Controlled few-cycle light waveforms find numerous applications in attosecond science, most notably the production of isolated attosecond pulses in the XUV spectral region for studying ultrafast electronic processes in matter. Scaling up the pulse energy of few-cycle pulses could extend the scope of applications to even higher intensity processes, such as the generation of attosecond pulses with extreme brightness from relativistic plasma mirrors. Hollow-fiber compressors are widely used to produce few-cycle pulses with excellent spatiotemporal quality, whereby octave-spanning broadened spectra can be temporally compressed to near-single-cycle duration. In order to scale up the peak power of hollow-fiber compressors, the effective length and area mode of the fiber has to be increased proportionally, thereby requiring the use of longer waveguides with larger apertures. Thanks to an innovative design utilizing stretched flexible capillaries, we show that a stretched hollow-fiber compressor can generate pulses of TW peak power, the duration of which can be continuously tuned from the input seed laser pulse duration down to almost a single cycle (3.5fs at 750nm central wavelength) simply by increasing the gas pressure at the fiber end. The pulses are characterized online using an integrated d-scan device directly under vacuum. While the pulse duration and chirp are tuned, all other pulse characteristics, such as energy, pointing stability and focal distribution remain the same on target. This unique device makes it possible to explore the generation of high-energy attosecond XUV pulses from plasma mirrors using controllable relativistic-intensity light waveforms at 1kHz.
Laser-plasma accelerators are usually driven by 100-TW class laser systems with rather low repetition rates. However, recent years have seen the emergence of laser-plasma accelerators operating with kHz lasers and energies lower than 10 mJ. The high repetition-rate is particularly interesting for applications requiring high stability and high signal-to-noise ratio but lower energy electrons. For example, our group recently demonstrated that kHz laser-driven electron beams could be used to capture ultrafast structural dynamics in Silicon nano-membranes via electron diffraction with picosecond resolution. In these first experiments, electrons were injected in the density gradients located at the plasma exit, resulting in rather low energies in the 100 keV range. The electrons being nonrelativistic, the bunch duration quickly becomes picosecond long. Relativistic energies are required to mitigate space charge effects and maintain femtosecond bunches.
In this paper, we will show very recent results where electrons are accelerated in laser-driven wakefields to relativistic energies, reaching up to 5 MeV at kHz repetition rate. The electron energy was increased by nearly two orders of magnitude by using single-cycle laser pulses of 3.5 fs, with only 2.5 mJ of energy. Using such short pulses of light allowed us to resonantly excite high amplitude and nonlinear plasma waves at high plasma density, ne=1.5-2×1020 cm-3, in a regime close to the blow-out regime. Electrons had a peaked distribution around 5 MeV, with a relative energy spread of ~30 %. Charges in the 100’s fC/shot and up to pC/shot where measured depending on plasma density. The electron beam was fairly collimated, ~20 mrad divergence at Full Width Half Maximum. The results show remarkable stability of the beam parameters in terms of beam pointing and electron distribution. 3D PIC simulations reproduce the results very well and indicate that electrons are injected by the ionization of Nitrogen atoms, N5+ to N6+, leading to the formation of an electron bunch of 1 fs duration.
The interaction of single-cycle pulses with the plasma also leads to new physical effects. We have observed experimental evidence that plasma dispersion cannot be neglected in this regime. This is due to the extremely broad bandwidth of the laser, extending from 400 nm to 1000 nm, and to the high electron density. Therefore, the acceleration process is optimal when small positive chirps are introduced: the negative dispersion of the plasma then causes the re-compression of the laser pulse inside the plasma. Simulations indicate that this help localizing the injection process, leading to single femtosecond electron bunch.
Such a kHz femtosecond electron source will pave to way to numerous innovative applications, such as sub-10 fs electron diffraction, radiolysis of water with unprecedented resolution or the generation of femtosecond X-ray at kHz.