High-power two-cycle ultrafast source based on hybrid nonlinear compression

: We demonstrate a hybrid dual-stage nonlinear compression scheme, which allows the temporal compression of 330 fs-pulses down to 6.8 fs-pulses, with an overall transmission of 61%. This high transmission is obtained by using a first compression stage based on a gas-filled multipass cell, and a second stage based on a large-core gas-filled capillary. The source output is fully characterized in terms of spectral, temporal, spatial, and short- and long-term stability properties. The system’s compactness, stability, and high average power makes it ideally suited to drive high photon flux XUV sources through high harmonic generation.

nm), which is typically required in combination with gating techniques to obtain isolated attosecond pulses, two stages of compression must usually be implemented [12]. This reduces the energy efficiency of Yb-based systems dedicated to attosecond physics.
In this article, we describe a two-stage nonlinear compression setup that provides enough compression ratio to reach sub-10 fs pulse duration from a high-energy ytterbium-doped fiber amplifier (YDFA) laser, while ensuring the highest transmission (61%) ever reported for two cascaded stages. This results in the generation of 6.8 fs, 140 µJ pulses at 150 kHz repetition rate, corresponding to 21 W average power. These performances are achieved by combining two nonlinear compression technologies, first a gas-filled multipass cell, and second a large diameter-core capillary. The described laser system is robust, compact, and power efficient, making it an ideal driver laser for application-ready high flux XUV and attosecond sources.

Rationale
Gas-filled capillaries are the most widespread nonlinear media used to temporally compress femtosecond pulses with energies above 100 µJ. Propagation in this lossy waveguide imparts spectral broadening through the self-phase modulation (SPM) effect along with a frequency chirp that can be removed by dispersive optics at the output of the capillary. The capillary diameter choice is bounded by two phenomena: it should be large enough to avoid significant ionization of the gas, and small enough to induce sufficient SPM, which directly translates to the compression ratio. The gas nature and pressure inside the capillary is limited by the threshold for self-focusing, which does not depend on the diameter. Moreover, for practical reasons and to ensure a reasonable footprint, the capillary length is often of the order of 1 m. As a consequence, for pulse durations of 300 fs and energies between 100 µJ and 1 mJ, as is standard at the output of YDFA systems, a capillary diameter of 250 µm is often used to obtain a compression ratio around 10. In a laboratory environment where longer capillaries with larger diameters can be used, a recent experiment demonstrates a compression ratio of 33 in a 6-m-long 500-µm-diameter capillary with a transmission of 70% [13].
On the other hand, the losses introduced by a capillary are related to two parameters: the spatial quality of the input laser, which determines the fraction of energy that can be coupled to the fundamental mode exhibiting the lowest losses, and the ratio of diameter to central wavelength, which determines the losses of each capillary mode. YDFA systems often exhibit close-to-perfect spatial quality, so that the losses are dominated by the capillary losses. Experimentally obtained transmission factors for such setups at the output of YDFA are in the range of 60% (single stage) to 30% (dual-stage) [12,14].
To increase this transmission, and allow energy scaling of compression setups, a recently demonstrated technique consists in propagating the pulses to be compressed in a multipass cell (MPC) that includes a nonlinear medium [15]. The MPC is formed by an arrangement of concave mirrors and the input beam is matched to the stationary beam in the cell. As it propagates through a large number of roundtrips, the input pulse is periodically focused. If the nonlinearity per roundtrip is kept sufficiently small, the spatial Kerr effect is redistributed over the whole beam, ensuring high spatial quality at the output, despite a possibly large accumulated temporal B-integral [16]. This technique can be considered as an extension of multi-plate setups [17][18][19] with a distribution of the nonlinearity over tens of passes in the material instead of few, inducing a better output spatial quality and allowing higher compression factors. Compared to capillaries, this technique provides several additional degrees of freedom in terms of geometry, nonlinear material used (solid [15,20] or gas [21,22]), and spectral phase control through the mirror coatings. The most obvious improvement is that, using commercially available mirrors, the transmission of the cell is above 90% in all reported demonstrations. One drawback of cell-based setups to reach very short pulse durations, however, is that the pulses undergo a large number of reflections on the cell mirrors (several tens), so that spectral phase aberrations induced by the mirrors are magnified. The design requirements on the mirrors in terms of bandwidth, spectral phase control, reflec duration decre We theref filled MPC to YDFA output as high as 85% on the MPC compared to diameter capi The use of a requirement f and allow acc 61%.

Experime
The experime with an YDF corresponding output of the second-harmo along with the We also in increase the p expands beyo measure the s spanning mor limit of this sp 6. Additional char two orthogonal ax : 0.5 W zoom arou utput, stability near processes rms of beam p am pointing de less than 2% o tted in Fig. 6