WEBVTT
00:00:00.000 --> 00:00:19.579
36c3 preroll music
00:00:19.579 --> 00:00:23.100
Herald: So who is excited about
photography or videography?
00:00:23.100 --> 00:00:28.320
applaus
Herald: Yeah? The title of the talk kind
00:00:28.320 --> 00:00:35.790
of gives us gives it away. OK. We bet we
are waiting for the last people to come in
00:00:35.790 --> 00:00:44.160
and take a seat. Last time, raise your
hands if you have a free seat next to you.
00:00:44.160 --> 00:00:55.289
Every one of you coming in, look for raised hands
and take your seat and then we will start.
00:00:55.289 --> 00:01:19.390
Yeah, very good. OK. Looks like the doors
are finally closed. Okay, so the next talk
00:01:19.390 --> 00:01:26.899
on the second day is about ultrafast
imaging. So many of you have done
00:01:26.899 --> 00:01:34.149
videography or photography. Have thought
about exposure time, how fast you can do
00:01:34.149 --> 00:01:39.869
your photography. And some of your might
have played with lasers and have built
00:01:39.869 --> 00:01:45.540
blinky stuff with it or have done
scientific experiments and Caroline Will
00:01:45.540 --> 00:01:51.140
now show us what happens if we take those
to combine them and take it to the
00:01:51.140 --> 00:01:58.849
extreme. Caroline is working at DESY since
four years. She has not done her PhD and
00:01:58.849 --> 00:02:05.619
is now working in a group for theoretical
fast modeling of inner workings of
00:02:05.619 --> 00:02:11.280
molecules and atoms. She is doing a
computational work and working together
00:02:11.280 --> 00:02:16.709
with experimentalists to verify their
observations, and now she is presenting
00:02:16.709 --> 00:02:22.310
the inner mechanics of what she is doing
and how we can actually maybe photograph
00:02:22.310 --> 00:02:27.720
molecules by their forming. Applause!
00:02:27.720 --> 00:02:31.160
applause
00:02:31.160 --> 00:02:35.189
Caroline: Great. Yeah. Thank you very much
for the introduction and thank you very
00:02:35.189 --> 00:02:40.170
much for having me here. I'm excited to
see this room so full. So I'm going to
00:02:40.170 --> 00:02:44.680
speak today about an ultrashort history of
ultrafast imaging. It's a really broad
00:02:44.680 --> 00:02:48.469
topic. And I'm just gonna present some
highlights, some background. Before I
00:02:48.469 --> 00:02:53.860
start, I'd like to give you a few more few
more words about myself. As we've already
00:02:53.860 --> 00:02:57.620
heard, I work at DESY, this is the DESY
campus you see here and in the Center for
00:02:57.620 --> 00:03:03.280
Free Electron Laser Science, circle in
orange. That's where I did my PhD. So this
00:03:03.280 --> 00:03:08.530
whole campus is located in Hamburg. This
is probably also a familiar place to many
00:03:08.530 --> 00:03:15.260
of you. And now this year we are in
Leipzig a bit further away for the 36th
00:03:15.260 --> 00:03:21.299
Congress. So I'd like to start with a very
broad question. What is the goal of
00:03:21.299 --> 00:03:25.519
ultrafast imaging? And we've heard already
that ultrafast imaging is related to
00:03:25.519 --> 00:03:31.550
photography. Now, as many of you know,
when you take a picture, with a quite long
00:03:31.550 --> 00:03:36.389
exposure time, you see just a blurry
image, for example, in this picture of a
00:03:36.389 --> 00:03:42.510
bowl of water. We can hardly see anything.
It looks a bit foggy. But if we choose the
00:03:42.510 --> 00:03:47.019
correct exposure time, which in this case
is 100 times shorter in the right picture
00:03:47.019 --> 00:03:51.250
than in the left picture, then we see a
clear image and we can see dynamics
00:03:51.250 --> 00:03:55.959
unfold. So we have here, a drop of water
that is bouncing back from the bowl and
00:03:55.959 --> 00:04:00.470
also some ripples that are forming on the
surface of this bowl. This is only visible
00:04:00.470 --> 00:04:06.329
because we chose the right exposure time.
And this is to me really the key of being
00:04:06.329 --> 00:04:13.230
successful in ultrafast imaging to take a
clear picture of an object that is moving.
00:04:13.230 --> 00:04:17.130
But it's not enough to say take just a
picture. So now imagine you're a sports
00:04:17.130 --> 00:04:20.680
reporter. You get these two pictures and
you're supposed to write up what happened.
00:04:20.680 --> 00:04:25.660
So it's complicated. So the top picture is
the start, the bottom pictures is the end.
00:04:25.660 --> 00:04:30.500
Just from these two pictures, it's hard to
see. But if we see before picture, we can
00:04:30.500 --> 00:04:35.169
see very complex dynamics unfold. There
are particles accelerating at high
00:04:35.169 --> 00:04:41.090
velocity laughing coming in from the back. And even
particles we did not see in the first
00:04:41.090 --> 00:04:46.330
picture at all somehow are very relevant
to our motion. And not only skiing races
00:04:46.330 --> 00:04:52.270
are very dynamic, but most processes in
nature are also not static. This is true
00:04:52.270 --> 00:04:55.889
for everything we see around ourselves,
but it's especially true for everything
00:04:55.889 --> 00:05:01.370
that is quite small in the microcosm. And
in general, we can gain a lot more insight
00:05:01.370 --> 00:05:08.360
from time resolved images. So from ultra
short movies. I'd like to show you the
00:05:08.360 --> 00:05:12.270
very first ultrafast movie that was ever
taken. Or maybe even the first movie that
00:05:12.270 --> 00:05:19.960
was taken at all. This guy, Eadweard
Muybridge lived in the 19th century. And
00:05:19.960 --> 00:05:24.770
very shortly after the invention of a
photography method, he tried to answer the
00:05:24.770 --> 00:05:29.900
question does a galloping horse ever lift
all of its feet off the ground? Why, it's
00:05:29.900 --> 00:05:34.460
running. To us, it may seem like not so
important question, but in the 19th
00:05:34.460 --> 00:05:39.560
century, the horse was the main method of
transportation, and horse races were very
00:05:39.560 --> 00:05:45.490
popular. So there was a lot of interest in
studying the dynamics of a horse, and this
00:05:45.490 --> 00:05:50.259
process is too fast to see with the naked
eye. But Muybridge implemented a stop
00:05:50.259 --> 00:05:55.040
motion technique where the horse as it is
running, cuts some wires, that then
00:05:55.040 --> 00:06:00.129
trigger photographs. And with this he was
able to take these twelve photographs of
00:06:00.129 --> 00:06:05.220
the horse in motion. That was published
under this title in Stanford in the 19th
00:06:05.220 --> 00:06:10.080
century. And we see very clearly in the
top row third picture and maybe also
00:06:10.080 --> 00:06:14.330
second picture that indeed the horse lifts
all of its legs off the ground, which was
00:06:14.330 --> 00:06:20.170
a new insight at that time. And when we
stitch all of these snapshots together, we
00:06:20.170 --> 00:06:24.860
have an ultra fast movie of a horse
galloping, which might be seen as the
00:06:24.860 --> 00:06:31.259
first movie that was ever made in the
history of mankind. Now, when I say
00:06:31.259 --> 00:06:35.550
ultrafast today, I'm no longer thinking
about horses, but about smaller things and
00:06:35.550 --> 00:06:40.570
faster things. But let's go there, very
gently. So the time scale that we are all
00:06:40.570 --> 00:06:44.909
familiar with that we can see with the
naked eye is something of the order of
00:06:44.909 --> 00:06:50.150
seconds. So, for example, the acceleration
of this cheetah, we can see with the naked
00:06:50.150 --> 00:06:56.349
eye. Now, if we zoom in on this motion, we
see that there are muscles inside of the
00:06:56.349 --> 00:07:00.930
animal that are contracting as it is
running. And this muscle contraction takes
00:07:00.930 --> 00:07:06.810
place within milliseconds. So that's a
part of a thousand in one second. But we
00:07:06.810 --> 00:07:11.319
can go even smaller than that to the
microsecond. So proteins inside of the
00:07:11.319 --> 00:07:17.180
muscles or in any biologic matter fold and
unfold on a timescale of microseconds.
00:07:17.180 --> 00:07:22.509
That's already a part in a million of a
second. Now going even smaller, to
00:07:22.509 --> 00:07:29.080
nanoseconds there's certain dynamics that
take place within these proteins, for
00:07:29.080 --> 00:07:34.091
example, of how they dissolve in water.
But the timescale that I'm interested in
00:07:34.091 --> 00:07:38.969
today is the femtosecond. It's even faster
than that it's the timescale where
00:07:38.969 --> 00:07:44.539
individual atoms move in molecules as
shown in this animation. Now a
00:07:44.539 --> 00:07:49.199
femtosecond is very short. It's a part in
a million of a billion of a second, or as
00:07:49.199 --> 00:07:55.039
we physicists like to call it, ten to the
minus 15 seconds because it's easier to
00:07:55.039 --> 00:08:01.689
spell laughing to us. We can - to us - , we can go even
faster than that. The time scale of
00:08:01.689 --> 00:08:05.389
electronic motion and in molecules would
be an attosecond. I'm just mentioning it
00:08:05.389 --> 00:08:12.340
here because we don't stop at molecules,
but nature is even faster than that. But
00:08:12.340 --> 00:08:15.030
for the purpose of this talk, I will
mainly focus on processes that take place
00:08:15.030 --> 00:08:21.741
within the femtosecond. So within ten to
the minus fifteen seconds. Now, this time
00:08:21.741 --> 00:08:26.410
scale is something that is not really
related to what we think about in everyday
00:08:26.410 --> 00:08:31.150
life. But there are certain processes in
chemistry, biology and physics that are
00:08:31.150 --> 00:08:36.419
really fundamental and that start at this
time scale. Just to give you an idea how
00:08:36.419 --> 00:08:42.070
short a femtosecond is, the width of a
human hair is about 100 micrometer. It's
00:08:42.070 --> 00:08:47.130
shown here in an electron microscopic
picture. And for light at the speed of
00:08:47.130 --> 00:08:53.750
light, it takes only thirty femtoseconds
to cross the hair. So that's how fast a
00:08:53.750 --> 00:08:58.950
femtosecond is. And even although this
timescale is so short, there are many
00:08:58.950 --> 00:09:03.730
important processes that start here, I'd
like to mention, just two of them. The
00:09:03.730 --> 00:09:07.630
first one is vision in our eyes and our
retina there sits a molecule called
00:09:07.630 --> 00:09:13.660
rhodopsin, that is shown here to the left.
And when light hits rhodopsin, it starts
00:09:13.660 --> 00:09:18.870
to isomorphise, which is a fancy word for
saying it changes its shape. And this
00:09:18.870 --> 00:09:24.760
transmits, in the end, electrical impulses
to our brain, which enables us to see. And
00:09:24.760 --> 00:09:28.850
this very first step of vision takes only
two hundred femtoseconds to complete. But
00:09:28.850 --> 00:09:33.060
without it, vision would not be possible.
Another very fast process that is
00:09:33.060 --> 00:09:39.649
fundamental in nature is photosynthesis,
where plants take light and CO2 and
00:09:39.649 --> 00:09:45.740
convert it to other things, among them
oxygen. And the very first excitation
00:09:45.740 --> 00:09:53.110
where light hits the plant and it starts
to make all this energy available. That
00:09:53.110 --> 00:09:56.690
also takes less than one hundredth
femtoseconds to complete. So really the
00:09:56.690 --> 00:10:01.930
fundamental questions of life lie at this
timescale. And I'd like to just mention
00:10:01.930 --> 00:10:06.820
that all of these processes are not only
very fast, but they also take place in
00:10:06.820 --> 00:10:11.540
very small objects, that are of a size of
a few atoms to nanometers, which makes it
00:10:11.540 --> 00:10:15.480
also hard to observe because we cannot see
them with the naked eye or with standard
00:10:15.480 --> 00:10:22.240
microscopes. Now, we've seen already that
it's important to choose the right
00:10:22.240 --> 00:10:28.190
exposure time to get a clear image of
something that's moving, but the kind of
00:10:28.190 --> 00:10:32.510
method that we need for taking such a
photograph of something that is moving
00:10:32.510 --> 00:10:37.220
depends a lot on the timescale. So for
stuff that is moving within seconds or
00:10:37.220 --> 00:10:43.459
fractions of a second, we can see that
with the naked eye, we can use cameras to
00:10:43.459 --> 00:10:48.180
resolve faster motion, very much like
Muybridge did with the very first camera.
00:10:48.180 --> 00:10:53.500
Today, of course, we can go much faster to
maybe a few microseconds. With very fancy
00:10:53.500 --> 00:10:57.230
cameras called opto-electronic street
cameras - i won't go into detail here - we
00:10:57.230 --> 00:11:01.560
can go down to picoseconds. So we are
already very close to the motion of
00:11:01.560 --> 00:11:06.899
molecules, but we are not quite there yet.
The timescale that we want to investigate
00:11:06.899 --> 00:11:11.060
is a femtosecond. So really a time
timescale of molecular motion and
00:11:11.060 --> 00:11:16.550
electronics are not fast enough to reach
this timescale. So we need something new.
00:11:16.550 --> 00:11:21.380
And fortunately, we can create light
pulses that serve as to say flashes, but
00:11:21.380 --> 00:11:26.199
take snapshots of our moving molecules
with femtosecond time resolution and light
00:11:26.199 --> 00:11:31.970
pulses can be made so short. So in the
following, I'm going to show you a bit
00:11:31.970 --> 00:11:37.550
more detail on how we can use these ultra
short light pulses to take snapshots of
00:11:37.550 --> 00:11:43.699
moving molecules. The first method that I
would like to briefly show you is X-Ray
00:11:43.699 --> 00:11:49.120
diffraction, where we have an ultra short
pulse, an X-Ray pulse coming in. It hits a
00:11:49.120 --> 00:11:54.899
sample shown here in the red bubbles.
That's essentially a molecule that that we
00:11:54.899 --> 00:12:01.541
just place in the beam and it produces a
so-called diffraction pattern that we can
00:12:01.541 --> 00:12:07.829
then record on a screen. Now, the whole
process is quite complicated. So I like to
00:12:07.829 --> 00:12:15.420
just sketch the very basics of it. We see
here X-Ray radiation hitting a crystalline
00:12:15.420 --> 00:12:23.040
sample here to the left and the sample is
excited, starts to radiate X-Ray back and
00:12:23.040 --> 00:12:29.199
on the right we can see the X-Rays leaving
the sample again. They will interfere and
00:12:29.199 --> 00:12:34.670
we can record this pattern on the screen.
So this is what we see here in this
00:12:34.670 --> 00:12:42.570
visualization to the right. With this, we
can feed a reconstructionalgorithm that
00:12:42.570 --> 00:12:47.110
allows us to transform back our
diffraction pattern that we've seen here
00:12:47.110 --> 00:12:53.570
for for in this case a bio molecule. We
can reconstruct from that the image as it
00:12:53.570 --> 00:13:00.899
was in real space. So this is some
protein, I believe. X-ray diffraction is
00:13:00.899 --> 00:13:09.149
very nice for resolving small structures
with atomic detail. Another method how we
00:13:09.149 --> 00:13:15.560
can take snapshots using ultra short
pulses, that I would like to briefly
00:13:15.560 --> 00:13:20.830
introduce is absorption spectroscopy. Now
you may know that light contains several
00:13:20.830 --> 00:13:27.540
colors. For example, you've surely have
held a prism in hand, and the prism can
00:13:27.540 --> 00:13:32.130
break white light up into all the colors
of a rainbow, that we can see with the
00:13:32.130 --> 00:13:38.149
eye. Now we can do the same with X-Ray
pulses. Then we cannot see the colors
00:13:38.149 --> 00:13:44.410
anymore. So just let's just stick with a
prism here. When we place a molecule in
00:13:44.410 --> 00:13:49.200
front of all these colors, the molecule
will block certain colors. That's quantum
00:13:49.200 --> 00:13:56.920
mechanics. You just have to believe it or
learn about it in long studies. So the
00:13:56.920 --> 00:14:03.110
molecule is placed in front of all these
colors. And to be right, the absorption
00:14:03.110 --> 00:14:08.110
spectrum is recorded and the parts of the
spectrum that are very bright correspond
00:14:08.110 --> 00:14:14.480
to the colors that have been blocked by
the molecule. And this is a very nice
00:14:14.480 --> 00:14:19.089
technique to investigate ultra short
dynamics, because where these lines are
00:14:19.089 --> 00:14:24.880
located is characteristic of the chemical
elements that we find in the molecule. For
00:14:24.880 --> 00:14:28.760
example, if we use X-Ray radiation for
this specific molecule, that I've shown
00:14:28.760 --> 00:14:33.810
here lysine, that's not so important which
molecule it is. We have three different
00:14:33.810 --> 00:14:39.269
atoms in this molecule that are important
carbon, nitrogen and oxygen and they
00:14:39.269 --> 00:14:43.880
absorb at very different colors so we can
keep them apart when we take the spectrum.
00:14:43.880 --> 00:14:49.120
But not only that, we can take the
spectrum at a later time when the molecule
00:14:49.120 --> 00:14:54.420
has moved around a bit and we will see
that the colors, the position of the lines
00:14:54.420 --> 00:14:59.829
have changed a tiny bit. So it's really
not much and I accelerated it already in
00:14:59.829 --> 00:15:06.070
this visualization quite a bit. But with
experimental methods, we can resolve this.
00:15:06.070 --> 00:15:11.420
And this allows us to then trace back to
how the molecule was moving in between
00:15:11.420 --> 00:15:16.889
when we took these two snapshots. There
are many more methods that you can use to
00:15:16.889 --> 00:15:21.820
take ultrafast images. So we call them
probe signals because we probe the
00:15:21.820 --> 00:15:26.889
ultrafast motion of a molecule with such
an ultra short pulse. For example, we can
00:15:26.889 --> 00:15:33.009
record photo electrons or we can record
fragments of a molecule and many more. But
00:15:33.009 --> 00:15:37.519
I won't go into further detail here
because this is not an exhaustive list of
00:15:37.519 --> 00:15:42.320
methods that we can use. I'd rather like
to show you how we can take molecular
00:15:42.320 --> 00:15:47.709
movies so how we can combine all these
ultrashort pulses to in the end film a
00:15:47.709 --> 00:15:55.050
molecule in action. Now we've already seen
in the movie of the horse that we need to
00:15:55.050 --> 00:16:00.990
stitch several snapshots together and then
we have a full picture, full motion of a
00:16:00.990 --> 00:16:06.740
molecule. So we just like to do the same,
but ten to the 15 times faster, should not
00:16:06.740 --> 00:16:13.029
be too difficult, right? So we use our
ultra short pulse. First ultrasound parts
00:16:13.029 --> 00:16:17.980
that we use as a trigger, parts that sets
off the motion and the molecule. This
00:16:17.980 --> 00:16:22.410
defines us a certain time zero in our
experiment and makes it sort of repeatable
00:16:22.410 --> 00:16:28.300
because we always start the same kind of
motion by giving it a small hit and now
00:16:28.300 --> 00:16:34.870
it's just moving around. So we wait for a
certain time, a time delay and then come
00:16:34.870 --> 00:16:40.940
in with a probe pulse. The probe pulse
takes a snapshot of a molecule. This goes
00:16:40.940 --> 00:16:45.320
to some detector, goes to a kind of
complicated reconstruction method that we
00:16:45.320 --> 00:16:51.709
just execute from our screen. And with
this, we reconstruct a snapshot of a
00:16:51.709 --> 00:16:56.850
molecule. But this is only one snapshot
and we want a whole movie. So we need to
00:16:56.850 --> 00:17:02.240
repeat this process over and over again by
shining and more and more probe pulses.
00:17:02.240 --> 00:17:08.040
And this will create more and more
snapshots of a molecule. And in the end,
00:17:08.040 --> 00:17:12.209
we could stitch all of these together and
we would arrive at the same image that you
00:17:12.209 --> 00:17:18.430
see in the in the middle where the
molecules is happily moving around. There
00:17:18.430 --> 00:17:24.790
is one little problem: The probe pulse
typically destroys the molecule. This is
00:17:24.790 --> 00:17:28.500
very different. This is very different
from taking pictures of a horse. The horse
00:17:28.500 --> 00:17:36.289
normally survives. laughting So the probe pulse
destroys the molecule. It just goes away.
00:17:36.289 --> 00:17:41.830
So for each of these snapshots we need to
use a new molecule. So we typically have a
00:17:41.830 --> 00:17:47.970
stream of samples that is falling from the
top to the bottom in our experiment. And
00:17:47.970 --> 00:17:51.690
then we have to carefully align two pulses
a trigger pulse and a probe pulse that
00:17:51.690 --> 00:17:57.610
come together and take a snapshot of this
molecule. And of course, we have to find a
00:17:57.610 --> 00:18:03.120
method on how to make identical molecules
available in - Yeah - you see, there's a
00:18:03.120 --> 00:18:07.860
lot of complications with doing these
experiments that I'm completely leaving
00:18:07.860 --> 00:18:15.090
out here. So now we want to take a
molecular movie and we know that we want
00:18:15.090 --> 00:18:19.700
to have ultra short pulses to do so. But I
didn't tell you yet what kind of light
00:18:19.700 --> 00:18:24.750
source we need. So there are many light
sources all around us. We have here lights
00:18:24.750 --> 00:18:29.450
from lamps. I have a light in my laser
pointer with light from the sun. But we
00:18:29.450 --> 00:18:35.380
need quite specific light sources to take
these snapshots of molecular motion. We've
00:18:35.380 --> 00:18:37.970
already established that we want
ultrashort pulses because else we cannot
00:18:37.970 --> 00:18:44.260
resolve femtosecond dynamics, but for the
proper kind of wavelength that we need I
00:18:44.260 --> 00:18:47.980
would like to quickly remind you of the
electromagnetic spectrum that you've
00:18:47.980 --> 00:18:54.090
probably seen at some point in high
school. So, so light, as you see here in
00:18:54.090 --> 00:18:58.970
the bottom picture is an electromagnetic
wave that comes in different wavelengths.
00:18:58.970 --> 00:19:05.440
They can be quite long as in the case of
radio waves to the very left. Then we have
00:19:05.440 --> 00:19:08.470
the region of visible light shown here as
the rainbow that we can perceive with our
00:19:08.470 --> 00:19:14.270
eyes. And then we have wavelengths that
are too short to see with our eyes. First,
00:19:14.270 --> 00:19:18.880
UV radiation, that gives us a tan in the
summer if we leave our house and then we
00:19:18.880 --> 00:19:25.400
have X-ray radiation, soft and hard X-ray
radiation that have atomic wavelength. So
00:19:25.400 --> 00:19:30.570
the wavelength is really on the order of
the size of an atom. So what kind of
00:19:30.570 --> 00:19:37.080
wavelength do we need to study ultra short
dynamics - ultra fast dynamics? We can
00:19:37.080 --> 00:19:43.321
first think about what kind of wavelength
we need when we want to construct an ultra
00:19:43.321 --> 00:19:49.600
short pulse. I've drawn here two pulses to
the left, a slightly longer pulse to the
00:19:49.600 --> 00:19:53.980
right, a shorter pulse. And now if you
think about squeezing the left parts
00:19:53.980 --> 00:19:58.610
together such that it becomes shorter and
shorter, you see visually that the
00:19:58.610 --> 00:20:03.890
wavelength also needs to shrink. So we
need shorter wavelengths for the shorter
00:20:03.890 --> 00:20:10.350
the pulse we want to make. So this will be
located somewhere here in this region of
00:20:10.350 --> 00:20:15.450
the electromagnetic spectrum. And another
important thing that we need to keep in
00:20:15.450 --> 00:20:21.250
mind is if we want to take pictures by
X-ray diffraction, we are limited, so we
00:20:21.250 --> 00:20:27.429
can only resolve structures that are about
the same size as the wavelength we used to
00:20:27.429 --> 00:20:31.919
take our diffraction image. So if we want
to take a picture of something with atomic
00:20:31.919 --> 00:20:37.549
resolution, our wavelength needs to be of
atomic size as well. And this places us in
00:20:37.549 --> 00:20:47.169
the region of X-Rays drawn here, that have
a wavelength of less than a nanometer. So
00:20:47.169 --> 00:20:51.510
we can establish that we want small
wavelengths in general. We have two
00:20:51.510 --> 00:20:55.840
additional requirements that would just
touch upon very briefly. First, we need
00:20:55.840 --> 00:21:01.460
very brilliant pulses because the pulses
are so short, we need to have a lot of
00:21:01.460 --> 00:21:06.760
light in the short pulse. You can think
about taking a picture in a dark room with
00:21:06.760 --> 00:21:12.270
a bad camera. You won't see anything. So
we need very bright flashes of light.
00:21:12.270 --> 00:21:16.159
Another requirement is we need coherent
laser light. So we cannot just use any
00:21:16.159 --> 00:21:20.130
light, but it needs to have certain
properties like laser light.
00:21:20.130 --> 00:21:25.110
Unfortunately, the lasers that you can buy
commercially do not operate in the region
00:21:25.110 --> 00:21:29.110
of the electromagnetic spectrum that we
are interested in. So we need to come up
00:21:29.110 --> 00:21:34.830
with something new. And I will show you
how we can generate ultra short pulses
00:21:34.830 --> 00:21:39.380
both in the laboratory where we can
generate pulses that are very short and
00:21:39.380 --> 00:21:46.650
extend up to maybe the soft X-ray region.
And another method to generate ultra short
00:21:46.650 --> 00:21:53.380
pulses is at free electron laser sources,
where we can go really to the hard X-Ray
00:21:53.380 --> 00:21:59.480
regime. But first I'd like to go to the
laboratory. So in the laboratory, it's
00:21:59.480 --> 00:22:03.380
possible to generate an ultrashort pulse by
using a process that's called high
00:22:03.380 --> 00:22:08.201
harmonic generation. In high harmonic
generation we start off of a high
00:22:08.201 --> 00:22:13.470
intensity pulse, that's a red pulse coming
in from the left, which which is focused
00:22:13.470 --> 00:22:19.350
in a gas cell. And from there, it
generates new frequencies of light. So the
00:22:19.350 --> 00:22:24.770
light that comes out is no longer red, but
it's violet, blue. We cannot see it with
00:22:24.770 --> 00:22:27.980
the naked eye. So that's an artist's
impression of how high harmonic generation
00:22:27.980 --> 00:22:33.470
works. Before going into more detail about
why this method is so good at producing
00:22:33.470 --> 00:22:38.770
ultra short pulses, I'd like to mention
that this is only possible because we have
00:22:38.770 --> 00:22:44.120
the high intensity driving pulses, the red
laser pulses available. This goes back to
00:22:44.120 --> 00:22:47.270
work by Donna Strickland and Gerard
Mourou, who were awarded the Nobel Prize
00:22:47.270 --> 00:22:54.950
in the year 2018 in physics for this work
that has been done in the 80s. Now we're
00:22:54.950 --> 00:22:59.690
coming to the only equation of his talk,
which is this equation that relates the
00:22:59.690 --> 00:23:07.500
energy width and the time duration of a
ultra short pulse. By the law of fourier
00:23:07.500 --> 00:23:13.150
limits we cannot have pulses that are very
short in time and at the same time very
00:23:13.150 --> 00:23:17.700
narrow in energy. But we need to choose
one. So if we want to have policies that
00:23:17.700 --> 00:23:23.640
are very short in time like the pulse that
I've shown here on the bottom, that is
00:23:23.640 --> 00:23:26.669
actually only two hundred fifty
attoseconds long, so even shorter than a
00:23:26.669 --> 00:23:33.059
femtosecond, then we need to have a very
broad width in energy. And this means
00:23:33.059 --> 00:23:37.450
combining a lot of different colors inside
of this pulse. And this is what makes high
00:23:37.450 --> 00:23:41.580
harmonic generation so efficient at
creating ultra short pulses, because the
00:23:41.580 --> 00:23:47.419
spectrum that the colors that come out of
high harmonic generation are shown here
00:23:47.419 --> 00:23:52.080
and they really span a long width. So we
get a lot of different colors with about
00:23:52.080 --> 00:23:57.669
the same intensity. And you can think of
it like putting them all back together
00:23:57.669 --> 00:24:04.990
into one attosecond pulse. That is very
short in time. This method has really made
00:24:04.990 --> 00:24:08.950
a big breakthrough in the generation of
ultra short laser pulses we see here a
00:24:08.950 --> 00:24:15.300
plot of a time duration of laser pulses
versus the year, and we see that since the
00:24:15.300 --> 00:24:23.520
invention of the laser, here in the mid
60s, there was a first technological
00:24:23.520 --> 00:24:28.920
progress and shorter and shorter pulses
could be generated. But then in the 80s,
00:24:28.920 --> 00:24:34.399
there was a limit that had been reached of
about five femtoseconds, I believe. And we
00:24:34.399 --> 00:24:39.710
could not really go farther than that and
only with high harmonic generation, that
00:24:39.710 --> 00:24:45.419
sets in here shortly before the year 2000,
we were able to generate pulses that are
00:24:45.419 --> 00:24:51.760
of a femtosecond duration. So that really
touch the timescale of molecular motion.
00:24:51.760 --> 00:24:57.200
The current world record is a pulse, that
is only 43 attoseconds long, established
00:24:57.200 --> 00:25:02.289
in the year 2017. So that's really the
timescale of electrons and we can do all
00:25:02.289 --> 00:25:06.649
sorts of nice experiments with it where we
directly observe electronic motion in
00:25:06.649 --> 00:25:12.470
atoms and molecules. This is all very
nice, but it has one limitation: We cannot
00:25:12.470 --> 00:25:17.169
go to hard X-rays, at least not right now.
So high harmonic generation cannot produce
00:25:17.169 --> 00:25:23.000
the kind of very short wavelengths that we
need in order to to do X-ray diffraction
00:25:23.000 --> 00:25:29.429
experiments with atomic resolution. So if
we want to have ultra short pulses that
00:25:29.429 --> 00:25:35.730
have X-Ray wavelengths, we need to build
right now very complex, very big machines,
00:25:35.730 --> 00:25:43.080
the so-called free electon lasers. Now,
this would be a specific light source that
00:25:43.080 --> 00:25:48.059
can produce ultra short pulses with X-ray
wavelengths in itself. The X-Ray
00:25:48.059 --> 00:25:52.720
wavelengths is not so new. We know how to
take X-ray images for about one hundred
00:25:52.720 --> 00:25:58.910
and thirty years and already in the 50s.
Rosalind Franklin, who is looking at a
00:25:58.910 --> 00:26:05.110
microscope here, was able to take a
picture of DNA, an X-ray diffraction
00:26:05.110 --> 00:26:11.880
pattern of a DNA double helix that was
successful in revealing the double helix
00:26:11.880 --> 00:26:19.370
structure of our genetic code. But this is
not a time resolved measurement. So think
00:26:19.370 --> 00:26:24.650
of it as you have a molecule that is in
crystalline form, so it's not moving
00:26:24.650 --> 00:26:32.480
around and we can just take an X-ray image
of it, it's not going anywhere. But if we
00:26:32.480 --> 00:26:36.860
want - if we want to take a picture of
something that is moving, we need to have
00:26:36.860 --> 00:26:43.400
very short pulses. But we still need the
same number of what we call photons, light
00:26:43.400 --> 00:26:48.890
particles. Or think of it as we need more
brilliant X-ray flashes of light than we
00:26:48.890 --> 00:26:55.140
could obtain before. And there was very
nice technological development in the past
00:26:55.140 --> 00:27:01.690
50 years or so, where we were able to go
from the X-ray tube to newer light sources
00:27:01.690 --> 00:27:07.260
called Synchrotron, and today, free
electro lasers that always increase the
00:27:07.260 --> 00:27:12.180
peak brilliance in an exponential way. So
we can take really brilliant, really
00:27:12.180 --> 00:27:17.270
bright X-ray flashes right now. I cannot
go into the details of all of that, but I
00:27:17.270 --> 00:27:21.350
found a very nice talk from two years ago,
but actually explains everything from
00:27:21.350 --> 00:27:28.590
Synchrotron to FELs still available online
if you're interested in this work. And as
00:27:28.590 --> 00:27:32.110
always, if something is failing scaling
exponentially, most of you will be
00:27:32.110 --> 00:27:37.840
familiar with Moore's Law, that tells us
about the exponential scaling of
00:27:37.840 --> 00:27:45.169
transistors. If something grows this fast,
it really opens up a new series of
00:27:45.169 --> 00:27:49.299
experiments of new technological
applications that no one has thought of
00:27:49.299 --> 00:27:55.600
before. And the same is true with free
electron lasers. So I'm going to focus
00:27:55.600 --> 00:28:00.450
just on the most brilliant light sources
for X-rays. Right now, the free electron
00:28:00.450 --> 00:28:06.230
lasers that are at the top right here of
this graph have been around for maybe 10
00:28:06.230 --> 00:28:13.220
years or so. I cannot go into a lot of
detail on how to generate ultrashort
00:28:13.220 --> 00:28:18.220
pulses with X-Rays. So I'd like to
give you just a very broad picture of how
00:28:18.220 --> 00:28:23.909
this works. First, we need a bunch of
electrons, that is accelerated to
00:28:23.909 --> 00:28:29.640
relativistic speed. This sounds very easy,
but is actually part of a two kilometer
00:28:29.640 --> 00:28:35.490
long accelerator, that we have to build
and maintain. Now we have this bunch here
00:28:35.490 --> 00:28:41.230
of electrons shown in red and it's really
fast and now we can bring it into
00:28:41.230 --> 00:28:46.519
something that is called an undulator.
That's a series of alternating magnets,
00:28:46.519 --> 00:28:52.270
shown here in green on blue for the
alternating magnets. And you may remember,
00:28:52.270 --> 00:28:57.100
that when we put an electron, that is as a
charged particle, into a magnetic field,
00:28:57.100 --> 00:29:02.820
the Lorence force will drive it away. And
if you have alternating magnets, then the
00:29:02.820 --> 00:29:08.890
electron will go on a sort of wiggly path
in this undulator. And the electron is a
00:29:08.890 --> 00:29:13.890
charged particle as it is wiggling around
wherever it turns around, it will emit
00:29:13.890 --> 00:29:18.309
radiation, that happens to be in the X-ray
region of the electromagnetic spectrum,
00:29:18.309 --> 00:29:23.260
which is exactly what we want. We can
watch this little movie here to see a
00:29:23.260 --> 00:29:28.800
better picture. So this is the undulating
seeing from the side. We now go inside of
00:29:28.800 --> 00:29:35.880
the undulator. We have a series of
alternating magnets. Now the electron
00:29:35.880 --> 00:29:42.430
bunch shows up and you see the wiggly
motion as it passes the different magnets.
00:29:42.430 --> 00:29:48.390
And you see the bright X-Ray flash that is
formed and gets stronger and stronger as
00:29:48.390 --> 00:29:54.779
the electron bunch passes the undulator. So
we need several of these magnet pairs to
00:29:54.779 --> 00:30:00.130
in the end, get the very bright X-Ray
flash. And at the end of the undulator we
00:30:00.130 --> 00:30:06.240
dump the electron, we don't really need
this electron bunch anymore and continue
00:30:06.240 --> 00:30:13.830
with a very bright X-Ray flash. This whole
process is a bit stochastic in nature, but
00:30:13.830 --> 00:30:18.890
it's amplifying itself in because of the
undulator. This is why the longer the
00:30:18.890 --> 00:30:29.020
undulator is, the more bright X-Ray
flashes we can generate. This whole thing
00:30:29.020 --> 00:30:33.149
is kind of complicated to build, it's a
very complex machine. So right now there
00:30:33.149 --> 00:30:37.740
are only very few free electron lasers in
the world. First one in California called
00:30:37.740 --> 00:30:43.850
LCLS 1, currently being upgraded to LCLS
2. There are several in Europe. There's
00:30:43.850 --> 00:30:49.330
one in Switzerland, in Italy and Hamburg.
So there's a Flash that does not operate
00:30:49.330 --> 00:30:54.559
in the hot X-ray regime, but was kind of
first free electron laser. That's the most
00:30:54.559 --> 00:30:58.919
recent addition to the free electron laser
zoo. It's the European XFEL also located
00:30:58.919 --> 00:31:04.399
in Hamburg. And then we have some of these
light sources in Asia, in Korea, South
00:31:04.399 --> 00:31:10.730
Korea, Japan, and one currently under
construction in Shanghai. I'd like to show
00:31:10.730 --> 00:31:15.740
you a bit more details about the European
X-ray free electron laser, because it's
00:31:15.740 --> 00:31:23.980
closest to us, and at least closest to
where I work. So the European XFEL is a
00:31:23.980 --> 00:31:30.049
three point four kilometer long machine
that is funded by in total 12 countries,
00:31:30.049 --> 00:31:36.350
So Germany and Russia paying the most and
then the other 10 countries also providing
00:31:36.350 --> 00:31:41.130
to the construction and maintenance costs.
This machine starts at the DESY campus,
00:31:41.130 --> 00:31:47.340
but as shown here to the right of the
picture. And then we have first an
00:31:47.340 --> 00:31:51.580
accelerator line for the electrons that
it's already one point seven kilometers
00:31:51.580 --> 00:31:58.450
long and where we add electrons reach
their relativistic speed. Then the
00:31:58.450 --> 00:32:04.600
undulate comes in, so the range of magnets
where we X-Ray flashes are produced. The x
00:32:04.600 --> 00:32:09.330
X-Ray flashes then cross the border to
Schleswig-Holstein, laughter shown here, on the
00:32:09.330 --> 00:32:16.450
other side in a new federal state. They
reach the experimental hall. We have in
00:32:16.450 --> 00:32:21.049
total six experimental end stations at the
European XFEL that provide different
00:32:21.049 --> 00:32:24.419
instrumentation, depending on which kind
of system you want to study, you need
00:32:24.419 --> 00:32:30.960
slightly different instruments. And it's
not only for taking molecular movies, but
00:32:30.960 --> 00:32:36.010
the XFEL is used, among others, for
material science, for the imaging of bio
00:32:36.010 --> 00:32:40.950
molecules, for femtosecond chemistry, all
sorts of things. So really wide range of
00:32:40.950 --> 00:32:46.580
applications. It's right now the the
fastest such light source can take twenty
00:32:46.580 --> 00:32:51.580
seven thousand flashes per second, which
is great because every flash is one
00:32:51.580 --> 00:32:56.110
picture. So if we want to take a lot of
snapshots, if you want to generate a lot
00:32:56.110 --> 00:33:01.179
of data in a short time, it's great to
have as many flashes per second as
00:33:01.179 --> 00:33:08.720
possible. And as you can imagine, it's
kind of expensive since there are so few
00:33:08.720 --> 00:33:15.450
free electron lasers in the world to take
measurements there. The complete price tag
00:33:15.450 --> 00:33:19.919
for constructing this machine, it took
eight years and cost one point two billion
00:33:19.919 --> 00:33:25.580
euros, which may seem a lot, but it's the
same amount that we spend on concert halls
00:33:25.580 --> 00:33:39.029
in Hamburg. loud laughter applause So kind of comparable. Now,
when you factor in maintenance and so on,
00:33:39.029 --> 00:33:45.639
I think a minute of X-Ray beam at such an
XFEL cost several thousands of tens of
00:33:45.639 --> 00:33:51.570
thousands of euros in the end. So getting
measurement time is complicated and there
00:33:51.570 --> 00:33:56.710
are committees that select the most
fruitful approaches and so on. So in order
00:33:56.710 --> 00:34:04.000
to not to waste or do taxpayers money.
With this, I'd like to make a small
00:34:04.000 --> 00:34:08.020
comparison of the light sources that I've
introduced now. So I introduced the
00:34:08.020 --> 00:34:12.810
laboratory light sources and the XFEL
light source. In general, in the
00:34:12.810 --> 00:34:17.600
laboratory we can generate very short
pulses of less than 100 attoseconds by now
00:34:17.600 --> 00:34:22.980
and in the XFEL we are limited to
something about 10 femtoseconds right now.
00:34:22.980 --> 00:34:29.630
In terms of brilliance the XFELs can go to
much more bright pulses, simply because
00:34:29.630 --> 00:34:32.960
they are bigger machines and high harmonic
generation in itself is a kind of
00:34:32.960 --> 00:34:38.870
inefficient process. In terms of
wavelength X-Ray free electron lasers
00:34:38.870 --> 00:34:42.620
enable us to reach these very short
wavelengths with X-Rays, that we need to
00:34:42.620 --> 00:34:48.290
get atomic resolution of defractive
images. In the laboratory we are a bit
00:34:48.290 --> 00:34:54.850
more limited to maybe the soft X-ray
region. There's another important thing to
00:34:54.850 --> 00:34:59.890
keep in mind when we do experiments,
that's the control of pulse parameters. So
00:34:59.890 --> 00:35:03.040
is every pulse that comes out of my
machine the same as the one that came out
00:35:03.040 --> 00:35:08.520
of my machine before. And since the XFEL
produces pulses by what is in the end, a
00:35:08.520 --> 00:35:13.630
stochastic process, that's not really the
case. So the control of possible
00:35:13.630 --> 00:35:20.400
parameters is not really given. This is
much better in the laboratory. And in
00:35:20.400 --> 00:35:23.280
terms of cost and availability, it would
of course, be nice if we could do more
00:35:23.280 --> 00:35:29.620
experiments in the lab. Then at the XFEL
simply because we XFEL ls so expensive to
00:35:29.620 --> 00:35:35.730
build and maintain and we have so few of
them in the world. And you can see this
00:35:35.730 --> 00:35:41.840
tunnel here. It stretches for two
kilometers or so, all packed with very
00:35:41.840 --> 00:35:53.260
expensive equipment. So I'd like to show
you a brief example of what we can learn
00:35:53.260 --> 00:35:58.500
in ultrafast science. So this is a
theoretical work that we did in our group.
00:35:58.500 --> 00:36:03.720
So no experimental data, but still nice to
see. This is concerned with an organic
00:36:03.720 --> 00:36:09.730
solar cell. So we all know solar cells.
They convert sunlight to electric energy
00:36:09.730 --> 00:36:14.480
that we can use in our devices. The nice
thing about organic solar cells is that
00:36:14.480 --> 00:36:20.900
they are foldable, very lightweight, and
we can produce them cheaply. The way that
00:36:20.900 --> 00:36:25.630
such a solar cell works is we have light
shining in and at the bottom of the solar
00:36:25.630 --> 00:36:29.410
cell there sits an electrode that collects
all the charges and creates an electric
00:36:29.410 --> 00:36:33.570
current. Now light creates a charge that
somehow needs to travel down there to this
00:36:33.570 --> 00:36:42.000
electrode and in fact, many of these
charges. So the important thing where we
00:36:42.000 --> 00:36:46.640
build such an organic solar cell is that
we need a way to efficiently transport
00:36:46.640 --> 00:36:55.240
these charges. And we can do so by putting
polymers inside. A polymer is just a
00:36:55.240 --> 00:36:59.670
molecule that is made up of two different
or two or more different smaller
00:36:59.670 --> 00:37:05.100
molecules. And one such polymer, which
should be very efficient at transporting
00:37:05.100 --> 00:37:10.170
these charges is BT-1T, that is shown here
of a name is not so important, it's an
00:37:10.170 --> 00:37:15.080
abbreviation. Because in BT-1T when we
create a charge at one end of a molecule
00:37:15.080 --> 00:37:20.180
here at the top, it travels very quickly
to the other side of a molecule and you
00:37:20.180 --> 00:37:26.880
can imagine stacking several of these
BT-1T or especially of the Ts together,
00:37:26.880 --> 00:37:31.170
putting it in this material. And then we
have a very efficient flow of energy in
00:37:31.170 --> 00:37:41.880
our organic solar cell. So what we did was
we calculated the ultrafast charge
00:37:41.880 --> 00:37:48.161
migration in BT-1T, shown here to the
right. The pink thing is the charge
00:37:48.161 --> 00:37:53.570
density that was created by an initial
ionization of the molecule. And now I show
00:37:53.570 --> 00:37:58.110
you the movie, how this charge is moving
around in a molecule so you can see
00:37:58.110 --> 00:38:02.820
individual atoms moving, the whole
molecules vibrating a bit. And the charge,
00:38:02.820 --> 00:38:10.570
if you look closely, is locating on the
right half of a molecule within about 250
00:38:10.570 --> 00:38:16.780
femtoseconds. Now, we cannot observe this
charge migration directly by looking at
00:38:16.780 --> 00:38:21.290
this pink charge density that I've drawn
here, because it's at least for us, not
00:38:21.290 --> 00:38:26.151
experimentally observable directly. So we
need an indirect measurement, an X-Ray
00:38:26.151 --> 00:38:30.050
absorption spectroscopy that I showed you
in the beginning could be such a
00:38:30.050 --> 00:38:35.310
measurement. Because in the X-Ray
absorption spectrum of BT-1T that I've
00:38:35.310 --> 00:38:41.030
shown here in the bottom left, we see
distinct peaks depending on where the
00:38:41.030 --> 00:38:47.070
charge is located. Initially the charge is
located at the top sulfur atom here and
00:38:47.070 --> 00:38:53.780
this molecule and we will see a peek at
this color. Once the charge moves away to
00:38:53.780 --> 00:38:58.060
the bottom of a molecule to the other
half, we will see a peak at the place
00:38:58.060 --> 00:39:02.680
where nothing is right now because the
charge is not there. But if I start this
00:39:02.680 --> 00:39:09.600
movie, we will again see very fast charge
transfer. So within about two hundred
00:39:09.600 --> 00:39:14.050
femtoseconds, the charge goes from one end
to the molecule to the other end of a
00:39:14.050 --> 00:39:19.470
molecule. And it would be really nice to
see this in action in the future XFEL
00:39:19.470 --> 00:39:25.720
experiment. But the process is very long.
You need to apply for time at an XFEL. You
00:39:25.720 --> 00:39:29.850
need to evaluate all the data. So maybe a
couple of years from now we will have the
00:39:29.850 --> 00:39:37.730
data available. Right now we are stuck
with this movie, that we calculated. Now,
00:39:37.730 --> 00:39:43.590
towards the end of my talk, I'd like to go
beyond the molecular movie. So I've shown
00:39:43.590 --> 00:39:47.860
you now how to generate the light pulses
and an example of what we can study with
00:39:47.860 --> 00:39:53.540
these light pulses. But this is not all we
can do: So when you think of a chemical
00:39:53.540 --> 00:39:58.420
reaction, you might remember high school
chemistry or something like this, which is
00:39:58.420 --> 00:40:03.460
always foaming and exploding and nobody
really knows what is going on. So a
00:40:03.460 --> 00:40:08.480
chemical reaction quite naturally involves
molecular dynamics, for example, the
00:40:08.480 --> 00:40:13.180
decomposition of a molecule to go from
here, from the left side to the right
00:40:13.180 --> 00:40:18.180
side, the molecules will somehow need to
rearrange so all the atoms will have moved
00:40:18.180 --> 00:40:24.340
quite a bit. We've seen already how we can
trigger these chemical reactions or these
00:40:24.340 --> 00:40:29.770
molecular motion that was part of a
molecular movie. But it would be really
00:40:29.770 --> 00:40:34.430
cool if we could control the reaction with
light. So the way to do this, it's not
00:40:34.430 --> 00:40:38.920
currently something that is possible, but
maybe in the near future, would be to
00:40:38.920 --> 00:40:44.440
implement a sort of optimisation feedback
loop. So we would record the fragments of
00:40:44.440 --> 00:40:48.970
our reaction, send it to an optimization
routine that will also be quite
00:40:48.970 --> 00:40:53.620
complicated and will need to take into
account the whole theory of how light and
00:40:53.620 --> 00:40:58.770
matter, interact and so on. And this
optimization routine would then generate a
00:40:58.770 --> 00:41:03.800
new sequence of ultra short pulses and
with this feedback loop, it might be
00:41:03.800 --> 00:41:10.100
possible to find the right pulses to
control chemical reactions, taking into
00:41:10.100 --> 00:41:17.640
account the quantum nature of this motion
and so on. Right now this is not possible.
00:41:17.640 --> 00:41:23.651
First, because the whole process of how we
can generate these ultra short pulses is
00:41:23.651 --> 00:41:28.760
not so well controlled that we could
actually implement it in such a loop. And
00:41:28.760 --> 00:41:33.000
also the step optimization routine is more
complex than it looks like here in this
00:41:33.000 --> 00:41:39.490
picture. So this is something that people
are working on at the moment, but this
00:41:39.490 --> 00:41:44.300
would be something like the ultra fast
wishlist for next Christmas, not this
00:41:44.300 --> 00:41:49.710
Christmas. So we've succeeded in taking a
molecular movie, but we would also like to
00:41:49.710 --> 00:41:55.270
be able to direct a molecular movie. So to
go beyond just watching nature, but
00:41:55.270 --> 00:42:01.460
controlling nature because this is what
humans like to best, laughing fortunately or
00:42:01.460 --> 00:42:06.230
unfortunately, it depends. So I'd like to
just show you that this is really an ultra
00:42:06.230 --> 00:42:12.220
fast developing field. There's lots of new
research papers every day, every week
00:42:12.220 --> 00:42:18.070
coming in, studying all sorts of systems.
When you just take a quick and dirty
00:42:18.070 --> 00:42:22.820
metric of how important ultrafast science
is this is the number of articles per year
00:42:22.820 --> 00:42:28.500
that mentioned ultrafast in Google
Scholar, it's exponentially growing. At
00:42:28.500 --> 00:42:30.780
the same time, the number of total
publications in Google Scholar is more or
00:42:30.780 --> 00:42:34.990
less constant, so the blue line here
outgrows the green line considerably since
00:42:34.990 --> 00:42:45.070
about ten years. So what remains to be
done? We've seen that we have light
00:42:45.070 --> 00:42:50.020
sources available to generate ultra short
pulses, but as always, when you have
00:42:50.020 --> 00:42:55.540
better machines, bigger machines, you can
take more fancy experiments. So it would
00:42:55.540 --> 00:43:00.580
be really nice to develop both lab based
sources and free electron laser sources so
00:43:00.580 --> 00:43:06.360
that we can take more, more interesting,
more complex experiments. Another
00:43:06.360 --> 00:43:10.060
important challenge, that's what people in
my research group where I work are working
00:43:10.060 --> 00:43:16.750
on is to improve theoretical calculations
because I did not go into a lot of detail
00:43:16.750 --> 00:43:21.730
on how to calculate these things, but it's
essentially quantum mechanics and quantum
00:43:21.730 --> 00:43:26.530
mechanics skales very unfavorably. So
going from a very small molecule like the
00:43:26.530 --> 00:43:32.490
glycene molecule here to something like a
protein is not doable. Simply, it cannot
00:43:32.490 --> 00:43:38.100
compute this with quantum mechanics. So we
need all sorts of new methodology to -
00:43:38.100 --> 00:43:45.100
Yeah - to better describe larger systems.
We would in general like to study not only
00:43:45.100 --> 00:43:49.410
small molecules and not only take movies
of small molecules, but really study large
00:43:49.410 --> 00:43:55.710
systems like this is the FMO complex, that
is a central in photosynthesis or solid
00:43:55.710 --> 00:44:01.410
states that are here shown in this crystal
structure simply because this is more
00:44:01.410 --> 00:44:07.790
interesting for biological, chemical
applications. And finally, as I've shown
00:44:07.790 --> 00:44:13.460
you, it would be cool to directly control
chemical reactions with light. So to find
00:44:13.460 --> 00:44:20.810
a way how to replace this mess with a
clean light pulse. With this, I'm at the
00:44:20.810 --> 00:44:26.240
end. I'd like to quickly summarize
femtodynamics, really fundamental in
00:44:26.240 --> 00:44:31.500
biology, chemistry and physics. So more or
less, the origin of life is on this
00:44:31.500 --> 00:44:36.840
timescale. We can take molecular movies
with ultrashort laser pulses and we can
00:44:36.840 --> 00:44:41.650
generate these pulses in the laboratory or
add free electron lasers with different
00:44:41.650 --> 00:44:46.820
characteristics. And we would like to not
only understand these ultra fast
00:44:46.820 --> 00:44:51.920
phenomena, but we would also like to be
able to control them in the future. With
00:44:51.920 --> 00:44:54.160
this, I'd like to thank you for your
attention and thank the supporting
00:44:54.160 --> 00:45:04.570
institutions here that funded my PhD work.
00:45:04.570 --> 00:45:10.820
applause
00:45:10.820 --> 00:45:17.380
Herald: Well, that was an interesting
talk. I enjoyed it very much. I guess this
00:45:17.380 --> 00:45:23.450
will spark some questions. If you want to
ask Caroline a question, please line up
00:45:23.450 --> 00:45:28.900
behind the microphones. We have three in
the isles between the seats if you want to
00:45:28.900 --> 00:45:36.070
leave, please do so in the door here in
the front. And until we get questions from
00:45:36.070 --> 00:45:39.570
the audience, do we have questions from
the Internet?
00:45:39.570 --> 00:45:45.311
Signal angel: Yes. Big fat random user is
curious about the design of the X-Ray
00:45:45.311 --> 00:45:48.460
detector. Do you have any information on
that?
00:45:48.460 --> 00:45:56.030
Caroline: That's also very complex. I'm
not a big expert in detectors. At this
00:45:56.030 --> 00:45:58.460
point, I really recommend watching the
talk from two years ago, that explains a
00:45:58.460 --> 00:46:05.140
lot more about the X-Ray detectors. So
what I know about the X-Ray detector is
00:46:05.140 --> 00:46:09.990
that it's very complicated to process all
the data because when you have 27000
00:46:09.990 --> 00:46:16.370
flashes of light, it produces, I think
terabytes of data within seconds and you
00:46:16.370 --> 00:46:21.030
need to somehow be able to store them and
analyze them. So there is also a lot of
00:46:21.030 --> 00:46:24.190
technology involved in the design of these
detectors.
00:46:24.190 --> 00:46:30.060
Herald: Thank you. So the first question
from microphone 2 in the middle.
00:46:30.060 --> 00:46:34.110
Microphone 2: So my question is.
Herald: Please go close to the microphone.
00:46:34.110 --> 00:46:38.860
Microphone 2: My question is regarding the
synchronization of the detector units when
00:46:38.860 --> 00:46:45.030
you're pointing to free electron laser so
you can achieve this in synchronization.
00:46:45.030 --> 00:46:50.700
Caroline: This is also very complicated.
It's easier to do in the lab. So you're
00:46:50.700 --> 00:46:54.800
talking about the synchronization of
essentially the first pulse and the second
00:46:54.800 --> 00:47:00.540
pulse. Right. So in the lab, you typically
generate the second pulse from part of the
00:47:00.540 --> 00:47:05.310
first pulse. So you have a very natural
alignment, at least in time of these two
00:47:05.310 --> 00:47:10.190
pulses. The X-ray free electron lasers
have special timing tools that allow you
00:47:10.190 --> 00:47:16.830
to find out how much is the time delay
between your two pulses. But it's true
00:47:16.830 --> 00:47:21.520
that this is complicated to achieve and
this limits the experimental time
00:47:21.520 --> 00:47:25.860
resolution to something that is even
larger than the time duration of the
00:47:25.860 --> 00:47:30.540
pulses.
Herald: So now next question from
00:47:30.540 --> 00:47:35.930
microphone number 3.
Microphone 3: Yes, i remember in the
00:47:35.930 --> 00:47:42.810
beginning, you explained that your
measuring method usually destroys your
00:47:42.810 --> 00:47:49.700
molecules. That's a bit of a contradiction
to your idea to control. laughing
00:47:49.700 --> 00:47:59.090
Caroline: In principle, yes. But, so in
the case of control, we would like to use
00:47:59.090 --> 00:48:04.650
a second pulse, that does not destroy the
molecule. But for example, at least
00:48:04.650 --> 00:48:11.140
destroys it in a controlled way, for
example. loud laughter So there's a difference between
00:48:11.140 --> 00:48:15.250
just blowing up your molecule and breaking
apart a certain part, but yet that we are
00:48:15.250 --> 00:48:19.770
interested in. And that's what we would
like to do in the control case. So we
00:48:19.770 --> 00:48:24.490
would like to be able to to control, for
example, the fragmentation of a molecule
00:48:24.490 --> 00:48:30.230
such that we only get the important part
out and everything else just goes away.
00:48:30.230 --> 00:48:34.430
Microphone3: Thank you.
Herald: So then another question for
00:48:34.430 --> 00:48:39.240
microphone 2 in the middle.
Microphone 2: So thank you for the talk. I
00:48:39.240 --> 00:48:44.470
was interested in how large structures or
molecules can you imagine with this lab
00:48:44.470 --> 00:48:48.520
contributions and with this XFEL thing?
Caroline: Sorry. Can you repeat?
00:48:48.520 --> 00:48:54.790
Microphone 2: So how large molecules can
you imagine in this laboratory with this
00:48:54.790 --> 00:48:59.120
high harmonic measures?
Caroline: So how large is not really the
00:48:59.120 --> 00:49:05.630
fundamental problem? People have taken
snapshots of viruses or bigger bio
00:49:05.630 --> 00:49:11.870
molecules. If you want to - the problem is
rather how small can we get? So yeah, to
00:49:11.870 --> 00:49:16.370
take pictures of a very small molecule.
Currently we cannot take a picture of an
00:49:16.370 --> 00:49:21.780
individual small molecule, but what people
do is they create crystals of a small
00:49:21.780 --> 00:49:25.260
molecule, sticking several of them
together and then taking images of this
00:49:25.260 --> 00:49:30.371
whole crystal for single particles, I
think right now about the scale of a virus
00:49:30.371 --> 00:49:34.180
nanometers.
Microphone 2: Thank you.
00:49:34.180 --> 00:49:38.190
Herald: OK. Do we have another question
from the Internet?
00:49:38.190 --> 00:49:43.740
Signal Angel: We have. So this is
concerning your permanent destruction of
00:49:43.740 --> 00:49:49.450
forces, I guess. How do you isolate single
atoms and molecules for analysing between
00:49:49.450 --> 00:49:55.860
the different exposures?
Caroline: Yes, excellent question. So
00:49:55.860 --> 00:50:01.440
molecules can be made available in the gas
phase by - so if you have them in a solid
00:50:01.440 --> 00:50:08.030
somewhere and you heat that up, they will
evaporate from that surface. This is how
00:50:08.030 --> 00:50:11.760
you can get them in the gas phase. This,
of course, assumes that you have a
00:50:11.760 --> 00:50:15.940
molecule, that is actually stable in the
gas space, which is not true for all
00:50:15.940 --> 00:50:22.450
molecules. And then the, the hard thing is
to align all three things. So the pump
00:50:22.450 --> 00:50:27.540
pulse, the probe pulse and the molecule
all need to be there at the same time.
00:50:27.540 --> 00:50:32.600
There are people doing whole PhD theses on
how to design gas nozzles that can provide
00:50:32.600 --> 00:50:37.670
this stream of molecules.
Signal Angel: So you basically really
00:50:37.670 --> 00:50:41.190
having a stream coming from a nozzle?
Caroline: Yes.
00:50:41.190 --> 00:50:43.300
Signal Angel: It's a very thin stream, I
guess.
00:50:43.300 --> 00:50:47.620
Caroline: Yes.
Signal Angel: Then you're exposing it like
00:50:47.620 --> 00:50:49.990
in a regular interval.
Caroline: And of course you try to hit as
00:50:49.990 --> 00:50:56.180
many molecules as possible. So this is
especially important when you do pictures
00:50:56.180 --> 00:50:59.760
of crystallized molecules because
crystallizing these molecules is a lot of
00:50:59.760 --> 00:51:04.480
work. You don't want to waste like 99
percent that just fall away and you never
00:51:04.480 --> 00:51:07.990
take snapshots of them.
Signal Angel: Thanks.
00:51:07.990 --> 00:51:10.990
Herald: So another question from
microphone 3.
00:51:10.990 --> 00:51:18.810
Microphone 3: How do you construct this
movie? I mean, for every pulse to have a
00:51:18.810 --> 00:51:25.240
new molecule and for every molecule is
oriented differently in space and has
00:51:25.240 --> 00:51:31.190
different oscillation modes. How do
correlate them? I mean, in the movie, I
00:51:31.190 --> 00:51:35.100
mean, every molecule is different than the
previous one.
00:51:35.100 --> 00:51:41.400
Caroline: Yes. Excellent question. That's,
So first, what people can do is align
00:51:41.400 --> 00:51:48.430
molecules. So especially molecules that
are more or less linear. You can force
00:51:48.430 --> 00:51:55.410
them to be oriented in a certain way. And
then there's also a bit of a secret in the
00:51:55.410 --> 00:52:00.220
trigger pulse that first sets off this
motion. For example, if this trigger pulse
00:52:00.220 --> 00:52:04.600
is a very strong proto ionization, then
this will kill off any sorts of
00:52:04.600 --> 00:52:08.910
vibrational states that you have had
before in the molecule. So in this sense,
00:52:08.910 --> 00:52:12.840
the trigger parts really defines a time
zero, that should be reproducible for any
00:52:12.840 --> 00:52:17.950
molecule that shows up in the stream and
the rest is statistics.
00:52:17.950 --> 00:52:23.580
Microphone 3: Thank you.
Herald: So there's another question on
00:52:23.580 --> 00:52:27.070
microphone 3.
Microphone 3: Are there any pre pulses or
00:52:27.070 --> 00:52:31.420
ghosts, You need to get rid of?
Caroline: Sorry. Again.
00:52:31.420 --> 00:52:36.550
Microphone 3: You have to control pre
pulses or ghosts during this effect for
00:52:36.550 --> 00:52:39.560
measurement.
Caroline: That I'm not really sure of,
00:52:39.560 --> 00:52:42.090
since I'm not really conducting
experiments, but probably.
00:52:42.090 --> 00:52:49.510
Herald: And another one from the middle,
from microphone 2 please.
00:52:49.510 --> 00:52:55.430
Microphone 2: I suppose if you apply for
experimentation time at the XFEL laser,
00:52:55.430 --> 00:53:02.000
you have to submit very detailed plans and
time lines and everything. And you will
00:53:02.000 --> 00:53:07.950
get the time window for your experiment, I
guess. what's going to happen if you're
00:53:07.950 --> 00:53:13.760
not completely finished within that time
window? Are they easy possibilities to
00:53:13.760 --> 00:53:19.180
extend the time or are they do they just
say, well, you had your three weeks,
00:53:19.180 --> 00:53:23.120
you're out apply in 2026?
Caroline: Yeah, I think it's a regular
00:53:23.120 --> 00:53:26.930
case, that you're not finished with your
experiments by the time your beam time
00:53:26.930 --> 00:53:31.870
ends. That's how it usually goes. It's
also unfortunatly not free weeks, but it's
00:53:31.870 --> 00:53:38.531
rather like 60 hours delivered in five
shifts of twelve hours. So, yeah, you
00:53:38.531 --> 00:53:43.250
write a very detailed proposal of what you
would like to do. Submit it to a panel of
00:53:43.250 --> 00:53:50.620
experts, both scientists and technicians.
So they decide, is it interesting enough
00:53:50.620 --> 00:53:54.980
from a scientific point of view and is it
feasible from a technical point of view?
00:53:54.980 --> 00:54:00.320
And then once you are there, you more or
less set up your experiment and do as much
00:54:00.320 --> 00:54:06.490
as you can. If you want to come back, you
need to submit an additional proposal. So,
00:54:06.490 --> 00:54:10.590
yeah, I think most experimental groups try
to have several of these proposals running
00:54:10.590 --> 00:54:15.490
at the same time, so that there is not a
two year delay between your data
00:54:15.490 --> 00:54:21.070
acquisition. But yes. No possibility to
extend. It's booked already for the
00:54:21.070 --> 00:54:27.890
complete next year. The schedule is fixed.
Herald: So I don't see any more people
00:54:27.890 --> 00:54:33.060
queuing up. If you want to pose a
question, please do so now. In the
00:54:33.060 --> 00:54:35.460
meantime, I would ask the signal angel if
there's another question from the
00:54:35.460 --> 00:54:38.240
Internet.
Signal Angel: I have a question about the
00:54:38.240 --> 00:54:44.650
dimensions of all those machines. The
undulator seems to be rather long and
00:54:44.650 --> 00:54:48.910
contain a lot of magnets. Do you have an
idea how long it is and how many of those
00:54:48.910 --> 00:54:55.221
electromagnets are in there?
Caroline: Yeah. Sorry, I didn't mention
00:54:55.221 --> 00:54:57.221
it. It's about, I think one hundred and
seventy meters long in the case of the
00:54:57.221 --> 00:55:03.390
European XFEL. I'm not sure about the
dimension of the individual magnets, but
00:55:03.390 --> 00:55:09.100
it's probably also in the hundreds of
magnets, magnet pairs.
00:55:09.100 --> 00:55:16.080
Herald: So is there more - excuse me,
there is a question on the microphone
00:55:16.080 --> 00:55:20.350
number 3.
Microphone 3: Yeah. Hi. It's regarding the
00:55:20.350 --> 00:55:22.350
harmonic light.
Herald: Please go closer to the
00:55:22.350 --> 00:55:24.350
microphone.
Microphone 3: The harmonic light generator
00:55:24.350 --> 00:55:26.870
that you were showing at the very
beginning, just before the one that won
00:55:26.870 --> 00:55:32.700
the Nobel Prize. And can you also produce
light in the visible range? Or it has to
00:55:32.700 --> 00:55:38.050
be in the visible range?
Caroline: The high harmonic generation? So
00:55:38.050 --> 00:55:43.160
in the in the visible range, you cannot
create pulses that are so short that they
00:55:43.160 --> 00:55:50.960
would be interesting for what I'm doing.
The pulse that comes in is already quite
00:55:50.960 --> 00:55:54.791
short. So it's already femtoseconds long.
They just convert it into something that
00:55:54.791 --> 00:56:01.850
is fractions of a femtosecond long. And
yeah. In the indivisible range that's kind
00:56:01.850 --> 00:56:05.800
of a limit how short your pulse can be.
Microphone 3: So it is not a good
00:56:05.800 --> 00:56:11.060
candidate for hyperspectral light source.
We need another kind of technique, I
00:56:11.060 --> 00:56:16.360
guess.
Caroline: Well, I mean you are kind of
00:56:16.360 --> 00:56:21.070
limited what short pulses you can generate
with which wavelength.
00:56:21.070 --> 00:56:26.441
Microphone 3: Thank you.
Herald: So again, a question to the Signal
00:56:26.441 --> 00:56:29.020
Angel. Are there more questions from the
internet?
00:56:29.020 --> 00:56:34.590
Signal Angel: Yes I have another one about
the lifetime of the molecules in the beam?
00:56:34.590 --> 00:56:39.800
How fast are they degrading or how fast
are they destructed?
00:56:39.800 --> 00:56:44.290
Caroline: So probably the question is
about how fast they are destructed before
00:56:44.290 --> 00:56:53.420
- so either before our pulses hit the
molecule the molecules should be stable
00:56:53.420 --> 00:56:57.330
enough to survive in the gas phase from
the point where they are evaporated until
00:56:57.330 --> 00:57:01.370
the point where the pump and the probe
pulse come together, because otherwise it
00:57:01.370 --> 00:57:07.560
doesn't make sense to study this molecule
in the gas phase. When the probe pulse
00:57:07.560 --> 00:57:12.690
hits and it flows apart, I guess pico
seconds until the whole molecules ...
00:57:12.690 --> 00:57:21.780
Microphone 3: Like instantaneous.
Herald: So I don't see any more questions
00:57:21.780 --> 00:57:26.040
on the microphones and we have a few
minutes left. So if there are more
00:57:26.040 --> 00:57:30.030
questions from the Internet, we can take
maybe one or two more.
00:57:30.030 --> 00:57:42.560
Signal Angel: Give me a second.
Herald: For people leaving already, please
00:57:42.560 --> 00:57:46.760
look if you have taken trash and bottles
with you.
00:57:46.760 --> 00:57:52.390
Signal Angel: So this one very, very
technical question. How do you compensate
00:57:52.390 --> 00:57:58.210
the electronic signal that the electronic
signal reaction is probably slower than x
00:57:58.210 --> 00:58:05.240
ray or light or spectrum changes at one
moment or at one particular moment, that
00:58:05.240 --> 00:58:10.170
was interesting to analyze. I do not
understand the question, though.
00:58:10.170 --> 00:58:13.240
Caroline: I think I understand the
question, but I don't have the answer
00:58:13.240 --> 00:58:16.860
because again, I'm a - that's the problem
of speaking to a technical audience, you
00:58:16.860 --> 00:58:23.710
get a lot of these very technical
question. Yeah. The data analysis is not
00:58:23.710 --> 00:58:32.810
instantaneous. So the data is transported
somewhere safe in my imagination. And then
00:58:32.810 --> 00:58:37.180
taken from there. So this data analysis
does not have to take place on the same
00:58:37.180 --> 00:58:41.330
timescale as the data acquisition, which I
guess is also because of the problem that
00:58:41.330 --> 00:58:47.530
was mentioned in the question.
Herald: I might interrupt here. Maybe it's
00:58:47.530 --> 00:58:55.680
also about the signal transmission, like
the signal rising of the signal of the
00:58:55.680 --> 00:59:01.900
electrical signal transmissions. Because
this would probably require bandwidths of
00:59:01.900 --> 00:59:06.680
several megahertz, gigahertz, I don't
know, to transport these very fast
00:59:06.680 --> 00:59:09.680
results.
Caroline: Yeah. I think that's also a
00:59:09.680 --> 00:59:15.130
problem of constructing the right
detector. That has been solved apparently
00:59:15.130 --> 00:59:20.690
because they can take these images. laughter
Herald: And on the other hand, we have 10
00:59:20.690 --> 00:59:28.760
gigabit either nets. So we get faster and
faster electronics. More questions from
00:59:28.760 --> 00:59:39.921
the Internet? Does not look like it, also
the time is running out. So let's thank
00:59:39.921 --> 00:59:42.760
Caroline for her marvelous talk. Applause, final music
00:59:42.760 --> 01:00:13.000
subtitles created by c3subtitles.de
in the year 2019. Join, and help us!