Max Planck Institute for the Structure and Dynamics of Matter- home CFEL home Condensed Matter Dynamics Department - MPSD/CFEL

Free electron lasers

In the future, we expect that extreme-timescale spectroscopy will make use of large-scale FEL sources reaching well beyond our original work and capability with tabletop laser-based sources. By many orders of magnitude, these are the brightest sources in the world. Furthermore, photon energies from the XUV to the soft-x-ray to the hard x-ray regimes can be accessed, with tunable pulse durations from near 100 femtoseconds in duration to as short as several femtoseconds, or below. In contrast, attosecond laser sources are currently limited to photon energies below ~150 eV.

However, while table-top attosecond laser pulses are inherently synchronized to a second pump laser pulse, used to trigger the dynamics, inherent synchronization is not yet possible at FELs. Therefore, to perform time-resolved experiments, the relative timing of the FEL x-ray pulses must first be measured or controlled on the sub-fs level. Furthermore, for few-fs or even attosecond FEL pulses, the precise temporal profile of each x-ray pulse, which is expected to be highly structured and vary randomly from shot to shot, must be determined to calculate the on-target x-ray intensity. → recent publication in Nature Photonics on the characterization of individual FEL pulses

Since 2009, we have participated in a collaborative effort to make single-shot measurements on the low-charge few-femtosecond x-ray pulses at LCLS. Over the first two experiment periods we have adapted techniques from attosecond metrology to characterize the x-ray pulses. Using NIR and IR dressing fields we have modulated the kinetic energy of a burst of photoelectrons emitted from a gas target by the x-ray pulse to map the photoemission temporal profile, which is assumed to be a replica of the x-ray pulse. Through these experiments, we have set an upper bound on the x-ray pulse duration of 4 femtoseconds and believe that we have seen evidence of single femtosecond substructure on the x-ray pulse profile.

We are now leading the pulse characterization collaboration and have been awarded additional beam time at LCLS in October 2011 to pursue more advanced streaking measurements. We have designed this experiment, and a second approach that is currently under review, to overcome the profound limitation of ~50 femtosecond timing jitter between the dressing laser pulse and the FEL x-ray pulse.

In the first of these experiments, dubbed “tandem streaking,” two streaking measurements are made in series on each x-ray pulse with a laser field that is identical in cycle-averaged amplitude but shifted by a quarter wave in phase. The phase shift will be achieved by utilizing the Gouy phase shift that all Gaussian beams undergo in a focus. The pair of measurements reveals the instantaneous dressing field parameters – which vary randomly from shot to shot – applied in that particular measurement. Only with this information can each single-shot spectra be mapped to its calibrated temporal profile. Our tandem streaking experiments are planned with 5µm IR laser which is critical because the x-ray pulse must be shorter than the dressing field half-cycle. A 5µm dressing field is suitable for measuring x-ray pulses to approximately 10 femtoseconds in duration, but not longer.

Our second approach utilizes high-intensity, single-cycle THz pulses. Here the THz field, produced by rectfication of Ti:Sapphire laser pulses in LiNbO3, has a rise time of approximately 500 fs, significantly longer than the timing jitter at LCLS. Therefore, the instant of temporal overlap and the instantaneous gradient of the streaking field can be determined from a single streaking measurement, rather than a phase-shifted pair, again resulting in a self-calibrated measurement. A further advantage of this technique is that it would allow for characterization of x-ray pulses up to and exceeding 100 femtoseconds due to the increased half-cycle duration. However, increased dynamic range in this measurement is achieved at the expense of attosecond resolution and it is unlikely that these measurements will reveal substructure on the sub-femtosecond timescale.

While pursuing measurements with hard x-rays at LCLS we are also planning to make similar measurements during the summer of 2011 on soft x-ray pulses at FLASH at DESY. Here, the same timing jitter limitations exist, complicating x-ray pulse characterization as well as use of this facility for time-resolved studies. Our experience at FLASH will be valuable for future work at LCLS and will provide a basis for continuing these efforts at the planned FLASH II and hard x-ray European XFEL facilities.

Finally, we have even more ambitious plans to overcome the problems of timing jitter and to improve the temporal resolution of FEL sources by controlling their emission with few cycle laser pulses. Here, the few-cycle laser pulse will be used to modulate the energy of the electron bunch that controls the FEL emission. In this way it is possible to produce well-behaved single-peaked FEL photon pulses with duration limited by the cooperation length of the FEL process. Because the FEL pulses are sliced by an external laser, the emission would be inherently synchronized to an external pump laser source, eliminating one of the barriers to full utilization of these user facilities.

We are currently working with the accelerator division at DESY to simulate the machine parameters of FLASH to determine what few-cycle laser pulse is required for beam manipulation and are planning the necessary physical infrastructure to couple this light into the machine. At FLASH where the cooperation length is ~5 femtoseconds, attosecond pulses cannot be expected, however, these proof-of-principle experiments will lead to future work at the hard x-ray FELS where the cooperation length is in the few-hundred attosecond regime. By modulating the energy of the driving electron bunch, we would be able to achieve a synchronized and shaped attosecond FEL pulse across the full spectral range for time-resolved studies.