Measurements on the attosecond timescale had been limited to the dynamics of electrons in an atomic gas. But a record has now been set in a quite different context — the photoemission of electrons from a surface.
The quest for faster and faster time-resolved measurements has reached a new level: Cavalieri et al. report (page 1029 of this issue)1 that they have measured a delay of 100 attoseconds in the emission of electrons ejected from a surface irradiated by light. This is not just the experiment with the best time resolution yet; it is also the first time that attosecond metrology has been applied to a solid, rather than a gaseous, system.
It was only in the 1990s that the trend to ever faster measurements produced laser sources with pulse durations below 5 femtoseconds (a femtosecond is 10−15 seconds). This is the timescale of the motion of atoms within molecules. Femtochemistry, in which a chemical reaction is followed through its transition state, became big news2. But that is now old hat. The attosecond (10−18 seconds) is the timescale of the motion of electrons within atoms: an electron takes about 150 attoseconds to orbit a hydrogen atom.
Attosecond pulses are created when intense laser pulses of femtosecond duration are focused into a gas sample. A process known as high-harmonic generation3 then kicks in to produce light at a range of frequencies that are precisely phased together, creating a train of very short, coherent pulses. In the past few years, the technology has evolved to the point where single pulses just 130 attoseconds long can be produced with tabletop-sized laser systems4. These pulses are so short that their frequency (and thus energy) lies in the extreme-ultraviolet or soft-X-ray portion of the electromagnetic spectrum.
Attosecond metrology has previously been applied to samples of atomic gases to observe excitation processes of electrons such as shake-up and Auger decay5. These are essentially 'pump–probe' measurements: an attosecond pulse excites the system, and the intense optical laser field that generated the pulse follows it and is used to sweep up the charged products — much as an oscilloscope streaks an electron beam across the screen to resolve an electrical pulse.
Cavalieri et al.1 focus their 90-electronvolt extreme-ultraviolet pulse at an angle on a tungsten metal surface (Fig. 1, overleaf). The lower-frequency optical pulse that created the attosecond pulse follows along the same path, but its passage can be delayed in steps of 300 attoseconds. Electrons liberated through the photoelectric effect by the first pulse are detected by a spectrometer that measures their kinetic energy. The optical laser field pushes these photoelectrons' energy up or down, depending on the precise position of both the attosecond pulse and the photoelectron in the laser field's cycle.
By varying the time delay between the pulse and the optical field, and measuring the shift in the up and down motion of the energy spectrum, the authors could precisely measure the emission time of the photoelectrons. They were able to distinguish electrons coming from different energy states in the surface, observing that electrons from the more deeply bound core states in the surface were emitted around 100 attoseconds after those from the conduction band.
The authors' great triumph is to apply the streaking technique to a metal surface, rather than a gas sample. This required an optical field of intermediate strength — intense enough to modify the energy of the photoelectrons significantly, but not so intense as to create photoelectrons directly.
When it comes to sources of soft X-rays of high brightness, synchrotrons still reign supreme. But the pulse duration of synchrotron light is in the picosecond range, and thus much too slow to observe the dynamics of fast electrons. Next-generation free-electron-laser facilities, such as FLASH at the DESY research centre in Hamburg, Germany, have pulse durations of around 50 femtoseconds, and will allow images to be obtained of molecular structures during chemical reactions, or protein structures to be determined6. The final kilometre of the SLAC linear accelerator at Stanford University is also being converted into a source of femtosecond X-ray pulses, the Linac Coherent Light Source, or LCLS.
But quite apart from the reduced cost, attosecond pulses produced from tabletop lasers have one significant advantage over these very expensive pulsed-X-ray facilities — they are exquisitely synchronized with the laser pulse that produces them. In our facility in Ottawa, we have found that attosecond pulses produced in two separate sources by the same laser are synchronized to even better than an attosecond — in the zeptosecond (10−21 second) range. By contrast, free-electron sources have significant timing 'jitter' that limits their resolution in pump–probe experiments. And, as any physicist living out of a suitcase will tell you, working in one's own laboratory is far preferable to travelling to a distant facility.
Cavalieri, A. L. et al. Nature 449, 1029–1032 (2007).
Zewail, A. H. http://nobelprize.org/nobel_prizes/chemistry/laureates/1999/zewail-lecture.html (1999).
Scrinzi, A., Ivanov, M. Yu., Kienberger, R. & Villeneuve, D. M. J. Phys. B 39, R1–R37 (2006).
Sansone, G. et al. Science 314, 443–446 (2006).
Uiberacker, M. et al. Nature 446, 627–632 (2007).
Neutze, R. et al. Nature 406, 752–757 (2000).