Post-compression schemes have made a rapid development in recent years through the advent of multipass cells (MPCs), enabling benchmark performances with kilowatt average power, more than 100 mJ of pulse energy and few-cycle pulse durations in the near infrared spectral region (NIR) [1–3]. The multipass platform allowed the existing kilowatt-scalable ytterbium-based amplifiers [4–6] to efficiently reach the 30 fs and shorter pulse duration regime which was previously led by Ti:sapphire (Ti:Sa) lasers. While the broad gain bandwidth of the Ti:Sa systems allows for sub-30 fs pulses from the laser system itself, the average power is generally limited to no more than a few 10s of watts [7]. Even though, these highly capable NIR sources are now more readily available, transferring these capabilities to the visible spectral region (VIS) would hugely benefit applications such as the machining of wide bandgap materials [8] and the generation of high-harmonic radiation (HHG) [9]. To capitalize on the highly developed ytterbium-laser architectures, an efficiently driveable nonlinear upconversion process is a clear choice. Second-harmonic generation (SHG) being one of the most favorable options for its simplicity, enabling the conversion to the green spectral range.

HHG in particular can benefit greatly from a driving source in the green spectral region delivering ultrashort pulses at high repetition rates, since the process struggles with notoriously low conversion efficiencies, typically below $10^{-5}$ [10,11]. While a higher repetition rate, thus a higher average power, of the driving laser improves the extreme ultraviolet (XUV) flux linearly, a change in driving wavelength allows for scaling of the single atom response in the HHG process by $\lambda ^{-6}$ [12]. Thus, transferring the driving laser from the NIR to the VIS opens the possibility for an increase of multiple orders of magnitude of XUV flux. Additionally shorter driving pulses enable phase-matching at higher intensities, resulting in an increased cutoff energy and efficiency [13]. Besides its multitude of applications such as coherent diffractive imaging [14], the investigation of electron dynamics in matter [15] and spectroscopic analyses [16–18], the extreme ultraviolet spectral region at 92 eV has garnered interest in the past few years for its use in lithography [19]. At wavelength metrology applications, (e.g., high-resolution imaging [20]) driven by HHG in the otherwise hardly accessible spectral range would greatly benefit from an increase in photon flux to decrease acquisition times and improve the signal-to-noise ratio.

Extending on the presented developments in laser post-compression through multipass cells in the NIR and the opportunities for high average, high peak power ultrafast VIS-sources, the next step to follow is their wavelength transfer. Even though one of the most remarkable features of nonlinear multipass cells is their simplicity, the wavelength transfer is not as straightforward. Multipass cells fundamentally rely on the capabilities of the employed cavity mirrors to achieve large bandwidths, i.e., ultra-short pulses, support high average powers and high efficiencies, all while maintaining a compact footprint.

Advancing to shorter wavelengths generally increases both linear and nonlinear absorption, severely limiting average power capabilities. A more careful selection of optics based on their material composition is required, and dispersive mirrors which play a vital role in post-compression schemes are particularly prone to multiphoton absorption [21].

Therefore, only one MPC demonstration in the green spectral range exists to date, reporting approximately 2 W of average power with 15 µJ of pulse energy [22]. So far, hollow-core fiber (HCF)-based compressions represent the state-of-the-art. The HCF-based compression of Klas et al. [9] delivered 18.6 fs with 51 µJ at 51 W of average power; Xia et al. [23] achieved 160 µJ at 30 fs and Decamps et al. [24] 14.8 fs with 69 µJ at 11.5 W of average power with an overall efficiency of around 20 %.

In this Letter, we demonstrate the generation of ultrashort pulses in the green spectral range utilizing a MPC-based post-compression delivering 15.7 fs full width at half maximum (FWHM) pulses with a pulse energy of 0.44 mJ at an average power of 22.4 W. The presented system is able to deliver mJ-class pulse energy at high repetition rates with an overall conversion efficiency from the IR driving laser to the compressed green pulses of more than 40 %; two times more than previously demonstrated HCF compression techniques.

The experimental setup is depicted in Fig. 1. The experiments were carried out utilizing an ytterbium-based fiber laser system emitting 280 fs pulses at a repetition rate of 50.8 kHz with an average output power of 55 W, corresponding to an energy of approximately 1.08 mJ. A 1.5 mm-long beta-barium borate (BBO) crystal cut for type-I phase matching is used to convert the collimated laser beam with a diameter of 5.6 mm ($1/e^2$) to the green spectral range via SHG. After separating the fundamental beam via two dichroic mirrors, 240 fs pulses with an average power of 29 W of second-harmonic radiation are accessible, corresponding to an SHG-efficiency of more than 52 %.

 

Fig. 1. Schematic of the experimental setup. Pulses generated from an Yb:fiber laser are frequency-doubled in a BBO crystal and spectrally broadened in a krypton-filled multipass cell. The temporal recompression to 15.7 fs is realized with dispersive mirrors (DM) and fused silica.

Download Full Size | PDF

The light is guided into a sealed chamber, which can handle rough vacuum and absolute pressures up to 1.3 bar, with two fused silica windows containing a krypton atmosphere of 0.6 bar. The enclosed Herriot-type MPC consists of two low dispersion quarter-wave stack dielectric mirrors with a radius of curvature (ROC) of 1 m and a diameter of 50 mm with a separation of approximately 1.95 m. The calculated Gaussian eigenmode has a $1/e^2$ diameter of 0.32 mm in the focus and 2.03 mm on the mirrors. Matching the collimated beam to the linear eigenmode of the MPC and recollimating the beam after passing through the cell is realized by two concave–convex mirror pairs. The beam is coupled in and out of the cavity via two scraper mirrors with dimensions of $10\times {10}\,\textrm{mm}$ on opposite sites of the cavity resulting in 19 focal passes. Gas type and pressure are chosen such that the bandwidth of the dielectric mirrors can be fully utilized through SPM introduced spectral broadening at the given laser input parameters, while maintaining a pressure $\leq$ 1 atm to avoid a more demanding overpressure setup.

A half-wave plate (HWP) in front of the MPC is used to rotate the polarization slightly to achieve the most ideal perpendicular polarization to the table plane which is maintained throughout the MPC. Without this correction we observed an undesirable elliptical polarization state at the output, which vanished when the cell was evacuated. We believe the consecutive reflections of non-0$^\circ$ mirrors and the out-of-plane propagation in the MPC introduce a weak elliptical polarization state which is enhanced due to the accumulated nonlinear phase shift resulting in nonlinear ellipse rotation [25].

The cell was designed such that the occurrence of ionization in the focal region can be avoided, which otherwise would introduce additional losses and could impair the beam quality. Choosing the appropriate cell length, i.e., a large enough radius of curvature for the cavity mirrors is the most straightforward option to achieve the desired focal intensity, as ROC and focal intensity are related by an inverse proportionality [26]. To estimate the onset of detrimental ionization effects, we adopted the criterion presented by Vozzi et al. in [27]. For a given set of laser parameters it allows for the calculation of a minimum beam radius, relating to a maximum intensity, for which ionization effects still remain negligible. Instead of the originally used ADK-rates, which cover only the tunnel ionization regime [28], we calculated the beam averaged ionization via the instantaneous YI-ionization rates which are applicable to arbitrary Keldysh parameters [29]. This is a necessity as the increase in photon energy in the VIS compared with the NIR results in an increased probability for multiphoton absorption. Simply using the formula presented in Ref. [27] results in a critical radius for ionization 2.5 times smaller than the calculated MPC eigenmode, while our calculation yields a difference of only 7 %, implying that presented cell is very close to the ionization limit but does not breach it. This is in line with the excellent cell transmission being close to 95 %, which can be directly linked to the linear reflective losses of the employed optics.

Finally the pulses are recompressed in time by the reflecting of 28 custom dispersive mirrors designed to provide −100 fs² group-delay dispersion (GDD) per reflection, over the spectral range from 490 to 540 nm. Additionally 4 mm of fused silica is introduced to fine-tune the dispersion and achieve the shortest pulse duration possible. Including the fused silica, this amounts to a total GDD of approximately −2525 fs². The compression stage has a transmission of 81%. After compression, 22.4 W remain, corresponding to a pulse energy of 0.44 mJ.

A fraction of the post-compressed pulses is split off by two fused silica wedges to carry out an M$^2$-measurement revealing a high beam quality of $1.19 \times 1.17$ (see Fig. 2).

 

Fig. 2. M$^2$ measurement of the compressed beam at 22.4 W in compliance with ISO 11146. Inset: intensity profile of the focus.

Download Full Size | PDF

The temporal profile of the pulse is characterized by guiding a sample of the beam to a commercially available (SourceLab) TIPTOE device (tunneling ionization with a perturbation for the time-domain observation of an electric field) [30]. Figure 3 shows the TIPTOE retrieved pulse and spectrum in conjunction with a spectrum, measured by a grating-based spectrometer and the corresponding calculated Fourier-limited pulse. The TIPTOE data shows good agreement between measured and retrieved spectrum, reporting a pulse duration of 15.7 fs, which is within 5 % of the Fourier-limited pulse duration (14.9 fs). The retrieved pulse shows a high temporal contrast. More than 93 % of the energy is contained in the main feature, resulting in a peak power of 24.9 GW.

 

Fig. 3. Temporal characterization of the compressed output pulses via the TIPTOE technique. (a) Retrieved temporal pulse profile and Fourier limit calculated from the measured spectrum in (b); both are normalized to the output pulse energy. (b) Measured and retrieved spectrum and the corresponding spectral phase.

Download Full Size | PDF

In summary, a multipass-cell-based post-compression in the green spectral range delivering 15.7 fs pulses, 0.44 mJ pulse energy at 22.4 W of average power was demonstrated. The 280 fs pulses of the Yb:fiber driving laser served as an ideal platform providing a narrow enough bandwidth to allow for a high SHG efficiency of more than 52 %, while producing a short enough pulse to reach sub-10 optical cycles in the green spectral range with a single multipass cell compression stage. This enables a high overall conversion efficiency of more than 40 % from the IR driving laser to the compressed green pulses, a twofold increase compared with state-of-the-art systems. The compressed output of the cell delivers pulses with an estimated peak power of close to 25 GW with a high beam quality of M$^2 <1.2$. Based on the here-presented results we believe that it will be possible to scale this concept efficiently into the 100 W average power-class and beyond. Developing capable post-compression schemes in the VIS spectral range will be transformative for the development of HHG-based XUV-sources and consecutive applications, due to the significant increase in the single-atom response compared with the NIR spectral range. More than 10 mW of HHG radiation at 26.5 eV in single harmonic line have been demonstrated utilizing a HCF-based compression delivering 19 fs pulses at 51 W of average power at 515 nm [9]. In recent years more than 10 kW of average power from an ultrafast fiber laser system has been demonstrated [6] as well as more than 1 kW of average power at 515 nm through a frequency doubled picosecond Yb-laser [31]. In conjunction with the highly average power scalable MPC platform the presented approach will be able to deliver the first Watt-class HHG table-top source at around 30 eV and boost the photon flux to unprecedented levels at the important 92 eV photon energy (13.5 nm wavelength). In a preliminary study, utilizing the presented sub-20 fs 515 nm source as the HHG driver and helium as the interaction medium, we were able to push the cutoff energy beyond 100 eV. A phase-matched conversion efficiency of $10^{-8}$ at the 39th harmonic line (93.9 eV) was achieved, which is comparable to the state-of-the-art high flux HHG systems around 90 eV driven by 1 µm wavelength ultrafast lasers [32]. Through optimization of the generation conditions (e.g., optimized gas target and focusing geometry), we are expecting a further improvement of up to two orders of magnitude in future experiments. Thus, pushing the photon flux to 100 µW per harmonic line around 90 eV appears to be in reach with a moderate increase of driving laser average power at 515 nm to 100 W, which would represent a flux improvement of more than two orders of magnitude compared with state-of-the-art systems [32,33].