic section where you can analyze the properties of our x-ray beams. And here, finally, we have the target. This is a liquid jet target. And it's not easy to design because we want a single molecule to get hit by our x-ray beam. We don't want to have two. And we don't want to have zero. And all of this has to happen in vacuum. And it's really not trivial to build these kind of experimental chambers. Now, how can we get to a molecular movie from this? So first of all, we have to be able to trigger our reaction. And we can do this with a UV laser pulse. So we hit our molecules with the UV laser, and the reaction starts. And then we can make a snapshot with our x-ray laser. And by setting up the delay time between the UV and the x-ray laser, we can make a snapshot from different stages of this reaction. And that's basically everything. But also, the readout of this detector is very sophisticated. So between the different layers, because between each pulse, there's only 200 nanoseconds. And then the detector has to be ready for the next picture. So it's really not trivial to build these. And this is basically the most powerful x-ray detector we have on Earth. But finally, we get our images. And from each image, we can calculate the structure of our molecule. And by putting them all together, we are able to make a molecular movie from a chemical reaction. You see what it takes to make something like this. And I guess you understand that it's a long way to get to something like this, right? But in principle, I think I have shown you not only how we are able to resolve the structures of these proteins, but also as how free electron lasers may enable us in a couple of years, maybe decades, to watch these kind of movies. But not as artist use, but as real experimental data. So thank you very much. And if there are any questions? Thorsten, thank you very much for this highly educational talk. If anything goes wrong with your postdoc in Berkeley, I recommend you go to science communication. OK, we already have a question from the internet, I have heard. Yeah, there's actually one question from Goyken. How well do the experiments replicate? I've seen the talk yesterday as well. And I think you mean, in general, the X-ray experiments or from the European XFL? It's on the internet, right? OK, I would say they replicate quite well. There are experiments made at different X-ray sources. And from time to time, they try to cross check at other X-ray sources or try to make the experiment a bit different. And I think this is kind of replicating it, right? But I'm not a photon expert, so I built the machine. I don't really care about the images. So I'm sorry. OK, microphone one, please. OK, hi. Yeah, really amazing talk. I have to also admit that. What's the current status of the X-FEL? So because you showed now at the end just this procedure how you would do a movie, how far are we actually to do that for a simple example? Something like a year, maybe? I mean, it really depends. I didn't tell you how difficult it is to make how many pictures you have to combine to make such a movie. So you have to combine several hundreds of thousands of X-ray images or diffraction images to make such a movie. So you need a lot of beam time. And especially right now, I think it's more difficult to prepare the samples and to get to full capacity because of some issues of the accelerator. I would guess something like one year to get to some. In general, the machine is ready and operational. It operates right now. And could start doing that. OK, thanks. It's just not that all subsystems are working. Like some of the experimental chambers are not ready or some beam properties can't be hit right now. OK, microphone number four, please. OK. How do you stop the molecules degrading by when they are hit by the free electron laser? Sorry again, please. So you showed before that if you don't have a crystal of molecules, it degrades instantly. And how do we stop it with the free electron laser? You mean how do we stop the molecule from exploding? Yes. Ah, we don't. Oh, OK. It gets obliterated in each shot. So that's why we have to make 100,000 pictures. Because after each, maybe let me show you this maybe. So each shot, this is our molecule and gets hit by this laser. And each shot gets destroyed. And it's more difficult because the orientation of the sample is random in each shot. So we need very sophisticated software to calculate this 3D diffraction image from this to finally resolve the structure. This is much more difficult than if you have a crystal because there you know your orientation and you can rotate it in a defined way. But finally, each shot is we need to get the data from one shot. OK, microphone number one, please. So this is more of a technicality. How is the power on the electron beam dump and what are you using for the electron beam dump to get the amount of bremsstrahlung emitted to acceptable levels that you don't destroy everything with that? Yeah, that's basically the limitation of this 900 kilowatts. It's the specification what we get from the Strahlenschutz Behörde to operate these machines. We use big blocks of, what is it, graphene, I think, and kind of rotating magnet such that the beam doesn't hit the same spot every time. But it's basically just a big, big block, very long. Like, how long is it? Maybe eight meters, like this big. And then we have several of them, which can be changed. And then they have to put away for some decades to cool it on. But it's safe. Microphone four, please. First, thank you again for this really amazing talk. This is a very greedy question. But is it anticipated that the growth in the ability of these will continue to go beyond what free electron lasers have achieved? And is there a glimpse into what the fifth generation of synchrotrons would be? Yeah. I asked a couple of guys in the scope of preparing this talk. And depending on where they are, they answer different things. So some of them answer no. It will be different techniques. So free electron lasers have the unique ability to make very short pulses. And this may become even better. So less than one femtosecond. But there are other tools like electron diffraction or also electron microscopy, which are maybe suited better for different samples. But actually, I don't know what's really the next step in synchrotron radiation sources. Thank you. OK, let's be fair to the internet. Is there any question? Yeah, we have some more questions. Bucking, is it all right? Bucking sheep is asking, how long does it take to run an experiment? As in writing the spec to the experiment, sending the beam, collecting all the images, and producing a picture? A beam time is something like so on flash or other free electron laser. The typical beam time slot is eight hours. And so the machine runs 24-7. But some experiments take eight, some 16, some two days. But that's the order. So let's say 10 hours. And setting up the experiment is actually the bottleneck. So this can take up to one week. So I don't, unfortunately, I don't have a picture from the experimental hall at flash. But we have different beam lines. And there are 10 people working there to build up the experiment for one week. And then they have eight hours of x-ray beam. And then they work half a year on reading the data and combining these images. So the beam time they are making the images is the smallest part. OK, microphone one, please. Yes, thank you for the great talk as well. My question is, I'm sure you are aware of these protein folding software projects, which try to make these images by calculation. How well do these work? And how much do you benefit from these approaches? I mean, that's the point. We don't know how well they work. I mean, we have these simulations. You can find them on YouTube. And they are nice, but no one knows. Thank you. All right, another microphone one, please. Yeah, and this was an amazing talk. Can I talk a little bit more about how to focus the x-ray pulse? Yes, but I don't know if I can answer your question. I should talk more. I think for discussions, if you want to have discussions, I think we can do that probably outside. So internet question. Unrestricted Eve would like to know if you could tell us some more details about how the x-ray camera manages to hold so many data in such a short period of time. OK, to the internet question, no, I can't really. I wanted to ask the guy who designed the detector or was the responsible for designing the detector, but he was in holiday already in the last week before Christmas. So I couldn't really get an answer to this question. I don't know it exactly. I just know that there are several layers. And it's not. I would talk bullshit, I think. But I guess very soon they wanted to write a big comprehensive, some comprehensive stuff about the x-ray detector on their homepage of the European XFL. So I would recommend you to look it up there. But to come to your question, we do this with basically diamonds or some diamond-like crystals. This is an x-ray mirror we have. And we have a gracing incident angle. So that's how we focus these beams. And it was in the news. The flatness of this mirror is really amazing. But I don't have the numbers right now. But look it up, it's crazy. And again, microphone one, yeah. This is, of course, an amazing piece of hardware. But as you mentioned, you showed us the control software. It's also an amazing piece of software and amount of software. Can you give us some numbers on number of programs, lines of code, many years, whatever, because you spend a billion in hardware. But software is also probably a lot. Yes, yes, yes. Yes, for sure. That would be an interesting number. No, I don't have the number of lines involved in this code. I know that the amount of CPU power we need is not that much. So it's the most difficult thing is to get all these channels appear on our control system. So the graphical interface is more challenging than work with the data. But I really don't know how many. No, I can't really tell you this. But if you write me at the end of the slide, I have my email address. I could ask some guys at easy. OK, microphone two, please. I also have a question about the control software. Do you have a query language to find the controls you need instead of having to step through all those windows? Yes, of course, of course. But usually when you have no clue of what you're looking, it's sometimes easier if you have a GUI where it's at least sorted. But of course, you can have access read and write also by just writing lines in your. Internet questions? No more questions. OK, microphone one, please. Yeah, my question is, is there any policy in place for publishing stuff, like only open access or something like that? At Daisy? My re-sector coming to your facility, applying for beam time, do I have some policy to fulfill to only? Yeah, you have to publish. I mean, you have to publish in an. Is it open access? That's the question. Yeah, that's a good point. I think it doesn't have to be. So you have to make sure that your results are published. But since it's not a good point, I know that a private company can also come and ask for beam time, but they have to pay a lot of money to get that. But if you are a scientific researcher or a university or whatever, you get it for free. Thanks. Thanks.