We present a high peak power degenerated parametric amplifier operating at 1030 nm and 97 kHz repetition rate. Pulses of a state-of-the art fiber chirped-pulse amplification (FCPA) system with 840 fs pulse duration and 410 µJ pulse energy are used as pump and seed source for a two stage optical parametric amplifier. Additional spectral broadening of the seed signal in a photonic crystal fiber creates enough bandwidth for ultrashort pulse generation. Subsequent amplification of the broadband seed signal in two 1 mm BBO crystals results in 41 µJ output pulse energy. Compression in a SF 11 prism compressor yields 37 µJ pulses as short as 52 fs. Thus, pulse shortening of more than one order of magnitude is achieved. Further scaling in terms of average power and pulse energy seems possible and will be discussed, since both concepts involved, the fiber laser and the parametric amplifier have the reputation to be immune against thermo-optical effects.
©2008 Optical Society of America
High-peak-power-ultrashort pulses have found numerous applications ranging from industrial to fundamental science. One of the most impressive nonlinear phenomenon is the generation of high harmonics (HHG) [1, 2]. Today, harmonics with orders above 100 and wavelengths below 10 nm are routinely generated [3, 4]. As an example this wavelength range is of great interest for microscopy purposes since the smallest details that can be observed are theoretically of the order of the imaging wavelength. The coherent-soft x - rays even allow for lensless imaging techniques, which have been experimentally demonstrated recently, using a tabletop-coherent-soft x - ray source based on Ti:Sapphire laser driven HHG . However, an increased laser repetition rate is desirable to decrease the necessary long detector integration times in such applications, which are currently of the order of hours. Unfortunately, the commonly used Ti:Sapphire based laser systems are limited to several watts of average output power due to thermal lensing .
In contrast, diode pumped Yb-based lasers which emit at the 1µm wavelength region provide enhanced efficiency. Additionally, Yb-doped fiber lasers offer average power scalability due to reduced thermo-optical distortions. Femtosecond fiber amplifiers based on chirped-pulse amplification with output pulse energies of 100 µJ  up to 1 mJ  and average powers up to 131 W  have been reported recently. The main drawback is the relatively small gain bandwidth of the Yb ions, which does not allow for high energy sub-100 fs pulse generation . To decrease the pulse duration, and therefore e.g. support efficient HHG, a pulse compression technique is needed.
In this regard, the application of optical parametric amplifiers (OPA) seems promising. They are capable of ultrashort-pulse generation due to their enormous amplification bandwidth of more than 200 THz, especially in a noncollinear geometry  and at degeneracy [12, 13]. Furthermore, OPA provides high gain (typically 102 to 104) using only very short crystal lengths of a few millimeters. Consequently, pulse distortions due to the accumulated-nonlinear phase are negligible. Thermo-optical distortions are minor, since the energy is preserved during the parametric process and the residual absorption of the nonlinear crystals is very low, hence, optical parametric amplifiers are capable to handle high average powers.
The combination of fiber chirped-pulse amplification systems with an OPA, to generate ultrashort pulses, has been demonstrated and reported in several publications. Up to now, the output pulse energies were limited to below 1 µJ [14, 15, 16]. In this contribution we report on a parametric amplifier, both seeded and pumped by a high repetition rate fiber chirped pulse amplification (CPA) system . A schematic overview of the setup is shown in fig. 1. The laser system has been operated at a repetition rate of 97 kHz and delivered 840 fs pulses with a pulse energy of 410 µJ corresponding to 40 W average power. The degenerated optical parametric amplifier is seeded by a spectrally broadened signal, generated in a large mode area fiber, and yielded output pulses with up to 41 µJ pulse energy. Recompression in a simple prism compressor leads to a pulse duration of 52 fs, which equals a compression ratio of more than one order of magnitude, compared to the initial 840 fs pulses. The resulting peak power of more than 0.5 GW, is to our knowledge the highest value ever reported for a fiber amplifier pumped OPA. In principle, this scheme can be adapted to other OPA configurations, such as noncollinear amplification of a broadband signal generated in bulk media, to achieve even shorter pulse durations and wavelength tuneability.
2. Fiber chirped-pulse amplification system and second harmonic pump generation
A high repetition rate fiber chirped-pulse amplification system, which has already been reported in  is used to drive the parametric amplifier. It consists of a Yb:KGW oscillator, a stretcher compressor unit, based on dielectric gratings, and two photonic crystal fiber amplifiers. In the herein presented setup the repetition rate is set to 97 kHz with a quartz based acusto-optical modulator (AOM), while the pulse energy is chosen to be 410 µJ as a compromise between pulse energy and pulse quality. The autocorrelation trace of the recompressed pulses is shown in fig. 2. The pulse duration can be calculated to 840 fs with a deconvolution factor of 1.41, assuming a Gaussian pulse shape. The resulting pulse peak power is 420 MW, when considering 90 % of the pulse energy in the main pulse.
A small fraction of the infrared pulses is used to generate a broadband signal in a photonic crystal fiber which is described in section 3. To pump the parametric amplifier the main part of the infrared pulses is frequency doubled in a 1 mm long type 1 critically-phase-matched BBO crystal. To achieve efficient conversion the beam size is chosen to be 1.0 mm in the nonlinear crystal by the means of a Galilean telescope. The corresponding peak intensity is 53 GW/cm2 yielding a conversion efficiency of 55% while maintaining an excellent beam profile. The autocorrelation trace of the frequency doubled pulses is shown in fig. 3. Note that the second harmonic generation process leads to a cleanup of the temporal pulse shape which is caused by lower conversion efficiency for the pre- and post-pulses. The pulse width of the frequency doubled pulses can be calculated to be 710 fs, estimating a Gaussian pulse shape. The average power has been measured to be 22 W corresponding to 227 µJ pulse energy. It has to be stressed that no thermal effects like thermal dephasing have been observed even at this high-average power level.
3. Broadband signal generation
Temporal synchronization of signal and pump pulses is inherently given using fractions of the same laser pulses to generate both the pump and the signal. Therefore, we have chosen the technique of nonlinear pulse compression to generate the ultrashort signal pulses. This approach lowers the complexity of our system in comparison to fiber based optical synchronization of high power fiber amplifier and broadband Ti:Sapphire oscillator . In principle, the technique of nonlinear compression is well known . The achievable minimum pulse duration strongly depends on the initial pulse shape, which determines the shape of the nonlinear phase, and especially on the higher dispersion orders of all involved optical components, which cannot easily be compensated for. On the other hand, the presence of group velocity dispersion (GVD) can also improve the quality of the compressed output pulse by reshaping the pulse during the nonlinear broadening . This leads to a nearly linear chirp, which can easily be compensated for, e.g. by a prism- or grating pair.
In practice, the maximum pulse energy that can be coupled to the broadening fiber is limited by the onset of surface damage and self focusing in the fiber. We have chosen a photonic crystal fiber with a mode field diameter of 33 µm which allows stable and robust fundamental mode propagation. For 800 fs pulses at 1064 nm wavelength a damage threshold of 120 GW/cm2 has been measured for fused silica . Based on these measurements fiber facet damage is expected for a coupled pulse energy of 1.7 µJ, assuming a 840 fs Gaussian pulse with 90 % of the energy in the central pulse. The corresponding peak power is 1.7 MW and therefore well below the self-focusing threshold (~4.3 MW) .
A numerical simulation solving the extended nonlinear Schrödinger equation based on the split step Fourier method is used to study the process in detail. The simulation includes self-phase modulation (SPM) in the fiber as well as group velocity dispersion (GVD) and third order dispersion (TOD) of the fiber, chirped mirrors and a SF 11 prism compressor, while dispersion of the OPA crystals, lenses, wave plates and polarizers are neglected. At first the pulses are spectrally broadened in a fiber with variable length followed by pre- compression with chirped mirrors to a pulse duration of ~500 fs, which is required for sufficient operation of the OPA. Finally, the pulses are compressed to minimum pulse duration by the means of a SF11 prism pair. The spectral bandwidth at the fiber end is shown in fig. 4 (blue) together with the compressed pulse duration (green).
A 20 cm long fiber leads to 134 nm bandwidth and a compressed pulse duration of 23.5 fs. However, the pulse quality of the compressed pulses is dramatically decreased at large spectral bandwidths due to residual higher order dispersion which cannot be compensated for with a simple prism pair. The peak powers of the compressed pulses normalized to the initial peak power (black) and the corresponding Fourier limited pulses (red) are plotted in fig. 5. For fiber lengths above 15 cm no significant enhancement in pulse peak power is achieved. Certainly, the pulse quality is decreasing and strong pre-and-post pulses arise. Hence, a compromise between pulse quality and compressed pulse duration has to be found.
Experimentally, a fiber length of 8 cm is chosen as a tradeoff between pulse duration and pulse quality, but also due to other reasons. Firstly, the pulses delivered by the fiber CPA system are already slightly distorted by nonlinearity so their temporal shape differs from an ideal Gaussian-or-sech2 shape and they create a more complex nonlinear phase with larger higher order phase terms. Secondly, the dispersive mirrors used in the experiment for precompression have a nearly constant GVD of -250 fs2 over a wavelength range of 1000 nm to 1080 nm. Outside this design wavelength range the GVD increases dramatically and gets even positive so that compression is not possible outside this wavelength range. Last but not least, a longer broadening fiber and a larger bandwidth require a large prism separation and a huge prism size which is in contrast to a simple compact experimental setup.
The measured spectrum at the output of the fiber (black) is shown in fig. 6 together with the corresponding numerical result (red dots).
A spectral bandwidth (FWHM) of 43 nm was achieved by coupling 1.7 µJ pulses to the fiber. In order to fulfill the requirement of the OPA (signal pulse duration < pump pulse duration) the spectrally-broadened pulses are pre- compressed to a pulse duration of 500 fs by 8 bounces on the chirped mirrors and final compression is achieved in a SF11 prism compressor. To suppress the small pre-and-post pulses delivered by the fiber CPA system a nonlinear loss, based on polarization rotation in the fiber, was introduced . It is accomplished by placing a polarizer in front of the fiber and a quarter wave plate, a half wave plate and a second polarizer at the fiber output. Carefully tuning the wave plates suppressed the side pulses and therefore cleaned the output pulse shape and especially removed a broad plateau underlying the compressed ultrashort pulse.
4. Optical parametric amplifier - preliminary considerations
The design of a high power optical parametric amplifier has to be optimized in terms of conversion efficiency. Especially for Gaussian beam profiles the effect of gain guiding becomes a serious problem . The amplified beams tend to be narrowed due to the presence of a non- uniform parametric gain caused by the pump beam profile. As a result, either poor beam quality is observed in the case of back conversion in the central part of the beam or low conversion efficiency is achieved because the peripheral parts are not converted efficiently. Diffraction can help to overcome this problem if the interacting beams are focussed tightly to spot sizes of less than 100 µm . However, this approach is limited in terms of pulse peak power and average power due to optical damage, nonlinear refraction and thermal problems arising from residual absorption. Nevertheless, high conversion efficiency and high beam quality can be combined by a two stage OPA. Figure 7 shows the experimental pump to signal conversion efficiency versus the seed signal average power for a 1mm long BBO crystal. These data are taken by using the frequency doubled pump pulses at three different intensities and injecting a fraction of the infrared fiber CPA system pulses as seed with varying power. The beam waist of the signal is carefully adjusted for maximum conversion efficiency and good beam quality. The achieved conversion efficiency versus the signal gain is shown in fig.8. Clearly, a conversion efficiency of more than 20 % can be achieved when the saturated amplifier gain is kept below 5. The highest conversion efficiency combined with the best beam quality is obtained with the lowest pump intensity. In conclusion our ultrashort-pulse-parametric amplifier should consist of a first amplifier stage which is operated in a high gain regime with moderate conversion efficiency followed by a power amplifier operated at low gain to avoid spatial gain narrowing and to achieve high conversion efficiency.
5. Ultrashort pulse optical parametric amplifier
The experimental setup of the ultrashort-pulse-parametric amplifier is shown in fig. 9. A two stage OPA, optimized according to section 4 is used to amplify the spectrally-broadened pulses. The seed power delivered by the broadening fiber was measured to be 50 mW due to undesired absorption losses and required intensity depended losses due to the nonlinear polarization rotation filter. The first parametric amplifier is driven by 22 W average power pump pulses and delivers 1.4 W of output power when a pump intensity of 15 GW/cm2 is applied. Due to the low total pump depletion of 12 % the spatial beam profile of the remaining pump is still nearly Gaussian and suitable for pumping the second OPA. The remainder of the pump power (~19 W) is slightly focussed into the second 1mm long BBO crystal, where the pump intensity is set to 15 GW/cm2 again. Careful mode matching led to 4.0 W output power which corresponds to a pulse energy of 41 µJ and a pump to signal conversion efficiency of 21%. The corresponding spectra are shown in fig. 10. Due to the saturated amplification regime and the large amplification bandwidth, which is characteristic for the degenerated OPA, the spectral bandwidth is increased and the spectral modulation depth is decreased during amplification. Finally, a stable output spectrum with a spectral bandwidth as large as 60.4 nm is achieved at the highest output power.
After collimating the beam, the pulses are compressed by a SF11 prism pair with a separation of ~1.06 m. The autocorrelation trace of the compressed pulses is shown in fig. 11 (black). With the Fourier transform of the measured spectrum, stretched to the measured autocorrelation width by TOD, a deconvolution factor of 1.64 is found. Based on this, the pulse duration is calculated to be 51.7 fs and the pulse peak power is found to be 598 MW. In fig. 11 a difference of the measured and theoretical expected AC trace can be seen. This is mainly caused by the pulse shape of the fiber CPA output pulses and higher order dispersion as discussed in detail in section 3. Comparison of the measured autocorrelation trace and the numerical result leads to an experimental pulse peak power of approximately 550 MW. The throughput of the prism compressor is measured to be 90 % so the compressed pulses have a pulse energy of 37 µJ. In summary, the compressed pulse peak power can be estimated to be about 500 MW.
The far field profiles of the amplified beams are shown in fig. 12. It has to be stressed that due to the moderate pump intensity and the short pump pulse duration no significant parametric superfluorescence is observed.
6. Conclusion and outlook
In conclusion we demonstrated a two stage optical parametric amplifier delivering 52 fs pulses with 37 µJ pulse energy at a repetition rate as high as 97 kHz. The system is driven by a chirped-pulse fiber amplifier delivering 840 fs pulses with 410 µJ pulse energy. Using an optimized two-stage amplifier design we were able to achieve a pump to signal conversion efficiency of up to 21 %. The compressed output power was measured to be 3.6 W. So the overall conversion efficiency, including the SHG the OPA process and the pulse compression is as high as 9 %. Hence, we demonstrated not only shortening of the pulse duration by more than one order of magnitude, but also increasing the pulse peak power from 420 MW to 500 MW. To our knowledge the reported pulse energy and pulse peak power is more than one order of magnitude larger than reported before for a fiber-laser-pumped-ultrashort-pulse OPA. The reported average output power is three times larger than the highest reported value for a Ti:Sapphire driven OPA . Further improvement of the total efficiency of the nonlinear processes could be achieved by applying a Gauss to Flattop beam shaper. Implementation of the already reported fiber-based synchronization of a Ti:Sapphire oscillator with a short pulse Yb-fiber amplifier  could result in sub-20 fs pulse with clean pulse shape and a pulse peak power of more than 1 GW at 100 kHz repetition rate. Scaling to higher average power will be achieved by applying a fiber CPA system, operating at kW average power, which is in preparation. Higher pulse energies may be achieved by using the technique of optical parametric chirped pulse amplification (OPCPA), i.e. longer pump pulses, to avoid limitations due to nonlinearity in the fiber based pump source. Overall short pulse parameters are in reach, which, together with the high repetition rates, will open the addressing of novel applications.
References and links
1. A. McPherson, G. Gibson, H. Jara, U. Johann, T. S. Luk, I. A. McIntyre, K. Boyer, and C. K. Rhodes, “Studies of multiphoton production of vacuum-ultraviolet radiation in the rare gases,” J. Opt. Soc. Am. B 4, 595 (1987). [CrossRef]
2. M. Ferray, A. L’Huillier, X. F. Li, L. A. Lompre, G. Mainfray, and C. Manus, “Multiple harmonic conversion of 1064 nm radiation in rare gases,” J. Phys. B 21, L31 (1988). [CrossRef]
3. E. A. Gibson, A. Paul, N. Wagner, R. Tobey, I. P. Christov, D. T. Attwood, E. Gullikson, A. Aquila, M. M. Murnane, and H. C. Kapteyn, “Coherent Soft X-ray Generation in the Water Window with Quasi-Phase Matching,” Science 302, 95–98 (2003). [CrossRef] [PubMed]
4. J. Seres, P. Wobrauschek, Ch. Streli, V. S. Yakovlev, E. Seres, F. Krausz, and Ch. Spielmann, “Generation of coherent keV x-rays with intense femtosecond laser pulses,” New J. Phys. 8, 1–11 (2006). [CrossRef]
5. R. Sandberg, A. Paul, D. Raymondson, S. Hädrich, D. Gaudiosi, J. Holtsnider, R. Tobey, O. Cohen, M. Murnane, H. Kapteyn, C. Song, J. Miao, Y. Liu, and F. Salmassi, “Lensless diffractive imaging using tabletop coherent high-harmonic soft-X-ray beams,” Phys. Rev. Lett. 99, 098103 (2007). [CrossRef] [PubMed]
6. S. Backus, C. Durfee, M. M. Murnane, and H. C. Kapteyn, “High power ultrafast lasers,” Rev. Sci. Instrum. 69, 1207–1223 (1998). [CrossRef]
7. A. Galvanauskas, G. C. Cho, A. Hariharan, M. E. Fermann, and D. Harter, “Generation of high-energy femtosecond pulses in multimode-core Yb-fiber chirped-pulse amplification systems,” Opt. Lett. 26, 935–937 (2001) http://www.opticsinfobase.org/abstract.cfm?URI=ol-26-12-935 [CrossRef]
8. F. Röser, T. Eidam, J. Rothhardt, O. Schmidt, D. N. Schimpf, J. Limpert, and A. Tünnermann, “Millijoule pulse energy high repetition rate femtosecond fiber chirped-pulse amplification system,” Opt. Lett. 32, 3495–3497 (2007). http://www.opticsinfobase.org/abstract.cfm?URI=ol-32-24-3495 [CrossRef] [PubMed]
10. R. Paschotta, J. Nilsson, A. Tropper, and D. Hanna, “Ytterbium-doped fiber amplifiers,” IEEE J. Quantum Electron. 33, 1049–1056 (1997). [CrossRef]
11. D. N. Schimpf, J. Rothhardt, J. Limpert, A. Tünnermann, and D. C. Hanna, “Theoretical analysis of the gain bandwidth for noncollinear parametric amplification of ultrafast pulses,” J. Opt. Soc. Am. B 24, 2837–2846 (2007). http://www.opticsinfobase.org/abstract.cfm?URI=josab-24-11-2837 [CrossRef]
12. G. Cerullo and S. Silvestri, “Ultrafast optical parametric amplifiers,” Rev. Sci. Instrum. 74, 1 (2003). [CrossRef]
13. J. Limpert, C. Aguergaray, S. Montant, I. Manek-Hönninger, S. Petit, D. Descamps, E. Cormier, and F. Salin, “Ultra-broad bandwidth parametric amplification at degeneracy,” Opt. Express 13, 7386–7392 (2005). [CrossRef] [PubMed]
14. C. Schriever, S. Lochbrunner, P. Krok, and E. Riedle, “Tunable pulses from below 300 to 970 nm with durations down to 14 fs based on a 2 MHz ytterbium-doped fiber system,” Opt. Lett. 33, 192–194 (2008). http://www.opticsinfobase.org/abstract.cfm?URI=ol-33-2-192 [CrossRef] [PubMed]
15. J. Rothhardt, S. Hädrich, D. N. Schimpf, J. Limpert, and A. Tünnermann, “High repetition rate fiber amplifier pumped sub-20 fs optical parametric amplifier,” Opt. Express 15, 16729–16736 (2007). [CrossRef] [PubMed]
16. T. V. Andersen, O. Schmidt, C. Bruchmann, J. Limpert, C. Aguergaray, E. Cormier, and A. Tünnermann, “High repetition rate tunable femtosecond pulses and broadband amplification from fiber laser pumped parametric amplifier,” Opt. Express 14, 4765–4773 (2006). [CrossRef] [PubMed]
17. W. J. Tomlinson, R. H. Stolen, and C. V. Shank, “Compression of optical pulses chirped by self-phase modulation in fibers,” J. Opt. Soc. Am. B 1, 139- (1984). http://www.opticsinfobase.org/abstract.cfm?URI=josab-1-2-139 [CrossRef]
18. R. M. Wood, The power and energy handling capabilities of optical materials components, and systems, (SPIE Press, 2003).
19. R. L. Farrow, D. A. V. Kliner, G. R. Hadley, and A. V. Smith, “Peak-power limits on fiber amplifiers imposed by self-focusing,” Opt. Lett. 31, 3423–3425 (2006). http://www.opticsinfobase.org/abstract.cfm?URI=ol-31-23-3423 [CrossRef] [PubMed]
20. D. Homoelle, A. L. Gaeta, V. Yanovsky, and G. Mourou, “Pulse contrast enhancement of high-energy pulses by use of a gas-filled hollow waveguide,” Opt. Lett. 27, 1646–1648 (2002) http://www.opticsinfobase.org/abstract.cfm?URI=ol-27-18-1646 [CrossRef]
21. G. Arisholm, R. Paschotta, and T. Südmeyer, “Limits to the power scalability of high-gain optical parametric amplifiers,” J. Opt. Soc. Am. B 21, 578–590 (2004). http://www.opticsinfobase.org/abstract.cfm?URI=josab-21-3-578 [CrossRef]
22. S. Adachi, H. Ishii, T. Kanai, N. Ishii, A. Kosuge, and S. Watanabe, “1.5 mJ, 6.4 fs parametric chirped-pulse amplification system at 1 kHz,” Opt. Lett. 32, 2487–2489 (2007) http://www.opticsinfobase.org/abstract.cfm?URI=ol-32-17-2487 [CrossRef] [PubMed]