 Imagine being able to engineer an artificial leaf that can harvest light to produce electricity or biofuel. To fully understand processes like this, we need to see what is going on at the super-small, super-fast scale of electrons. I will show you now how autoscience and science can help. All matter from molecules to subatomic particles is in constant motion. We know the state of a molecule before and after a reaction, but we don't know how and how fast the transition occurred. These transitions are dictated by fast electron movements, but we cannot see them. Microscopy has made enormous progress in the last century. We can now routinely image on the atomic scale. But the imaging challenge is not only being able to see small, in order to see motion, we need to see small and fast. At the speed of electrons, we can no longer talk in micro, nano, or picoseconds. We need a new unit now, the autosecond. So one autosecond is to one second, like one second is to the age of the universe. Being able to image at autosecond speed matters. Imaging at the nanosecond speed helped us to engineer modern computers. Imaging at the picosecond speed allowed us to develop new, ultra-strong materials. But increasingly, to make further progress, we need to see at the autosecond speed. The problem is that even our fastest cameras are too slow to directly observe atoms and electrons in motion. This is like trying to capture a photo of this maglev train. The problem is with a regular camera, we just see a blur. In my lab at Imperial College, we are taking a different approach. If we can't speed up the camera itself, what if we instead use a fast flash of light that illuminates our scene? And now rather than trying to film a continuous scene, we take individual snapshots from hundreds of reshoots. And afterwards, piece them back together to make a microscopic movie. We already know this can work. In 1872, Edward Mybridge used a combination of trip buyers and powder flashes to capture the motion of a galloping horse. Proving that all four legs to leave the ground. This freeze frame technique allowed us to visualize motion that was previously too fast to see. If we can prove that horses fly, with a few tweaks, we can also use this technique to observe ultra-fast changes in matter. Even chemical reactions in progress. The key is to produce a bright enough, fast enough light flash. And to do this, we need to use lasers. Lasers can bundle a large amount of power into a precisely controlled beam of light. In our lab, we had to engineer laser light to a degree never seen before. We compressed the entire visible spectrum of light and more into a short part lasting only 3 film per second. So in the very moment we fire our laser, it contains the power of 100 nuclear power plants. To see electrons, we have to use extremely short wavelength light. So we convert our powerful, visible laser pulses into X-rays. But above all, we have to be very fast. Through careful engineering, we can now produce X-ray pulses lasting less than 300 auto seconds. Auto-second laser pulses are the fastest man-made events in history. Our technology now means that we can start to explore some of the fastest events in nature. By combining static images captured using our ultra-fast camera, we can now use the freeze-frame technique of Mybridge to make movies of electrons. Using auto-second lasers, we've already discovered that electrons do not instantaneously escape from atoms. But instead they come out at different times, dependent on their initial state. This was a very surprising result, which forced refinement of theories dating back to Einstein. We're also starting to look at larger molecules, where electron dynamics are kicking off chemical reactions. We're also starting to observe electron dynamics in solid state matter. So who knows, perhaps more of our theories need to be refined in the future. The frontier of imaging is also the frontier of science in multiple disciplines. Understanding auto-second electron motion will lead to better understanding of chemical reactions and processes like photoemission. Ultimately, this fundamental understanding will lead to improved machines for light harvesting, better understanding of radiation damage, and might also have an impact on future optical computing.