We demonstrate a scheme for nonlinear pulse compression at high average powers based on self-phase modulation in a multi-pass cell using fused silica as the nonlinear medium. The scheme is suitable for compression of pulses with peak powers exceeding the threshold for critical self-focusing. At of input power, the pulses of a Yb:YAG-Innoslab laser system (10 MHz repetition rate, 850 fs pulse duration) are spectrally broadened from 1.6 to bandwidth while maintaining almost diffraction-limited beam quality. The chirp is removed with a dispersive mirror compressor, and pulse durations of 170 fs at an output power of 375 W are achieved. The compression unit reaches an overall transmission of .
© 2016 Optical Society of America
Aside from applications in science and research, ultrafast lasers have recently seen a spark of interest in material processing. To maximize efficiency and throughput for industrial micromachining applications, high average output powers combined with pulse energies in the 10–100 μJ range are required. At the same time, ever shorter pulse durations are favorable to increase the precision and enable novel manufacturing mechanisms based on multi-photon-absorption or filament formation [1,2]. Ultrafast lasers with average powers of several 100 W have been demonstrated based on Yb:YAG [3–5] and Yb-doped fibers [6,7], but the pulse durations of these laser systems are limited to several 100 fs due to the gain bandwidths of the laser media.
A route to shorter pulse durations is nonlinear pulse compression, which is commonly used in high-power laser systems (Fig. 1). When propagating through a nonlinear medium, a laser pulse is spectrally broadened due to self-phase modulation (SPM) based on the optical Kerr effect. After removing the chirp of the spectrally broadened pulse, a temporally compressed pulse can be obtained. SPM is accompanied by self-focusing due to the spatial action of the Kerr effect, which can lead to a degradation of beam quality. If the peak power exceeds a material-dependent threshold, catastrophic self-focusing can occur.
Nonlinear pulse compression setups for high average powers can be distinguished according to the nonlinear medium for spectral broadening, with different media being suitable for different pulse energy and peak power regimes, respectively.
Dielectrics—most prominently fused silica—have been used as nonlinear media for spectral broadening in the form of both active [8,9] as well as passive waveguides [7,10–12]. These schemes are limited by the onset of catastrophic self-focusing, which occurs at peak powers of for fused silica and linearly polarized light. The highest average power so far compressed with a dielectric nonlinear medium is 250 W (23 fs compressed pulse duration, 69% compression efficiency) and has been obtained using a passive large-mode-area photonic crystal fiber .
Pulses with higher pulse energies can be compressed using noble gases as the nonlinear medium, which exhibit about three orders of magnitude higher self-focusing thresholds. In this context, most experiments have been conducted with waveguide-based spectral broadening in gas-filled hollow-core capillaries [13–15]. Owing to the significantly lower nonlinearity and the practical dimensions of these capillaries, this approach is—in the context of input pulses with multi-100 fs durations—typically used for input pulse energies exceeding 100 μJ. The highest average power compressed with this approach is 400 W ( output pulse duration, 58% compression efficiency, pulse energy) .
The aforementioned pulse compression schemes leave a pulse energy gap in the 10–100 μJ range (for sub-ps input pulse duration), in which pulse compression with neither scheme is possible. In recent years, much effort has been made to close this gap both by upscaling dielectric-based schemes and by downscaling noble gas-based schemes.
With respect to the latter, one promising approach is pulse compression using gas-filled hollow-core Kagome photonic crystal fibers, which allow for smaller diameters than capillaries and have been used to demonstrate nonlinear pulse compression in the aforementioned energy gap [16–18]. Spectral broadening at 118 W of output power (18 μJ output pulse energy, compression of 8.4% of the power to 88 fs)  and—in a different experiment—a compressed output power of 76 W (7 μJ output energy, 31 fs, compression efficiency)  have been demonstrated. These results indicate the average power scalability of this approach.
With upscaling dielectric-based compression schemes, concepts have been introduced in which prior to spectral broadening a pulse is split into several copies with peak powers below the threshold for catastrophic self-focusing, which are coherently combined after the nonlinear medium [19,20]. Compressed 7.5 μJ pulses (71 fs pulse duration, 750 mW average output power) have been achieved , but the average power scalability of this approach has not yet been demonstrated.
Alternatively, nonlinear spectral broadening in a single bulk dielectric instead of waveguides has been proposed to compress pulses with peak powers exceeding the threshold for critical self-focusing . In this case, the length of the nonlinear medium, as well as the input beam diameter and divergence require precise adjustment to avoid catastrophic self-focusing. Experimentally, the compression of 250 fs pulses to 43 fs at an average input power of 50 W and a pulse energy of 1.38 μJ has been demonstrated , but the scheme suffers from beam quality degradation and spatially inhomogeneous spectral broadening due to the spatial action of the Kerr effect, which necessitates spatial filtering and inherently limits the compression efficiency to .
In this Letter, we report on a novel, bulk dielectric-based scheme for nonlinear pulse compression in the 10–100 μJ pulse energy range, which mitigates the disadvantages of spectral broadening in a single bulk medium, and features unprecedented robustness and average power scaling capabilities. Instead of coupling to a single bulk nonlinear medium or conventional waveguide, laser pulses are repeatedly propagated through a thin bulk nonlinear medium by means of a multi-pass cell (MPC). Each pass through the nonlinear medium corresponds to a -Integral and is followed by a long propagation without nonlinearity. In this way, pulses with peak powers above the threshold for critical self-focusing can be spectrally broadened and beam quality degradation during propagation through the nonlinear medium is mitigated. The nonlinear phase accumulates during multiple passes so that total -Integrals and significant spectral broadening can be achieved. The MPC can be designed to have sufficient clear apertures to make the setup insensitive to alignment, pointing, and beam quality of the input beam. Aside from limitations due to the MPC’s dimensions, this approach is limited only by laser-induced damage of the nonlinear medium, the MPC optics, and their coatings. Using antireflective (AR)-coated fused silica as nonlinear medium, and dielectric- coated mirrors as MPC and compressor optics, this approach suits pulse compression of ultrafast laser sources with kW-level average power.
To demonstrate this approach, we use a Herriott-type MPC with two concave-convex mirrors of curvature and diameter 50 mm (Fig. 2). The concave faces are oriented toward the center of the MPC and AR-coated, while the convex faces are highly reflective (HR)-coated. In this way, the mirror substrates can be used as the nonlinear medium. Here, fused silica substrates of thickness 13 mm are used. To compensate for the substrate material dispersion, the HR mirror coatings are dispersive with a GDD of per bounce. The mirror distance is chosen to be , i.e., in the absence of Kerr-lensing, the mode has a beam radius at the curved mirrors of 460 μm. The beam is coupled in and out of the MPC using two right-angle fused silica prisms, which are placed close to the MPC mirrors at opposite sides of the MPC. The coupling is adjusted, such that the ray bundle has a circular shape on the MPC mirrors, and the beam reflects 18 times off each mirror during a complete pass through the MPC. Therefore, including the propagation through the in- and out-coupling prisms, a complete pass through the MPC comprises 38 passes through nonlinear media with a total propagation length of in the nonlinear media and in free-space.
The MPC is operated with a two-stage Yb:YAG-Innoslab amplifier system seeded by a commercial Yb-doped fiber-based oscillator . The system comprises a spatial filtering module to improve beam quality . It provides up to 530 W of output power at 10 MHz repetition rate with pulse durations of 850 fs (FWHM-bandwidth 1.6 nm) and almost diffraction-limited beam quality (). The amplifier output is mode-matched to the MPC using three spherical lenses. Using the transmission through one of the MPC mirrors, the beam bundle at this mirror is imaged to a camera. Thus, the beam radii after each roundtrip can be measured to optimize mode-matching. After coupling the beam out of the MPC, it is collimated with two spherical lenses. With an aperture, the output beam is separated from residual reflections from the AR-coated surfaces of the MPC mirrors.
Up to 500 W of average power and 50 μJ pulse energy are coupled to the MPC. Based on the experimentally determined beam-radii at maximum power, we estimate a maximum -integral of . The output power of the MPC increases linearly with input power [Fig. 3(a)], and a maximum transmitted power of 446 W is reached. The transmission is 91%. This transmission exceeds the transmission of typical waveguide-based spectral broadening units. A complete pass through the MPC comprises 72 passes through AR-coated mirror surfaces and 36 reflections from HR-coated mirror surfaces. Therefore, the losses of the MPC are dominated by linear losses due to the coatings of the MPC optics. The power incident to the MPC can be varied without the need to realign the MPC beam path. The mode-matching to the MPC changes slightly when changing the input power, due to Kerr-lensing. Moreover, we observe localized heating of the mirror substrates, presumably due to absorption in the dispersive HR coatings. The maximum temperature increase is at an input power of and results in thermal lensing, which affects the mode-matching comparable to the Kerr-lensing. For all measurements presented here, mode-matching as well as the collimating optics at the MPC output have been optimized for output powers of 375 W, and are not adapted when changing the input power. By measuring and fitting a caustic, the output -parameter after the MPC collimating optics is determined for different input powers to the MPC. Based on these -parameters, we calculate a mode overlap with the reference output mode obtained at 354 W of output power [Fig. 3(b)], which changes by less than 6% over the operating range of the MPC. Thus, the output -parameter of the MPC is barely affected by the input power, respectively, of Kerr and thermal lensing in the nonlinear elements. This is a significant advantage compared to spectral broadening with a single pass through the bulk nonlinear medium, where the output -parameter strongly depends on the -Integral in the nonlinear medium.
The experimental setup for MPC spectral broadening is based on standard commercial opto-mechanics and can be operated for several hours without realignment. Within several weeks of operation of the MPC setup, neither coating nor material damage of any MPC optical element has been observed.
Up to a transmitted power of , the beam quality at the output of the MPC is almost unchanged compared to the input beam quality [Fig. 3(c)]. For transmitted powers , the beam quality starts to deteriorate rapidly. The beam quality is affected by aberrations of the Kerr and thermal lenses in the mirror substrates, and we assume that the complex addition of aberrations of different passes through the nonlinear media cause the highly nonlinear dependence between output power and beam quality. At a transmitted power of 354 W, beam quality is measured to . When the transmitted power is increased toward this value, the beam quality factors in - and -directions assimilate, i.e., the beam quality factor in -direction decreases compared to the beam quality at the input of the cell while the beam quality factor in -direction increases. The cause of this assimilation will be investigated further.
In the following, the bandwidth of the modulated output spectra is defined as the width between the points on the outer slopes where the spectral intensity has fallen to half of the maximum intensity of the outer peaks. At 350 W of output power, a spectral bandwidth of 13.4 nm is observed, and at the maximum transmitted power a bandwidth of 16.1 nm is reached (Fig. 4). Compared to the amplifier output bandwidth of 1.6 nm, this corresponds to an 8.4- and 10-fold increase in bandwidth. This increase in bandwidth matches the expectation for the maximum -Integral estimated above. From the measured spectra, transform-limited pulse durations of, respectively, 160 fs and 140 fs (FWHM) are calculated.
To characterize the MPC behavior when operating with reduced input beam quality, we remove the spatial filtering module after the Innoslab amplifier. In this case, the beam quality incident to the MPC is , and a similar assimilation of beam quality in - and -directions with increasing output power is observed [Fig. 3(d)], leading to a beam quality of at 370 W of output power. At this power, the spectrum at the output of the MPC is depicted in Fig. 4(a). Its FWHM bandwidth matches the bandwidth of the corresponding spectrum with spatially filtered input beam showing that MPC-based spectral broadening is not sensitively affected by input beam quality.
For pulse compression subsequent to spectral broadening, a dispersive mirror compressor with three reflections on a mirror with a nominal GDD of ( and a bandwidth of 15 nm (manufacturer specification) is integrated. At a power of 375 W and a bandwidth of 13.5 nm optimum compression is observed [Fig. 5(a)]. The measured autocorrelation has an FWHM width of 200 fs and closely matches the shape of the transform-limited autocorrelation trace, which has been calculated from the spectrum. Based on the calculated transform-limited pulse duration, a pulse duration of is inferred. Thus, the pulse duration of the laser system has been reduced by a factor of five. From the autocorrelation, we estimate that of the pulse energy is contained in the main peak, which corresponds to an estimated peak power of for a pulse energy of 37.5 μJ.
At this power level, the beam quality after the compressor is measured to [Fig. 5(b)]. This reduction in beam quality compared to Fig. 3(c) originates from thermal effects in the compressor mirrors, presumably due to absorption in the coating. Using a mirror compressor with lower absorption by means of a lower GDD per reflection can improve output beam quality.
In conclusion, we have demonstrated nonlinear spectral broadening at average output powers of up to 450 W. This corresponds to pulse energies of 45 μJ. An -fold increase in bandwidth and an ∼5-fold decrease in pulse duration have been shown with excellent transmissions of the compression setup . Up to a threshold output power ( for the investigated configuration), beam quality is unchanged by the spectral broadening unit.
The achievable bandwidths and compressed pulse durations are—for a fixed compression factor—limited only by the bandwidths of the dispersive coatings of the MPC optics. Thus, compressed pulse durations are feasible. The demonstrated scheme can be adapted to different peak powers of the input pulses (i.e., different input pulse energies and input pulse durations) by changing the spot size in the nonlinear medium via the mirror distance and curvature and by changing the nonlinear medium thickness. This scalability is limited by the damage thresholds and dimensions of the MPC. For MPC dimensions , the scheme is suitable for input peak powers between 5 and 300 MW corresponding to –250 μJ pulse energies at input pulse durations. Thus, output peak powers exceeding 1 GW appear feasible using this approach.
For average power scalability, the compression scheme is limited only by thermal aberrations by absorption of coatings and the nonlinear media, so that scaling to kW-level average powers seems feasible.
Similarly, the achievable compression factor is limited by beam quality deterioration due to Kerr lensing in the nonlinear media. Further investigation is needed, but an increase of the compression factor should be feasible by increasing the number of passes through the nonlinear media.
Bundesministerium für Bildung und Forschung (BMBF) (13N11628 “FOKUS”).
1. D. Kiselev, L. Woeste, and J.-P. Wolf, Appl. Phys. B 100, 515 (2010). [CrossRef]
2. R. R. Gattass and E. Mazur, Nat. Photonics 2, 219 (2008). [CrossRef]
3. P. Russbueldt, T. Mans, J. Weitenberg, H. D. Hoffmann, and R. Poprawe, Opt. Lett. 35, 4169 (2010). [CrossRef]
4. P. Russbueldt, H. D. Hoffmann, M. Höfer, J. Löhring, J. Luttmann, A. Meissner, J. Weitenberg, M. Traub, T. Sartorius, D. Esser, R. Wester, P. Loosen, and R. Poprawe, IEEE J. Sel. Top. Quantum Electron. 21, 447 (2015). [CrossRef]
5. C. J. Saraceno, F. Emaury, C. Schriber, A. Diebold, M. Hoffmann, M. Golling, T. Südmeyer, and U. Keller, IEEE J. Sel. Top. Quantum Electron. 21, 106 (2015). [CrossRef]
6. M. Müller, M. Kienel, A. Klenke, T. Gottschall, E. Shestaev, M. Plötner, J. Limpert, and A. Tünnermann, Opt. Lett. 41, 3439 (2016). [CrossRef]
7. C. Jocher, T. Eidam, S. Hädrich, J. Limpert, and A. Tünnermann, Opt. Lett. 37, 4407 (2012). [CrossRef]
8. Y. Liu, W. Li, D. Luo, D. Bai, C. Wang, and H. Zeng, Opt. Express 24, 10939 (2016). [CrossRef]
9. W. Liu, D. N. Schimpf, T. Eidam, J. Limpert, A. Tünnermann, F. X. Kärtner, and G. Chang, Opt. Lett. 40, 151 (2015). [CrossRef]
10. A. Vernaleken, J. Weitenberg, T. Sartorius, P. Russbueldt, W. Schneider, S. L. Stebbings, M. F. Kling, P. Hommelhoff, H. D. Hoffmann, R. Poprawe, F. Krausz, T. W. Hänsch, and Th. Udem, Opt. Lett. 36, 3428 (2011). [CrossRef]
11. I. Pupeza, S. Holzberger, T. Eidam, H. Carstens, D. Esser, J. Weitenberg, P. Rußbüldt, J. Rauschenberger, J. Limpert, Th. Udem, A. Tünnermann, T. W. Hänsch, A. Apolonski, F. Krausz, and E. Fill, Nat. Photonics 7, 608 (2013). [CrossRef]
12. H. Carstens, M. Högner, T. Saule, S. Holzberger, N. Lilienfein, A. Guggenmos, C. Jocher, T. Eidam, D. Esser, V. Tosa, V. Pervak, J. Limpert, A. Tünnermann, U. Kleineberg, F. Krausz, and I. Pupeza, Optica 3, 366 (2016). [CrossRef]
13. J. Rothhardt, S. Hädrich, H. Carstens, N. Herrick, S. Demmler, J. Limpert, and A. Tünnermann, Opt. Lett. 36, 4605 (2011). [CrossRef]
14. S. Hädrich, A. Klenke, J. Rothhardt, M. Krebs, A. Hoffmann, O. Pronin, V. Pervak, J. Limpert, and A. Tünnermann, Nat. Photonics 8, 779 (2014). [CrossRef]
15. S. Hädrich, M. Kienel, M. Müller, A. Klenke, J. Rothhardt, R. Klas, T. Gottschall, T. Eidam, A. Drozdy, P. Jojart, Z. Varallyay, E. Cormier, K. Osvay, A. Tünnerman, and J. Limpert, in High Brightness Sources and Light-Driven Interactions, OSA Technical Digest (Optical Society of America, 2016), paper JT3A.1.
16. F. Emaury, C. J. Saraceno, B. Debord, D. Ghosh, A. Diebold, F. Gerome, T. Südmeyer, F. Benabid, and U. Keller, Opt. Lett. 39, 6843 (2014). [CrossRef]
17. F. Guichard, A. Giree, Y. Zaouter, M. Hanna, G. Machinet, B. Debord, F. Gerome, P. Dupriez, F. Druon, C. Hönninger, E. Mottay, F. Benabid, and P. Georges, Opt. Express 23, 7416 (2015). [CrossRef]
18. S. Hädrich, M. Krebs, A. Hoffmann, A. Klenke, J. Rothhardt, J. Limpert, and A. Tünnermann, Light Sci. Appl. 4, e320 (2015). [CrossRef]
19. F. Guichard, Y. Zaouter, M. Hanna, F. Morin, C. Hönninger, E. Mottay, F. Druon, and P. Georges, Opt. Lett. 38, 4437 (2013). [CrossRef]
20. A. Klenke, S. Hädrich, M. Kienel, T. Eidam, J. Limpert, and A. Tünnermann, Opt. Lett. 39, 3520 (2014). [CrossRef]
21. C. Rolland and P. B. Corkum, J. Opt. Soc. Am. B 5, 641 (1988). [CrossRef]
22. M. Seidel, G. Arisholm, J. Brons, V. Pervak, and O. Pronin, Opt. Express 24, 9412 (2016). [CrossRef]