We present the extension of high-power pulse compression deeper into the challenging IR spectral range around 3.2 μm wavelength, where the effects of material absorption, dispersion, and free electron disturbance on nonlinear propagation become increasingly limiting parameters. 5 mJ, 80 fs pulses from a KTA parametric amplifier were spectrally broadened in a large-core hollow fiber with argon as the nonlinear medium. Subsequent compression through anomalous dispersion in yielded 2.5 mJ close-to-transform-limited two-cycle pulses exhibiting a passively stabilized carrier envelope phase (CEP). Furthermore, we outline the feasibility of generating sub-two-cycle pulses with good spatial and temporal characteristics.
© 2016 Optical Society of America
High-energy few-cycle pulses in the mid-wave IR (mid-IR) spectral range are attractive for a number of high field applications, e.g., high-order harmonic generation, and allow us to extend the cutoff to the kiloelectron volt range due to scaling. A scalable and efficient way to generate femtosecond mid-IR pulses in the 3–4 μm spectral range is via optical parametric amplification . KTA is an especially attractive nonlinear crystal that allows amplification of pulses in the 3–4 μm spectral range when pumped by a high-power ytterbium laser centered at 1.03 μm and supports pulses typically of around 80 fs. However, there is great demand for quasi-monocycle pulses, which directly enable the generation of isolated attosecond pulses. Here, we extend a pulse post-compression technique based on spectral broadening in hollow-core fiber (HCF) [2,3] to the mid-IR spectral range and show that high transmission through a HCF, despite strong cladding absorption, can be achieved by choosing an appropriate core diameter. Previously, such pulse post-compression down to a few-cycle duration was successfully demonstrated for wavelengths as long as 1.8 μm, where a bulk material was used to compensate the phase of the pulse after spectral broadening caused by self-phase modulation (SPM) in the capillary . Recompression of positively chirped pulses in the mid-IR range can be simply achieved by passing the beam through a bulk material, as most of the materials exhibit anomalous dispersion in this spectral range . This is a significant advantage compared to pulse compression in the near-IR range, where specially designed dispersive mirrors or/and a prism compressor is needed.
One of the key bottlenecks hindering the extension of the technique to the mid-wave IR mid-IR spectral range is the increasing loss that scales rapidly with wavelength as . In this Letter we experimentally show that such losses can be offset when an appropriately large-core waveguide is used and present 5 mJ, 3.2 μm pulse post-compression in a large diameter 3 m long HCF.
The schematic of the optical parametric amplifier (OPA) source and the pulse post-compression setup are depicted in Fig. 1. The 3.2 μm input pulses with energy up to 5 mJ and 80 fs duration are generated in a KTA parametric amplifier pumped by a 200 fs, 120 mJ multi-pass amplifier. The seed pulses for the OPA centered at 1.5 μm are generated directly from 1.03 μm pump pulses via supercontinuum generation in a YAG plate. The 200 fs pulses allow implementation of a simple OPA scheme and passive carrier envelope phase (CEP) stabilization  of the idler pulses at 3.2 μm. In order to achieve Fourier-limited-duration pulses at the input of the HCF, the chirp due to the dispersion in the optical elements after the OPA ( dichroic optics, focusing lens, gas cell windows) was pre-compensated using a positive dispersion of a 5 mm silicon plate in the seed before the last amplification stage.
For the optimal coupling into the 1 mm core diameter HCF (few-cycle Inc.), the 3.2 μm input beam is loosely focused using a lens so that the waist diameter is , as shown in Fig. 1(c), and the input/output windows were placed far enough from the capillary to avoid unwanted SPM in them (). A large-core HCF allows us to reduce the linear propagation losses that depend critically on the waveguide core diameter , as depicted in Fig. 1(b). For comparison, the transmission of a 200 μm, 1 m long capillary is above 80% at 800 nm; however, such a HCF is not suitable at a 3.2 μm wavelength range due to very high losses. In contrast, a 1 mm core diameter 3 m long HCF has over 90% transmission at 3.2 μm wavelength, and the transmission is still sufficiently high for wavelengths as long as 6 μm, even when the material absorption in the cladding is taken into account (see Section 3 of Supplement 1 for details). The use of the large-core HCF also allows us to avoid the ionization losses caused by the reduced intensity, . The reduction of the nonlinear phase shift can be compensated by increasing the HCF length , where is the wavenumber and is the intensity. The strength of the nonlinearity can be changed by varying the argon filling pressure at the HCF output: .
The straightness of the waveguide becomes a critical parameter with the increase in the core diameter, as it leads to (i) increased probability of linear mode coupling; (ii) increased attenuation due to bending losses, , where is the radius of curvature; and (iii) higher losses of the fundamental mode as compared to higher order modes in the presence of bending of the fiber. However, maintaining HCF straightness for a multi-meter-long waveguide becomes technically challenging. We employ a recently developed stretched waveguide technique  using two specially designed mounts that keep the fiber straight over this long length and allow differential pumping. The flexible HCF has a 1 mm inner diameter and is 3 m long, with 300 μm thick fused silica cladding surrounded by a polymer layer. Fiber holders allow us to change the HCF length depending on the given experimental conditions, and tension applied to both ends of the fiber keeps it straight [8,9].
The key advantage of pulse post-compression in HCF is the uniform pulse compression across the beam and the high-quality beam profile at the output due to spatial filtering effects, as shown in Fig. S6 in Supplement 1. The use of a long wavelength allows us to scale the hollow-core post-compression scheme to much higher pulse energy and peak power pulses because of the favorable scaling of the critical power for self-focusing, . The measured spectral evolution of 3.2 μm pulses at different filling gas pressures was measured using a mid-IR spectrometer based on an acousto-optic dispersive filter (Fastlite) and is summarized in Fig. 2.
First, the pulses at the output were measured under evacuated fiber (no spectral broadening) and approximately correspond to the 85 fs input pulse. After increasing argon gas pressure to 1 bar, a significant spectral broadening is observed. The optimal SPM nonlinearity that corresponds to the chirp of that compensates a 2 mm window is given in Fig. 4(c). We find that the optimal argon pressure for pulse compression and chirp compensation in a 2 mm window is 1.3 bars. Temporal characterization of post-compressed pulses using second-harmonic generation frequency-resolved optical gating (SHG-FROG) is shown in Fig. 3 and indicates that the compression down to a two-cycle duration was achieved at 1.3 bars argon filling gas pressure. The corresponding spectrum and group delay dispersion (GDD) retrieved from SHG-FROG measurement are shown in Fig. 2(c). A shorter compressed pulse duration could be achieved at higher gas pressures in combination with a thinner window; however, the pulse duration measured here was limited by the geometric smearing in the SHG-FROG apparatus (see Supplement 1 for details). Using a pulse characterization method free of geometric smearing  will allow us to measure even shorter pulses in the sub-two-cycle regime. The graph in Fig. 2(b) shows the calculated Fourier limit of the pulse duration from the measured spectra at different filling gas pressures and suggests that a spectral bandwidth close to a single optical cycle is achievable.
The total throughput of the HCF was above 50% and did not depend on the filling gas pressure even for the high input pulse energies, which is attributed to the large-core fiber. Most of the losses are due to coupling and distortions of the input beam profile. The large core of the capillary allows us to keep the peak intensity at the output of the fiber at , which does not cause any significant ionization of the argon gas used for spectral broadening. Furthermore, the peak power of 22 GW is more than a factor of 6 below the critical power for self-focusing at this long wavelength (at 1.3 bars argon pressure, ). After compressing the pulse in the 2 mm output window down to 22 fs, the peak power is increased up to . The attractive property of the HCF pulse compression in the IR is that the chirp induced via spectral broadening can be compensated using bulk material dispersion. Most of the optical materials transparent in the 2–4 μm range exhibit anomalous dispersion. The material of choice for our experiments in the 3–4 μm spectral range is . Figure 4(c) shows the dependence of the optimal thickness of as a function of fiber filling gas pressure under our experimental conditions. Figures 4(a) and 4(b) compare the case of pulse compression at 3.2 μm with that at 1.8 μm as described in , where the 4-mm-thick fused silica (FS) plate at 1.8 μm exhibits similar GDD to the 2 mm plate used in the present experiment.
Here we analytically estimate the GDD () and third-order dispersion (TOD) () needed to compensate the phase of the spectrally broadened pulses and show that pulse compression close to a single optical cycle is possible in the mid-IR range. Calculations  show that the self-steepening effect yields the opposite sign of the TOD phase and the control of input pulse TOD can be used to minimize the residual TOD further.
The desired compression ratio ( is the input pulse duration and is the Fourier-limited duration of the compressed pulse) determines the required amount of nonlinear phase shift induced by the SPM in the HCF. The nonlinear phase shift, in turn, defines the GDD of the output pulses: . As illustrated in Fig. 4(c), the required thickness for GDD compensation decreases with increased HCF filling gas pressure. For example, the GDD of 2 mm is and corresponds to the chirp due to SPM at nonlinear phase shift. For the given nonlinear phase shift , the resulting compression ratio can be calculated for the case of perfect phase compensation,
In the sub-two-optical-cycle regime, the TOD dispersion compensation becomes critical and has to be taken into account in the estimations. has a similar GDD/TOD ratio at 3.2 μm to that of FS at 1.8 μm. It has been shown that self-steepening helps achieving pulse compression to a short pulse duration, as it induces a negative third-order component on the spectrally broadened pulse, which partially compensates the positive TOD of the material [4,11]. Figure 4(d) shows that the residual TOD increases with increasing pressure, which is one of the limiting factors of the pulse post-compression. The TOD could be reduced further by introducing some positive TOD on the input pulse , as described in Supplement 1. The self-steepening effect yields negative TOD of the spectrally broadened output pulses, which partially compensates the material dispersion, as illustrated in Fig. 4(d). Moreover, a larger amount of TOD can be tolerated at longer wavelengths because of the longer cycle duration and the pulse envelope broadening due to uncompensated TOD scales as . For a given number of cycles , the bandwidth-limited pulse of duration is proportional to the wavelength. See Supplement 1 for the calculations of pulse envelope stretching dependence on TOD and wavelength.
The passive CEP stabilization of the 3.2 μm pulses is achieved by using a femtosecond laser for seeding and pumping the OPA, as illustrated in the Fig. 1(a) optical scheme. The seed at 1500 nm is obtained via supercontinuum generation in a YAG plate, which is subsequently amplified, and the passively CEP-stable idler is obtained at 3200 nm . The idler is then amplified in the subsequent two amplification stages.
The measurement at the output of the HCF was performed in order to verify the CEP stability of the OPA source and to show that the CEP is preserved during the SPM spectral broadening in the HCF . The spectral fringes in the interferogram shown in Fig. 5 depict the CEP of the spectrally broadened pulses. The CEP of the input pulses was modulated using a saw-tooth profile in order to demonstrate the CEP control capability. This measurement confirms that the OPA idler pulses are CEP stable and the phase is preserved after SPM broadening in the capillary.
In conclusion, we report on multi-millijoule 3.2 μm two-cycle pulse post-compression via spectral broadening in a 1 mm core diameter 3 m long argon-filled hollow-core capillary with transmission. The spectrally broadened pulses were compressed in a bulk material down to 22 fs. At higher filling gas pressures, a pulse bandwidth supporting a Fourier-limited pulse duration below 15 fs was obtained, which corresponds to a sub-two-cycle pulse at this wavelength. The mid-IR input pulses are generated in a femtosecond OPA based on KTA crystals, which delivers 5 mJ, 80 fs, 3.2 μm CEP-stable idler pulses. We show that the losses of the waveguide at long wavelengths can be minimized when an appropriately large-core HCF is used, despite the absorption of the HCF cladding material. Remarkably, in simulations we find that material absorption is not an ultimate limiting factor due to the mode guiding through grazing incidence reflection, and we suggest that the HCF post-compression technique can be applied for wavelengths up to 6 μm with high transmission.
Austrian Science Fund (FWF) (P 27577-N27, SFB NextLite F4903-N23); Natural Sciences and Engineering Research Council of Canada (NSERC); Fonds de Recherche du Québec—Nature et Technologies (FRQNT); Canada Foundation for Innovation (CFI); Marketing Science Institute (MSI); Ministère de l’Économie Science et Innovation (MESI).
See Supplement 1 for supporting content.
1. G. Andriukaitis, T. Balčiūnas, S. Ališauskas, A. Pugžlys, A. Baltuška, T. Popmintchev, M.-C. Chen, M. M. Murnane, and H. C. Kapteyn, Opt. Lett. 36, 2755 (2011). [CrossRef]
2. M. Nisoli, S. DeSilvestri, and O. Svelto, Appl. Phys. Lett. 68, 2793 (1996). [CrossRef]
3. S. Bohman, A. Suda, T. Kanai, S. Yamaguchi, and K. Midorikawa, Opt. Lett. 35, 1887 (2010). [CrossRef]
4. B. E. Schmidt, P. Béjot, M. Giguère, A. D. Shiner, C. Trallero-Herrero, E. Bisson, J. Kasparian, J.-P. Wolf, D. M. Villeneuve, J.-C. Kieffer, P. B. Corkum, and F. Légaré, Appl. Phys. Lett. 96, 121109 (2010). [CrossRef]
5. P. Béjot, B. E. Schmidt, J. Kasparian, J.-P. Wolf, and F. Legaré, Phys. Rev. A 81, 063828 (2010). [CrossRef]
6. Z. Huang, D. Wang, Y. Dai, Y. Li, X. Guo, W. Li, Y. Chen, J. Lu, Z. Liu, R. Zhao, and Y. Leng, Opt. Express 24, 9280 (2016). [CrossRef]
7. A. Baltuška, T. Fuji, and T. Kobayashi, Phys. Rev. Lett. 88, 133901 (2002). [CrossRef]
8. T. Nagy, V. Pervak, and P. Simon, Opt. Lett. 36, 4422 (2011). [CrossRef]
9. V. Cardin, N. Thiré, S. Beaulieu, V. Wanie, F. Légaré, and B. E. Schmidt, Appl. Phys. Lett. 107, 181101 (2015). [CrossRef]
10. G. Fan, T. Balciunas, C. Fourcade-Dutin, S. Haessler, A. Voronin, A. M. Zheltikov, F. Gérôme, F. Benabid, A. Baltuska, and T. Witting, Opt. Express 24, 12713 (2016). [CrossRef]
11. A. Suda and T. Takeda, Appl. Sci. 2, 549 (2012). [CrossRef]
12. A. Baltuška, M. Uiberacker, E. Goulielmakis, R. Kienberger, V. S. Yakovlev, T. Udem, T. W. Hänsch, and F. Krausz, IEEE J. Sel. Top. Quantum Electron. 9, 972 (2003). [CrossRef]
13. F. Lücking, A. Trabattoni, S. Anumula, G. Sansone, F. Calegari, M. Nisoli, T. Oksenhendler, and G. Tempea, Opt. Lett. 39, 2302 (2014). [CrossRef]