We report on a high repetition rate noncollinear optical parametric amplifier system (NOPA) based on a cavity dumped Ti:Sapphire oscillator providing the signal, and an Ytterbium-doped fiber amplifier pumping the device. Temporally synchronized NOPA pump pulses are created via soliton generation in a highly nonlinear photonic crystal fiber. This soliton is fiber amplified to high pulse-energies at high repetition rates. The broadband Ti:Sapphire laser pulses are parametrically amplified either directly or after additional spectral broadening. The approach of fiber-based pump-pulse generation from a femtosecond laser, that emits in the spectral region of NOPA–gain, offers enhanced long-term stability and pulse quality compared to conventional techniques, such as signal pulse generation from a high power laser system via filamentation in bulk media. The presented system produces high-energy ultra-short pulses with pulse-durations down to 15.6 fs and pulse-energies up to 500 nJ at a repetition rate as high as 2 MHz.
©2007 Optical Society of America
The availability of ultra-short and high peak power optical pulses has driven breathtaking advances in applications ranging from industrial to fundamental science. Traditionally high peak power pulses are generated by Ti:Sapphire laser systems. Their scalability to repetition rates beyond the kHz regime, and hence, to higher average power is limited to the watt-power-level by thermo-optical problems . Even advanced approaches such as cryogenic cooling of the laser crystal could not substantially solve this issue . Unfortunately, a number of ultra fast processes initiated by high peak power pulses (e.g. high harmonic generation) is characterized by a low probability or a low conversion efficiency. As a consequence, detection systems comprise very sophisticated and sensitive apparatuses precluding them from real world applications. An increase of few orders of magnitude in the repetition rate could dramatically decrease the necessary measurement times.
Diode pumped solid-state lasers, which emit 1 µm wavelength radiation, provide enhanced efficiency compared to Ti:Sapphire lasers and even more important in advanced geometries, such as disk or fiber, they offer average power-, i.e. repetition rate-, scalability due to reduced thermo-optical distortions. Though, their gain-bandwidth doesn’t support pulse-durations of few-10 fs .
Optical parametric amplification (OPA) in nonlinear crystals such as BBO or LBO offers a large amplification bandwidth using either non-collinear configuration (NOPA) [4,5] or amplification at degeneracy [6,7]. Hence, the amplification of few cycle optical pulses is feasible [8,9]. Furthermore, OPA provides a high gain (typically 104 to 106) using only very short crystal lengths (few mm). Consequently the pulse-distortions due to accumulated nonlinear phase are negligible. Based on these facts, OPA is a promising way in order to generate high peak-power pulses at high repetition rates. Our approach involves the transfer of the high pulse–energy, at high average power, of a 1 µm laser source to pulses of significantly shorter pulse-durations via parametric amplification. This approach is supported by the inherent immunity of parametric amplifiers against thermo-optical problems due to the energy conservation during the instantaneous nonlinear process.
It is well known that fiber-laser-systems are average power scalable due to the fiber design itself, and femtosecond fiber-amplification-systems have been demonstrated with average powers above 100 W and pulse energies above 100 µJ [10, 11]. Hence, they are suitable pump sources for ultra-fast OPAs.
On the other hand, the achievable peak-power of short-pulse fiber-laser-systems is limited by nonlinear effects, mainly in form of self-phase modulation. A significant increase of peak-power can be expected by an ultra-fast OPA, which is driven by a fiber-laser.
Traditionally, to provide a synchronized broadband signal white light is generated by filamentation in a bulk substrate such as fused silica or sapphire. In particular, the signal needs to overlap with the gain-bandwidth of a NOPA.
This has been reported recently for ultra-fast OPA systems, which are pumped by µJ-pulse-energy-level high repetition-rate laser-systems [12, 13]. The main drawbacks of this approach are instabilities and degradation of the filament and hence the generated continuum. This can be partly avoided by applying very short (sub-100 fs) pulses in order to drive the filamentation, what restricts the performance of the 1 µm laser-system or requires a pre-compression of the laser pulses before filamentation . Providing a seed signal by supercontinuum-generation in highly nonlinear photonic crystal fibers has been reported as well . These investigations have shown that broad continua possess limited compressibility due to phase offsets inherent to the broadening process [14, 15].
In this contribution we report on a parametric amplifier, operating at 2 MHz repetition rate and delivering up to 500 nJ pulses with sub-20 fs pulse duration, which is pumped by a short pulse fiber laser. The starting point is a cavity dumped Ti:Sapphire laser, which serves as the signal for the OPA, producing 20 fs pulses with few nJ of pulse-energy at variable repetition rates. Temporally synchronized pump pulses are created by soliton generation in a photonic crystal fiber. Subsequently, the soliton is amplified in a fiber-based chirped-pulse-amplification (CPA) system. This approach is possible due to the fact that the phase of the signal is not influenced by pump pulses’ phase, which is dissipated to the generated idler. Hence, in our case the quality of the Ti:Sapphire laser pulses is not degraded by any imposed nonlinear phase of the fiber amplified pump pulses. The generation of synchronized pump pulses via soliton formation in a highly nonlinear photonic crystal fiber has been already reported for a NOPA pumped by a bulk 10 Hz regenerative amplifier . Herein we report, for the first time to our knowledge, on an ultra-fast NOPA that is pumped by a high repetition-rate fiber-laser-system producing long term stable high quality pulses with duration as short as 15.6 fs.
2. Inherent pump synchronization via soliton generation
In principle, it should be possible to seed a short pulse Yb doped fiber amplifier with an octave-spanning Ti:Sapphire oscillator . However, these oscillators, based on highly advanced broadband chirped mirrors, cannot ensure picojoule energy level within the bandwidth of an Yb-doped fiber amplifier which is required to compete efficiently with the amplified spontaneous emission (ASE). To enhance the seed pulse energy a resonant frequency conversion is needed. Due to the low intensities, bulk parametric frequency shifters become extremely inefficient when seeded directly with the oscillator output .
The special properties of soliton phenomena in optical fibers suggest an interesting alternative. Photonic-crystal fibers substantially enhance nonlinear optical processes due to a strong field confinement in a small-size fiber core. Furthermore, dispersion can be tailored for the soliton process. Sub-ps pulses propagating in these fibers are spectrally broadened via self phase modulation. The spectral location within the anomalous dispersion region leads to formation of fundamental solitons which experience continuous frequency downshifting due to the Raman-effect . This, so-called soliton self-frequency shift (SSFS) provides a convenient way of tuning the central wavelength of the generated ultra-short pulse. The nonlinear wavelength shift is strongly dependent on the parameters of the input pulse and therefore sensitive to noise of the seed pulse.
In the experiment we used the soliton formation in order to create a synchronized seed pulse for the fiber amplifier. To do so, pulses with an energy of 1.1 nJ, a pulse duration of 55 fs and a bandwidth of 50 nm at 810 nm central wavelength are coupled into a photonic crystal fiber (Crystal Fibre, PCF NL-3.7-975). The fiber has a core diameter of 3.7 µm and a zero dispersion wavelength of 975 nm providing phase-matched soliton generation at 1030 nm central wavelength. A soliton frequency shift is not necessary and we observe very stable output pulses which are capable of seeding a high power amplifier chain. The measured output spectrum of the PCF (red) together with the result of a numerical simulation is shown in Fig. 1. The numerical calculations are based on a split-step Fourier method, which includes nonlinear effects like self-phase modulation, Raman-effect, self-steepening and dispersion up to the 6th order. The spectral evolution of the pulses in the PCF is shown in Fig. 2. A fundamental soliton with a central wavelength of 1035 nm is generated and no self frequency shift is observed. The experimental and numerical data indicate a pulse energy of about 3 pJ within the amplification bandwidth of Ytterbium-doped silica fibers.
3. Chirped-pulse fiber-amplifier as high repetition rate pump source
The generated pump pulses are amplified in a three-stage fiber–amplifier-system while the repetition rate is set to 2 MHz with the cavity dumper of the Ti:Sapphire oscillator. The experimental setup is shown in Fig. 3. The highly nonlinear PCF is spliced directly to a double pass single-clad Yb-doped fiber amplifier. This device is pumped by a fiber coupled 200 mW single mode laser diode, emitting at a wavelength of 975 nm. The fiber has a length of 30 cm and an active core diameter of 6µm. A second preamplifier consists of a 125 µm pump core double-clad fiber with 10 µm active core diameter. It is pumped by a fiber coupled 975 nm laser diode (d=200 µm, NA=0.2). This boosts the average power to 0.36 W. For efficient nonlinear conversion a clean and pedestal-free pulse is required. In addition, keeping the compressed pulse duration below 1 ps ensures a high damage threshold of the nonlinear crystals and is therefore recommended. In high energy chirped-pulse fiber amplifiers the temporal shape of the compressed output pulses is mainly influenced by self phase modulation (SPM) . To keep the B-Integral low and therefore avoid distortions due to SPM the pulses are stretched to 340 ps by a grating stretcher (gold-coated 1740 line/mm diffraction grating) while the bandwidth is cut to 6nm.
The main amplifier is based on a 1.7 m long polarizing double-clad Ytterbium-doped photonic crystal fiber  possessing an active core mode-field diameter of 30 µm. Its 200 µm inner cladding (NA=0.55) is pumped by a fiber-coupled laser-diode emitting at 976 nm. A launched pump power of 58 W leads to a maximum output power of 31 W. The corresponding pulse energy is 15.5 µJ.
The amplified pulses are recompressed by using a compressor-stage consisting of two (1740 line/mm) gold coated reflection gratings. Their diffraction angle is slightly increased in order to compensate for third order dispersion of the fibers. This ensures a clean shape of the pulse. To avoid thermally induced beam distortions the beam-diameter on the gratings is increased to 20 mm by a telescope. The double-pass compressor efficiency is 65 % resulting in a pulse energy of 10 µJ of the compressed pulses. Figure 4 shows the autocorrelation trace of the compressed output-pulse at a pulse energy of 1µJ and 10 µJ. The 1.2 ps temporal width of the autocorrelation trace at maximum pulse energy leads to a pulse duration of 640 fs using a deconvolution factor of 0.53, which is given by Fourier transformation of the measured spectrum.
According to the simulation, the peak power of the pulses is 10 MW with approximately 20 % of the pulse energy in the wings. The temporal broadening and formation of wings is attributed to nonlinear phase distortions in the main amplifier fiber with a B-Integral of 3.2. Focusing the beam to a spot size of 80 µm in a 2 mm long Type I critically phase-matched LBO crystal results in efficient frequency doubling of the infrared pulses. Figure 5 shows the power of the second harmonic (red) and the conversion efficiency (blue) versus compressed output power of the fiber amplifier. The maximum conversion efficiency of 49 % is obtained at 9 W compressed output power of the fiber amplifier. Because of the slightly distorted pulse shape the conversion efficiency is decreasing to 41 % at the maximum pulse energy. Therefore the maximum pulse-energy of the second harmonic pulses is 4.15 µJ corresponding to 8.3 W average power. A numerical simulation done with the freely available SNLO software  indicates a pulse-duration of 620 fs for the second harmonic pulses.
4. Non-collinear optical parametric amplifier
It is well known that in a non-collinear beam geometry the group-velocity mismatch between pump, signal and idler waves can be efficiently compensated for Type I phase matched BBO parametric amplifiers, resulting in a very broad phase-matching bandwidth [4,5]. If a pump wavelength of 518 nm is chosen and the internal angle between pump and signal is set to 2.6°, the gain bandwidth should range from 700 nm to 900 nm . In our case this permits the amplification of the whole Ti:Saphhire oscillator spectrum. The experimental setup is shown in Fig. 6.
For efficient conversion, the Ti:Sapphire pulses are stretched to 180 fs in a 25mm fused silica substrate. The signal spot size in the 5 mm BBO crystal is chosen to be 80 µm in order to match the pump spot size. Temporal overlap of signal and pump pulses is realized by means of a delay line using a mechanical translation stage.
Figure 7 shows the signal spectrum (red) and the amplified spectrum (blue). Note that due to tuning of the phase matching angle and the temporal overlap, the amplified spectrum is modified to a smooth and less asymmetric shape. A very clean pulse shape, shown in Fig. 8, is observed after the compression with a fused silica prism pair. The measured autocorrelation width is 31.0 fs corresponding to a pulse duration of 20.1 fs when a sech2-temporal pulse shape is assumed. The best compromise of conversion efficiency and beam quality is found when placing the focus (80 µm diameter, corresponding to 100 GW/cm2) of the second harmonic beam slightly behind the BBO crystal. The highest obtained average output power was 1,0 W, which corresponds to a pulse energy of 500 nJ. This equals a conversion efficiency of 12,0 %. It should be mentioned that the beam profile of the amplified signal is strongly degraded (elliptical shape) at higher conversion efficiencies. As a consequence of reflection losses at gold mirrors, that are used to fold the prism compressor, we measured 90 % throughput of the pulse compressor. Therefore, the compressed output pulses have a pulse peak power as high as 20.0 MW.
5. Nonlinear spectral broadening in photonic crystal fiber followed by amplification and recompression
To increase the bandwidth of the signal we launched the Ti:Sapphire pulses into a 1.5 cm long piece of a highly nonlinear photonic crystal fiber (Crystal Fibre, PCF NL-4.7-1030). Spectral broadening is dominated by self phase modulation because the zero dispersion wavelength is located at 1030 nm. At launched pulse energy of 5 nJ the spectral width is increased from 39 nm to 124 nm. Since the seed pulses experience only small amounts of dispersion during propagation through the OPA setup, recompression down to nearly the transform limit should be possible with a simple prism pair. To compensate a part of the second order dispersion, added by the lenses and the fiber itself, the pulses pass a chirped mirror pair (-35 fs2/bounce) five times. The signal pulse duration is therefore kept below 180 fs to ensure amplification of the whole spectrum.
During parametric amplification the temporal components in center of the stretched signal pulse are favored because of the stronger pump intensity at the pump pulse peak. This mechanism helps to reduce the “batman” shape of the spectrum during the amplification process. Figure 9 shows the normalized spectra of the Ti:Sapphire oscillator (green) the broadened spectrum (red) after propagating through 1.5 cm fiber and the parametrically amplified spectrum (blue). A direct comparison of amplified and un-amplified signal is shown in Fig. 10, indicating a gain factor as high as 100. Compared to the previous experiment, a lower average output power of 600 mW is measured because of the lower signal power. Figure 11 shows the measured autocorrelation trace (red dots) and the Fourier transformation of the amplified spectrum (green). The small wings in the autocorrelation trace are caused by the spectral structure, nonlinear phase and uncompensated higher order dispersion.
Deriving a deconvolution factor of 0.74 by Fourier transformation of the measured spectrum results in a pulse duration of 15.6 fs which is only slightly above the transform limit (13.2 fs) of the measured spectrum. Compression is again realized by a simple prism pair made of fused silica.
We have presented an ultra-fast high repetition rate parametric amplifier that combines the advantages of a broadband Ti:Sapphire oscillator and a high average power Yb-doped fiber-amplifier-system. Soliton generation in a highly nonlinear photonic crystal fiber is a versatile technique for stable signal-pump-synchronization.
Using the frequency doubled output pulses of a fiber amplifier with up to 4µJ pulse energy, efficient parametric amplification of the Ti:Sapphire oscillator pulses has been realized. Output energies up to 500 nJ were achieved using a 5mm BBO amplifier crystal. The amplified pulses were compressed by a simple fused silica prism pair, resulting in nearly transform-limited pulses with 20.1 fs pulse duration. The corresponding peak power is as high as 20 MW. Additional spectral broadening by self-phase modulation in a 1.5 cm photonic crystal fiber has been used to enlarge the spectral bandwidth to 124 nm. We have demonstrated parametric amplification up to 300 nJ and recompression to 15.6 fs pulse duration. To our knowledge, such ultra short intense pulses have not been generated at a repetition rate as high as 2 MHz so far. The approach appears to be scalable in terms of average power and pulse energy because both fiber lasers and parametric amplifiers are relatively immune against thermal distortions.
References and links
1. S. Backus, C. Durfee, M. M. Murnane, and H. C. Kapteyn. “High power ultrafast lasers,” Rev. Sci. Instrum. 69, 1207–1223 (1998). [CrossRef]
3. R. Paschotta, J. Nilsson, A. Tropper, and D. Hanna, “Ytterbium-doped fiber amplifiers,” IEEE J. Quantum Electron. 33, 1049–1056 (1997). [CrossRef]
4. G. Cerullo and S. Silvestri, “Ultrafast optical parametric amplifiers,” Rev. Sci. Instrum. 74, 1 (2003). [CrossRef]
5. L. Hongjun, Z. Wei, C. Guofu, W. Yishan, C. Zhao, and R. Chi, “Investigation of spectral bandwidth of optical parametric amplification” Appl. Phys. B 79, 569–576 (2004). [CrossRef]
6. K. Yamakawa, M. Aoyama, Y. Akahane, K. Ogawa, K. Tsuji, A. Sugiyama, T. Harimoto, J. Kawanaka, H. Nishioka, and M. Fujita, “Ultra-broadband optical parametric chirped-pulse amplification using an Yb: LiYF4 chirped-pulse amplification pump laser,” Opt. Express 15, 5018–5023 (2007). [CrossRef] [PubMed]
7. 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]
8. A. Baltuška, T. Fuji, and T. Kobayashi, “Visible pulse compression to 4 fs by optical parametric amplification and programmable dispersion control,” Opt. Lett. 27, 306–308 (2002). [CrossRef]
9. A. Shirakawa, I. Sakane, M. Takasaka, and T. Kobayashi, “Sub-5-fs pulse generation by pulse-front-matched noncollinear optical parametric amplification,” Appl. Phys. Lett. 74, 2268–2270 (1999). [CrossRef]
11. F. Röser, D. Schimpf, O. Schmidt, B. Ortaç, K. Rademaker, J. Limpert, and A. Tünnermann, “90 W average power 100 µJ energy femtosecond fiber chirped-pulse amplification system,” Opt. Lett. 32, 2230–2232 (2007). [CrossRef] [PubMed]
12. A. Killi, A. Steinmann, G. Palmer, U. Morgner, H. Bartelt, and J. Kobelke, “Megahertz optical parametric amplifier pumped by a femtosecond oscillator,” Opt. Lett. 31, 125–127 (2006). [CrossRef] [PubMed]
13. C. Schriever, S. Lochbrunner, and E. Riedle, “Tunable 20 fs red pulses with up to 200 nJ energy from a 2 MHz Yb-doped fiber laser oscillator/amplifier system,” in Conference on Lasers and Electro-Optics/Quantum Electronics and Laser Science Conference and Photonic Applications Systems Technologies 2007 Technical Digest (Optical Society of America, Washington, DC, 2007), CMT5.
14. C. Aguergaray, T. V. Andersen, D. N. Schimpf, O. Schmidt, J. Rothhardt, T. Schreiber, J. Limpert, E. Cormier, and A. Tünnermann, “Parametric amplification and compression to ultrashort pulse duration of resonant linear waves,” Opt. Express 15, 5699–5710 (2007). [CrossRef] [PubMed]
15. J. Dudley and S. Coen, “Fundamental limits to few-cycle pulse generation from compression of supercontinuum spectra generated in photonic crystal fiber,” Opt. Express 12, 2423–2428 (2004). [CrossRef] [PubMed]
16. F. Tavella, A. Marcinkevicius, and F. Krausz “90 mJ parametric chirped pulse amplification of 10 fs pulses,” Opt. Express14, 12822–12827 (2006). T.M. Fortier, D.J. Jones, and S.T. Cundiff, “Phase stabilization of an octace-spanning Ti:sapphire laser,” Opt. Lett.28, 2198–2200, (2003). [CrossRef] [PubMed]
17. H. Zheng, J. Wu, H. Xu, K. Wu, and E. Wu, “Generation of accurately synchronized pump source for optical parametric chirped pulse amplification,” Appl. Phys. B 79, 837–839, (2004). [CrossRef]
18. G. P. Agrawal, Nonlinear Fiber Optics (Academic Press, San Diego, 2001).
19. J. Limpert, F. Röser, T. Schreiber, and A. Tünnermann, “High-power ultrafast fiber laser systems,” IEEE J. Sel. Top. Quantum Electron 12, 233–244 (2006). [CrossRef]
20. T. Schreiber, F. Röser, O. Schmidt, J. Limpert, R. Iliew, F. Lederer, A. Petersson, C. Jacobsen, K. Hansen, J. Broeng, and A. Tünnermann, “Stress-induced single-polarization single-transverse mode photonic crystal fiber with low nonlinearity,” Opt. Express 13, 7621–7630 (2005). [CrossRef] [PubMed]