Nobel Prize lecture: Anne L'Huillier, Nobel Prize in Physics 2023

Scientists won the Nobel Prize
12 followers
1 Video View·Jun 27, 2025
Honored Nobel laureates, your excellencies, distinguished colleagues, ladies and gentlemen,
a warm welcome to the Nobel Lectures in Physics. And the Nobel Prize in Physics this year
is about attosecond physics. And attosecond physics is about very, very, very small time scales.
So small that the number of attoseconds in one second is the same number as the second since
our universe was born. Even though the time scale is so small, what is taking place on that small
time scale is affecting our daily life because it determines properties in many different aspects.
Attosecond physics reveals the life of the electrons and their dynamics. And through
the knowledge of what is happening on that small time scale and with the electrons, we get the
ability and the knowledge to make new tailored materials. We also can design electronics using
new concepts. And we can develop methods for medicine of most importance for human mankind.
And our laureates, they discovered that it was possible to generate these very short
light pulses. They developed methods to measure the duration of the light pulses.
And they also developed methods for generating pulse trains and also individually isolated light
pulses of the attosecond duration. And our Nobel laureates in physics we will see here.
And I get some help to switch to the next slide. And while we're figuring out how to go there,
so our Nobel laureates in physics this year is Pierre Agostini, Ferenc Kraus and Anne Lollier.
And it is my pleasure now to introduce our first speaker. And this is Anne Lollier. So please
join me in welcoming Anne. So good morning. It's a pleasure to be here. And let me start by
thanking the Royal Swedish Academy of Sciences and the Nobel Foundation for this prize and
for the pleasure, the privilege to give this Nobel lecture. So I will tell you about how we
arrived to this attosecond pulses. This is a history of attosecond science with the first years,
progress in technology and in understanding, measurement of attosecond pulses and then
opening to research fields and applications. And I will go through the first two parts and leave
part three to my co-laureates, measurement of attosecond pulses. I will in the end just
say a few words about some applications. Nothing would have happened without the invention of the
laser in 1960 by Theodor Meinmann. And the Nobel prize in physics was given to the background for
laser in 1964 to Towns, Bazov and Prokhorov. And you have here a picture of one of the first lasers,
a ruby laser. Now what we are going to use in our research, the properties of the laser
are first the possibility to generate short laser pulses and especially the possibility
to by focusing these pulses to achieve a very high intensity.
So the invention of the laser led to new research fields and the first one is non-linear optics.
This is a picture of an experiment performed by Peter Franken in 1961 immediately after the
invention of the laser. And you see a ruby laser which is focused by a lens into a material and
our experiment will look very much the same. And then the radiation is analyzed by a prism
and detected by a detector. In this process an atom absorbs two photons, it's a non-linear optical
process, and emits a photon at frequency two omega. This is second harmonic generation,
the first non-linear optical processes that was discovered.
Now another research field that started after the invention of the laser is atoms in strong fields
or multi-photon processes that were predicted already in 1934 by Maria Gopertmeier.
And in multi-photon processes an atom absorbs several photons simultaneously. It can be excited,
it can be ionized. In 1979 Pierre Agostini discovered the process of above threshold
ionization where an atom is ionized by absorbing more photon that is necessary for ionization.
And this is detected by observing electrons and one observes several peaks.
Now an atom can also be multiply ionized and this was actually my thesis work at the beginning of
the 80s. Now at that time the question arised, is it possible to produce excited atoms or excited
ions that then would decay by emitting fluorescence light? So we set up an experiment,
this was in France at the Commissariat de l'Energie Atomique. We set up an experiment
to observe this fluorescence light. Fluorescence light is emitted in all directions, so we had
two axes of detection, one along the direction of the propagation axis of the laser
with a grating and a detector and one perpendicular to it. And this is the laser that was used at that
time, a Neodymium YAG laser, one micrometer wavelength and pulse duration was 40 picoseconds.
You have an oscillator here and two amplifiers. And this is a picture of the experimental setup.
You have a laser is coming from the right, you have a lens focusing the laser into a chamber
and the radiation is detected along the propagation axis with a grating and a detector
and perpendicular to it. The intensity reached in the medium, so here we have also a pulsed jet
delivering some gas of atoms. So the intensity in the medium reached several times, 10 to the 13
watts per square centimeter, which is quite high. What did we see? Well, we never saw
anything perpendicular to the detection axis, so I'm not going to discuss that anymore.
And we didn't see much fluorescence. What we saw was very high order harmonics of the laser field
and this is one of the first spectra obtained in argon gas. And what you see here are harmonics 13,
15 up to 31st harmonic of the laser field. So when it says 13 here, that means that the frequency
is equal to 13 times the laser frequency. Now if we plot the intensity distribution of these
high harmonics, we get the typical behavior with the decreased plateau behavior where all the
harmonics have the same intensity and this is ending with a cutoff. And this was very unexpected.
First, it was not expected to see so many high order harmonics. Then, it was not expected to
have this plateau behavior. One thought that it would be more and more difficult to generate
high order harmonics. This was based on some kind of perturbation theory thinking at the time,
that based on an expansion of the polarization induced in the medium, in powers of the electric
field, for this expansion to be valid, one has to assume that the first order is much larger than
the third and much larger than the fifth. And this was not the case. You can notice that I'm only
talking about odd orders of the laser and this is due to a symmetry property of the medium. It has
the inversion symmetry, so the even order harmonics are actually killed. Now if you go from a
frequency of the laser light which is in the infrared region and you multiply this frequency
by let's say 33, you end up in the extreme ultraviolet region of the spectrum. It's below x-rays
but above ultraviolet light. So this light is invisible and it does not propagate in air, so we
have to evacuate. Now let's make a little bit analogy with music here. When we think about harmonics,
usually think about music. What we do is actually very similar to what happens when the bow of a
violin overtones and this is what actually makes the clang, the timbre of a music instrument, is how many of
these overtones we have. In our experiment we produce not harmonics of sound but harmonics of light
and the violin is the gas of atoms and the bow of the violin is the laser pulse.
Now this process requires at the border of two fields, atomic physics and non-linear optics.
Atomic physics is the description of the response of an atom to the laser field
and this is, we are in the quantum world, this is described by a Schrodinger equation that you have
here, where you have here the atomic potential and the laser atom interaction. In this problem
we can describe the laser field classically and we use the dipole approximation.
Now the second aspect is non-linear optics, the response of many atoms to radiation.
It's not enough that each atom generates these high harmonics, you also need to have efficient
harmonic generation to have the emission from all of the atoms in the medium in phase
at coherently and this is described by Maxwell's equations of electromagnetism
which can simplify to a wave equation as what is written here. E here is the generated field
in the medium and p is the polarization in the medium induced by the laser field.
So the first attempt, let me say one more thing, again analogy to music, this many atom response,
you can think of a gigantic orchestra, all of the musicians in this orchestra need to play with the
same pace and the number of atoms in this orchestra is trillions of atoms, many more than
the number of people on earth, just to give you a little bit an order of magnitude. Now the first
attempt to describe these processes was to simply solve these two equations which I just wrote and
these are calculations of this single atom response solving numerically the Schrodinger equation
and you see that the agreement between theory and experiment is quite good.
In itself it's a bit surprising because at that time we thought that the many atom response
would be very different for the different harmonic order, it would be more and more difficult to
phase match the process. So it was really necessary to also solve the second equation,
the wave equation, and this is what was done here. You see that we obtain a rather good agreement
again with the experimental behavior and we can reproduce this plateau and cutoff behavior.
So we can simulate the experiment but we don't really understand the physics behind.
Approximately at the same time was asked the question could it be that these high order
harmonics form in the time domain a series, a train of attosecond pulses and the idea is the
following. Consider let's say three harmonics 17, 19 and 21 and assume that these harmonics
are all in phase at a certain time, they are phase locked. If they are in phase synchronized
at a certain time they will be again in phase after half the laser cycle. So now if you sum
these three waves you see that we get constructive interference when they are in phase and
destructive interference where they are not in phase. We can confine the light into short pulses
and if we plot the intensity corresponding to this electric field you get a train of
attosecond pulses. If you sum not only three harmonics but let's say 10 harmonics and you
assume they are phase locked you obtain a train of very short light pulse. Why is it interesting?
Because the evolution of the pulse duration of lasers had come to a stop. This is the evolution
of the pulse duration of dye lasers decreasing from picoseconds down to a few femtoseconds and
a couple of years later titanium sapphire laser would also arrive at the stop of about two
femtoseconds. The reason for this stop is that we are talking about light pulses
and to talk about a light pulse you need to have at least one optical period in this pulse and
the optical period of for example a titanium sapphire laser is 2.6 femtoseconds. So in order to
decrease the pulse duration you need to decrease the optical period which means to go to a higher
frequencies. Now the second condition to get short pulses is that you need a broad bandwidth.
This is just an expression of Heisenberg principle to get a short time you need a broad bandwidth
and this high order harmonics provided both higher frequencies in the extreme ultraviolet
and the broad bandwidth this plateau behavior. So this was a very exciting question.
So now I'm coming to the second part of this talk which is progress in understanding and progress
in technology and I would like to start by progress in technology. At the beginning of the 90s there
was a huge progress in laser technology. First with the chirp pulse amplification which led to
a Nobel Prize to Gérard Moreau and Donna Strickland in 2018. There was the discovery of a new material
titanium sapphire by Peter Moulton and then a new way to generate short pulses in the laser
oscillator which was Kerr-Lenz mode locking. And here you have three swedish scientists
Sune Svanberg, Anders Persson, Klaus-Johan Wallström that look at this new laser technology
and they decide to really invest into this new laser technology. This was very risky because
it was not clear that it would work but they really invest into this technology and build the
high power laser facility in Lund with a terawatt titanium sapphire laser. This was really
the perfect laser to study high order harmonic generation with a short duration of the order of
100 femtosecond and at that time a high repetition rate 10 light laser shots per second, 10 hertz.
In parallel to that there was also a progress in instrumentation. In Saclay we build a
dedicated instrument to study high order harmonic generation which you see here
and these instruments include gas jet optics to analyze the radiation and a detector. And here you
have myself much younger but together with Philippe Balcou and Pascal Salier, my first
PhD students, we are here in the audience today. And this is also the time where personal computers
came into the laboratory. The first experiment that I mentioned we were using a printer to record
the signal. Now personal computers started to really come into the lab and we could,
thanks to these computers, really take data in a very completely different way.
So this led to very nice experiments combining this dedicated instrumentation
and this new laser system titanium sapphire. And I show here one of the first spectrum obtained in
NEON using this laser by a group in Stanford and there you see a beautiful harmonic spectrum
with a very large plateau here. This decrease at the beginning is simply due to the decrease
in efficiency of the grating and you have the cutoff in this region. So you see that we go to
the 111th harmonic. And in Lund, together with the collaboration Lund and Saclay, we could not only
take harmonic spectra but really study these harmonics as a function of different parameters.
For example here as a function of the laser intensity. Now approximately at the same time
came really a breakthrough in the theoretical understanding of this process with the three-step
model. So I would like to say here that this prize is given to the experimental method in
Atos against Science but the theorists have played a very very big role in Atos against Science and
in particular in the understanding and this three-step model. And this was shown
approximately at the same time by Ken Cullander and Ken Schaefer, by Paul Corkum in Canada,
and Maciej Lewenstein and co-workers here, Misha Ivanov, presented a quantum formulation of this
three-step model called the strong field approximation which was going to have a very
strong impact in the field. So this three-step model is presented here. What happens is that you
have an atom in a strong laser field and this laser field is really strong. It distorts the atomic
potential. It bends this potential and here you have a barrier. So an electron in the ground state,
this is quantum mechanics, there is a certain probability for the electrons to tunnel through
this barrier. Then the electron is driven away by the laser field. The laser field changes sign,
the electron is pushed back towards the atom and then there is a certain probability
of recombination. In this excursion the electron has gained some kinetic energy and the excess
energy is emitted in the form of a high energy photon. And this process is repeated every half
laser cycle on one side and the other side and again and again leading to emission of light
every half cycle of the laser. Now what about attosecond pulses? I just remind you that we have
discussed that if we had the phase-locked harmonics in the time domain we would have
one pulse per half cycle. So what was this new theoretical understanding telling us?
And here is the bad surprise. This was the result of the calculation. Light everywhere,
no one pulse per cycle, so no phase-locked harmonics. So the attosecond idea at that time
was quite controversial and not so clear. How can we understand this multiple pulse emission?
Actually we can understand very simply using this three-step model and looking at the possible
electron trajectory. This is just a very simple picture of these trajectories, electron going out
and coming back. I put in colors the energy that the electron have when they return. So you have
higher and higher energy depending on when the electron is born. But there are also trajectories
which takes longer time and leads to a similar type of energy. So when you look at the possible
trajectory you find that there are short trajectory and long trajectory leading to the
same kinetic energy. And this complex picture leads to this complex result in the time domain.
Now we could actually verify the existence of these trajectories in the laboratory. This is
an experiment performed in Lund in 1998 and you have some of the participants here.
Mette Gorder who is in the audience, Ted Hensch, Marco Bellini and others.
And what we did in this experiment was to look at the coherence of this high harmonic. Is it
coherent light? And to do that we duplicate the harmonic sources and we look at whether
this light interferes in the far field. If this is coherent we should have interferences.
And indeed you see here on the 15th harmonic that there are beautiful interference fringes.
What you can also see in this picture is that there are now two contributions. There is collimated
radiation and more divergent contribution which we could identify to the contribution from the
short trajectory and from the long trajectory. Now we could also delay one of the sources
relative to the other and look at what happens to the interference fringes. And you see that
in this case the interferences are still present for the collimated radiation while they are gone
for the more divergent contribution which shows different properties of these two contributions,
short and long trajectory. What happens to the other side? Strong field non-linear optics,
the response of many atoms. There is also progress in this part as well. So to achieve
fast matching here we need the phase of the pulse which is generated to be equal to the
phase of the polarization induced in the medium. So we need these two wave vectors here to be equal.
And this difference in wave vector depends on the dispersion in the partially ionized medium,
depends on the laser focusing, but it also depends, and that's the important thing,
on the electron trajectory whether it's short or long. And this gave us the idea or maybe in some
condition only one of these trajectories can survive, only one of these trajectories get phase matched.
And indeed this is what we got. In dashed line you have the single atom response
while in blue you have the temporal structure of the total attosecond pulse train. So we
recover one attosecond pulse per laser high cycle and phase locked harmonics.
So at this point we think that, or maybe we have attosecond pulses, but what is very important
is to measure attosecond pulses. Before I come to the conclusion of this talk let me point out that
at the end of the 90s there is a lot of progress both in laser and in high order harmonic
technology. And let me just flash some of this progress. First now we can generate laser pulses
that are very short, few cycles, and then the importance of the carrier envelope phase is
pointed out. And this carrier envelope phase is the phase offset between the electric field and
the envelope. And this was going to be very important for the generation of single attosecond pulses.
Then you have novel medium geometries like cell or capillaries. You have also new
media that are being explored, molecules. You have a push toward higher photon energy
using short laser pulse and especially long wavelength lasers. And you have also a push
toward higher pulse energy in the microjoule range using clever focusing geometries.
So now I am going to jump over this part three and I would like to finish this presentation by
just giving you two examples of applications. Very different examples. One is very fundamental.
This is about electron dynamics and one is much more applied. So electron dynamics. I think that's
exciting thing about attosecond pulses. We can now follow the dynamics of electrons in matter.
And here I take one example which is very simple. It's photoionization, a process that was understood
theoretically by Albert Einstein in 1905. And he actually got the Nobel Prize just for the
discovery of the law of the photoelectric effect. So what we can do, and I'm going to talk about
photoionization of atoms. What we can do thanks to attosecond pulse is we can now answer a question.
How long time it takes for an electron to propagate in the continuum? What are the
wave quantum properties of the photoelectron? And I'm not going to explain how we do these
measurements. I just want to try to make you understand what we can do again by using an
analogy with music. In music you keep the pace by the conductor is keeping the pace
or the metronome can be used to keep the pace and the metronome has a motion like this.
So this is a wave motion, both the conductor and the metronome, which is due to the mechanical
motion of the tip of the metronome. And this is exactly what we do. We are looking at oscillations
such as this one here, and we are going to compare the position of the maxima of this
oscillation and this oscillation. Here I'm talking about photoionization of neon atoms
in the 2s shell, which I'm comparing to photoionization of neon in the 2p shell.
And what do we measure? It's always interesting to talk about time by showing a clock.
In our measurements, one hour on a clock is one half period of the laser, which is 1.3 femtosecond.
So one hour is here 1300 attosecond. And what we can measure very accurately is the position
of two arrows on this clock, which correspond to the time it takes for the electron to ionize from
the 2s shell and from the 2p shell. We don't know where these two clocks are on this clock, but
we can measure the difference in time between these arrows. And now I'm showing some
results, showing that indeed we can measure this relative time quite accurately within
tenths of attosecond. The second application I would like to mention is really going into a
completely different direction now, very applied, and this is now belonging not to the research
laboratories anymore, but to industry. And this is an application not so much about the temporal
structure of this high-order harmonics, not so much about attosecond path, but about simply an
application of this radiation, which is in the extreme ultraviolet range and which is very broad
in frequency. So now harmonic generations are being used to control wafers, silicon wafers,
which contain integrated circuits. And this technique is used for the next generation
instruments that will control this integrated circuit, which is very, very small dimension
of the order of 10 nanometer or below. And this control is made not only in 2D, but in 3D.
And what is really used here is the fact that we have so many frequencies, and this is developed by
an industry, ASML in Holland. Now to conclude, I would like to thank many people, all of my
colleagues during all of these years, and I'm showing here some group pictures, not since 1987,
just the last 15 years or so. I would like to also thank the organizations that have funded this
research, the Swedish Science Council, Swedish Research Council, and the European Research Council,
the Wallenberg Foundation, and last but not least, I would like to
thank my family, Klaus-Joran, Oskar and Viktor. Thank you for your attention.
