It was the first paper that conclusively demonstrated the possibility to generate pulses shorter than a femtosecond. The generation of such pulses was considered an important quest ever since it was theoretically predicted that this might be possible, since pulses this short could allow a more than 10-fold improvement in a physicist's ability to do time-resolved measurements.

  Does it describe a new discovery or new methodology that's useful to others?

Yes it does; the described way of generating and characterizing the attosecond pulses is sufficiently simple that it could be used as a source delivering pulses for further time-resolved studies in many other areas of physics or chemistry.

  Can you give us some background on this research?

In 1988 it was discovered in Saclay that focusing an infrared laser into a gas jet led to the production of very high harmonics (of odd order) of the driving laser radiation. These harmonics are in the vacuum UV / soft X-ray range (XUV) of the electro-magnetic spectrum. Around 1993, by the work of Paul Corkum from NRC, Ottawa, it became understood that the harmonics are produced because photoelectrons, that are first pulled from the gas atoms by the laser, are accelerated by that same laser to high velocities, after which they can recollide with the ion that they originally left behind. In such a recollission the electron can give up its kinetic energy as radiation, after which it is recaptured by the atom.

Recollision with sufficient energy to generate XUV light can occur only during certain parts of the laser cycle, and this led theorists (P. Antoine and M. Lewenstein) to predict that the XUV would come in bursts much shorter than an optical period of the driving laser. This effect was predicted to persist in the total emission from a huge number of atoms as present in the laser focus.

The labs of P. Agostini (CEA, Saclay), H.G. Muller (FOM, Amsterdam) and Ph. Balcou (ENSTA, Palaiseau) have a long-standing (EC-funded) cooperation for studying the behavior of atoms subjected to short, intense XUV pulses in combination with laser light. Previous results from this cooperation encompass the discovery of 'Laser-Assisted Auger Decay', XUV pulse-duration measurements by cross-correlation techniques and the development of a novel high-order cross-correlation method dubbed 'ponderomotive streaking' that allowed characterization of XUV pulses as short as 10 fs. In 2000 we realized that the setup developed for the ponderomotive-streaking experiment would be capable of resolving the time structure of the XUV pulse with attosecond resolution, provided we could perform it with interferometrically stable beams. (i.e. path lengths traversed by the various beams should fluctuate significantly less than a wavelength for the duration of the experiment). We put the experiment on our backlog, because the probability for success seemed remote, and gave priority to the ponderomotive streaking. In one of our measurement sessions, the laser did not perform well enough to do the ponderomotive streaking (which required high intensity), and to avoid wasting beam time we tried to measure the attosecond pulses. Once we got to analyze the data, the results were above expectation, and led to the Science publication.

  Could you summarize the significance of your paper in layman's terms?

The significance of our paper is that we devised and developed a method to measure the time structure of ultra-violet light pulses with an unprecedented time resolution. This allows us to reveal details that are only a small fraction of the duration of the cycle time of a vibration of ordinary light. Using this method (which we now call RABBITT, for 'Resolution of Attosecond Beating by Interference of Two-photon Transitions') we could demonstrate that ultra-violet light generated from a laser in a gas jet indeed has a very pronounced structure on this time scale: it comes as a sequence of strong bursts, separated by time intervals where there is almost no light at all. Such trains of pulses could be used to study very fast phenomena, like the motion of electrons in molecules, atoms or solids.

The method itself makes use of 'beating' between various 'colors' of ultra-violet light. If something (like an atom) is subjected to two different vibrations of different frequency simultaneously, the vibrations sometimes enhance each other, but at other times they oppose each other and their effects cancel. The combination of the vibration periodically alternates these times of strong and weak action. This alternation is known as 'beating', and piano tuners can use it to compare the pitch of the piano string to that of a tuning fork (the beating stops if the pitches are equal).

The effect of ultra-violet light on atoms is that it kicks out electrons. The electrons only appear when the different kinds of UV light enhance each other. In our method the electrons, once out of the atom, are shoved around by the laser, but in a way that depends on the timing of their appearance compared to that of the laser vibration. This allows us to measure during which part of the laser vibration the UV colors enhance each other, and we could do that for each pair of UV colors present in our pulse. It turned out that they all enhance each other at one particular time, and this leads to a strong, ultra-short pulse of XUV light.

Prof. Dr. H.G. Muller,
FOM-Institute for Atomic and Molecular Physics
Kruislaan 407,
1098 SJ Amsterdam, 
The Netherlands