We have generated femtosecond pulses with a peak power as high as 0.7 GW at the repetition rate of 100 kHz from a downchirped-pulse amplification (DPA) Ti:sapphire laser. For the high-energy amplification with high repetition rate, we employ a regenerative amplifier, acousto-optically switched and pumped by a Q-switched green laser. The DPA-based dispersion control is achieved up to the third-order term by use of a prism-grating stretcher and a glass compressor. We have obtained 28 μJ, 39 fs laser pulses with a compression efficiency of 95 %. Temporal and spatial characteristics of the laser pulses are investigated.
© 2006 Optical Society of America
Since 1990s, Ti:sapphire laser technology has provided practical solutions of solid-state femtosecond light source with high average power as well as high peak power. Especially, high-repetition-rate Ti:sapphire amplification laser systems are a good tool for basic and applied sciences because of the moderate average and peak powers and their compactness. Typical high-repetition-rate Ti:sapphire amplification systems using electro-optic pulse selection method operate at the range from 10 Hz to 20 kHz [1, 2, 3] with an energy of ~mJ, whereas those using acousto-optic pulse selection method can operate at the range from 100 kHz to 250 kHz with an energy of ~ μJ. Recently, an electro-optically switched regenerative amplification has been reported at 200 kHz with a Yb:KY(WO4)2 amplifier using a β-barium borate Pockels cell. However, the piezo-electric ringing in a Pockels cell is still a fundamental problem for higher-repetition-rate operation and this method has not been demonstrated with Ti:sapphire systems yet. On the other hand, the multipass amplification of laser pulses from a cavity-dumped Ti:sapphire oscillator has been also demonstrated up to 40 kHz with sub-μJ energy.
For the 100-250 kHz amplification system, the acousto-optically switched regenerative amplifier suggested by Norris  has been a conventional configuration. Several features of the lasers based on this configuration can be described as follows. First, the high-repetition-rate pulse injection and dumping is achieved by the cavity dumper composed of a Brewster-angle Bragg cell and an rf driver. Second, they are pumped by a continuous-wave (CW) green laser and intracavity Q-switched for the suppression of prelasing. Third, the dispersion control of the amplification system has been achieved either without or with a chirped-pulse amplification (CPA) technique. In the case of non-CPA systems[4, 8], pulses are naturally broadened and compressed by use of prism pairs, whereas, in the case of CPA system, transmission gratings are used for both stretching and compression. Adaptive control has been also applied for the non-CPA system to compress the laser pulse down to 35 fs . The compression efficiency for those systems is 60-80%.
Despite higher repetition rate, 100-250 kHz systems have no advantage with average power because of much lower energy per pulse or even lower peak power: the highest energy and peak power reported at 100 kHz is 7 μJ and 0.2 GW  so far. Thus, the enhancement of the energy per pulse will make this kind of lasers be a more valuable tool not only for high-average-power applications such as micromachining and medical surgery but also for high-peak-power applications such as high-order harmonic generation  and above-threshold ionization study. For the generation of higher peak and average powers at 100 kHz, we demonstrate an efficient acousto-optically switched regenerative amplification system, which is modified from the conventional scheme in terms of pumping source and dispersion compensation method. First, we have employed a Q-switched pump laser, instead of a CW one, to obtain higher pumping flu-ence and simplify the cavity by removing an intracavity Q-switcher. Use of a pulsed source allows more efficient pumping than CW source because the portion of the pump energy participating in the population inversion process is higher than that of CW source. Also, accumulated dispersion and spectral narrowing due to an intracavity Q-switcher can be reduced. Even though this method cannot be used for the repetition rate as high as 250 kHz because of the limitation of pumping repetition rate, it is a good approach in the range from multi-tens of kHz to 100 kHz.
Second, we have applied downchirped-pulse amplification (DPA) technique for higher compression efficiency. In a DPA configuration, the laser pulse is stretched with negative dispersion elements (down-chirp), such as prism pair and grating pair, and then compressed by positive dispersion element which is just an optical glass block. The concept of DPA is essentially the same as that of CPA, but it was recently demonstrated at a 10 kHz system. The main advantage of this technique is high compression efficiency (>95 %) and simple compressor alignment. Since the pulse duration and phase are insensitive to the compressor alignment, a DPA technique is also a good approach for the amplification of carrier-envelope phase stabilized pulses [13, 14]. It should be noted, however, that a material compressor is more suitable to relatively low-peak-power pulses than high-peak-power pulses because of the probable third-order nonlinear effect at the compressor. Therefore, the DPA technique is an optimal method of dispersion control for the development of a 100-kHz femtosecond laser whose peak power is relatively low compared to those of conventional kHz lasers. In this paper, we describe the configuration and operational characteristics of a 100-kHz DPA Ti:sapphire laser system producing the energy as high as 28 μJ and the pulse duration as short as 39 fs.
2. Design of a DPA regenerative amplification laser
2.1. Regenerative amplifier
Basic design parameters of an amplifier, such as output energy, the spot size of a pump source at a gain medium and the number of roundtrips, are mainly determined by a pumping energy. We have utilized a diode-pumped, frequency-doubled, Q-switched Nd:YAG laser (HSQG5000, Goldenlight Co.) operating at 10-100 kHz, in contrast to a CW pump laser utilized in previous 100-250 kHz regenerative amplifiers[4, 8, 9]. The energy of the pump laser is 200 μJ at 100 kHz and its pulse duration is ~400 ns. When the spot size at the Ti:sapphire crystal is 150 μm in diameter, the pumping fluence becomes 1.1 J/cm2, which is slightly higher than saturation fluence of Ti:sapphire (0.9 J/cm2). A regenerative scheme is more suitable than a multi-pass one because of the small spot size at the crystal and a large number of passes (much larger than 10) through the gain medium. When designing a femtosecond regenerative amplifier, we have to carefully consider the dispersion characteristic of the amplifier because the large number of roundtrips inside the amplifier can significantly change the stretcher and compressor configurations.
A home-made mirror-dispersion-controlled sub-10-fs Ti:sapphire laser is used as a front-end femtosecond oscillator. The laser pulses have the energy of 5 nJ with a repetition rate of 90 MHz. Calculation with above pumping fluence, based on Frantz and Nodvick model, shows that the amplification in saturation regime allows the output energy of 55 μJ and that the number of roundtrips required for the gain saturation is 30, where the experimentally measured seed energy loss of 90 % at the stretcher is included. Actually, we estimate the output energy to be 33 μJ considering that the dumping efficiency of our fast acousto-optic modulator (fused silica Bragg cell) is 60%. The nonlinear phase (B-integral) accumulated inside the amplified is about 0.7 rad when the pulse is stretched to 20 ps. Thus, the stretched pulse duration of 20 ps should be safe against optical damage as well as nonlinear effect.
To realize the calculated parameters, we have designed a regenerative amplifier similar to a X-folded cavity-dumped laser oscillator as shown in Fig. 1(a). Both Ti:sapphire crystal and Bragg cell (BC of Fig. 1) are installed with a Brewster angle in a confocal geometry. The four curved mirrors (DM1, DM2, M1, and M2) in Fig. 1(a) have the same focal length of 10 cm and their folding angles are determined so as to compensate for the astigmatism induced by the Brewster-cut Ti:sapphire crystal and Bragg cell. The seed beam and amplified beam are diffracted at the Bragg cell in the vertical plane with a small angle (~50 mrad). They are either injected to or dumped from the cavity via a folding mirror (FM of Fig. 1). Double pass scheme at the Bragg cell is used for higher dumping efficiency. The crystal temperature is kept ~5 °C by a thermo-electric cooler to minimize a thermal lens effect. The crystal mount holds the top and the bottom of the Ti:sapphire crystal whose vertical dimension is 2 mm. It is expected that the thin crystal enhances the thermal dissipation from the crystal to the cooled mount. To maximize the absorption at the crystal, we have installed a re-focusing mirror (RM of Fig. 1) for the pump beam. The pump beam absorption is enhanced from 75 % to 94 % with this mirror. The chirped mirrors (CM1-4 of Fig. 1) are for the dispersion-compensated broadband amplification. Net group-delay dispersion (GDD) per roundtrip of the regenerative amplifier is only -5 fs2, which makes the accumulated GDD insensitive to the number of roundtrips. The dispersion-compensated cavity is also very useful for the DPA configuration because pulses are automatically shortened and can cause damage on the optical components in case of a normal positive-dispersion cavity. The Faraday isolator and polarizing beam splitter (FI and PBS of Fig. 1) are for the discrimination between injected and dumped beams.
Injection and dumping signals for driving Bragg cell as well as the pump laser trigger signal are synchronized to the mode-locked pulse train generated by the laser oscillator. Figure 2 illustrates the synchronization scheme. The 100 kHz trigger signals are generated by frequency dividing the 90 MHz RF signal from the oscillator pulse train with a synchronous countdown and a digital delay pulse generator. The rf signal of the Bragg cell is also generated using the harmonics of 90 MHz pulse train. As a result, injection and dumping triggers, pump laser trigger, and RF signal of Bragg cell are all synchronized to the mode-locked pulse train.
2.2. Dispersion control of DPA system
Dispersion calculation is very important with a DPA laser design because the compressor, i.e., a glass block with a finite length, has no other degree of freedom than the combination of different lengths for the control of laser pulse durations. To minimize the inconvenience of changing the stretcher parameters or compressor length, we have to accurately calculate the dispersion of all the optical elements in the system. It has been already known that a stretcher consisting of a prism pair and a grating pair is good for the dispersion compensation up to the third order term in a DPA configuration. Here, we utilize a SF10 prism pair combined with a 600 grooves/mm ruled grating pair as a stretcher and a SF10 glass block as a compressor, respectively.
Table 1 shows the calculated dispersion of the laser system up to the fourth-order term at the wavelength of 800 nm. The material dispersion of the prisms is included in the calculation for the prism pair. The prism separation is given by tip-to-tip space while the grating separation is by the space along the beam path. Additionally considered optical elements are Faraday isolator composed of a TGG rotator (thickness of 14 mm) and two SF2 polarizing beam splitters (thickness of 12.7 mm), a fused silica Bragg cell (path length of 3.7 mm), Ti:sapphire crystal (path length of 2.3 mm), and chirped mirrors (total GDD of -520 fs2 per roundtrip). The dispersion compensation is considered up to the third order term as found in Table 1. As for the dispersion characteristics of chirped mirrors, the net value of the third-order dispersion (TOD) is zero, whereas the fourth-order dispersion (FOD) has a small but unknown positive value. However, we set the FOD values of the chirped mirrors as zero for the convenience of calculation because the FOD is dominated by the prism pair and its compensation is not attempted in this system. Based on the parameters shown in Table 1, we have designed and installed a stretcher and a compressor for DPA scheme.
3. Amplification and compression of the laser pulses
The 5-nJ femtosecond pulses generated in the master oscillator were strongly downchirped by a prism-grating stretcher, as shown in Fig. 1(b), to be stretched up to 20 ps. Because of the large incidence angle to the grating of the stretcher (70°), the actual energy of the seed pulse out of the stretcher was only ~0.4 nJ. The laser spectra from the master oscillator and after the stretcher are shown in Fig. 3. The spectral bandwidth of the stretcher was 180 nm, limited by the size of the second prism in Fig. 1(b). The spectral loss in the range from 800 nm to 860 nm was also observed due to the grating pair.
For the mode matching between the seed beam and the free-running beam of regenerative amplifier, the seed beam was down-collimated using a 2:1 telescope and injected into the regenerative amplifier through the Faraday isolator. The buildup time inside the regenerative amplifier was minimized by cavity alignment and seed beam alignment. The dumped output energy of amplified pulses was maximized by controlling the trigger timing and the phase of rf signal to the Bragg cell.
The number of roundtrips for the saturation of the regenerative amplifier was 30-33, de- pending on the cavity and seed pulse alignment. With the pump energy of 200 μJ, we have obtained an amplified energy of 30 μJ at maximum. The main limiting factor for a higher output energy is a relatively low dumping efficiency of our Bragg cell (60%). Both the number of roundtrips and the output energy were closely matched to the calculated results. The rms jitter of shot-to-shot energy of the amplified pulses is about 3%.
Amplified laser spectrum is shown in Fig. 4(a). The spectral bandwidth in full-width at half maximum (FWHM) is 36 nm, centered at 780 nm due to strong peak at short-wavelength part of the seed spectrum. This spectrum supports the transform-limited pulse duration as small as 20 fs in FWHM. Two strange spikes at around 770 nm cannot not be clearly identified, but we believe they are from an optical defect of some optics, such as reflectivity modulation and dispersion modulation, inside the amplifier. A nonlinear effect such as self-phase modulation does not explain these two peaks because they still appear even in a low-power operation.
Amplified pulse was finally up-collimated using a 1:2 telescope to avoid any nonlinear effect at the optical glass and compressed by two anti-reflection-coated SF10 blocks with an efficiency of 95 % (total three passes). The final beam size was ~4 mm in Gaussian diameter and the resultant output energy was 28 μJ after compression.
The pulse duration was monitored at a frequency-resolved optical gating (FROG) apparatus and minimized by changing both the path length through the SF10 medium and the grating separation of the stretcher. The shortest pulse duration measured was 39 fs in FWHM, as shown in Fig. 4(b), which was 2 times larger than the transform-limited one. As a result, the peak power of the pulses reached 0.7 GW at 100 kHz. To our knowledge, those energy and peak power are the highest ever reported with a Ti:sapphire amplification system operating at 100-kHz-level repetition rates.
We investigated the limiting factor of further pulse compression. The experimental parameters for generating the shortest pulse duration were slightly different from those seen in Table I. The grating separation was about 63 mm and the total path length at the SF10 block was 380 mm. The center wavelength of the spectrum, i.e., 780 nm, was also different from that used for the original design, i.e., 800 nm. Recalculation of the residual high-order dispersion along with those three different parameters showed that the residual TOD and FOD were -8700 fs3 and -269000 fs4, respectively. The phase curve with those values is plotted as solid line in Fig. 4(a), while the experimental phase curve measured with FROG is shown as square dots. Two curves are relatively well matched to each other in the central part of the spectrum. It is clear that the main contribution for the pulse broadening was the FOD even though the small TOD still remained. As for the FOD, it is not easy to compensate for in a prism-grating stretcher. However, it is possible to minimize this term by properly choosing prism material, grating groove number and compressor material and by searching for another combination of dispersion parameters such as grating angle, separation, prism separation, and compressor length. In addition, spectral shaping and adaptive phase control technique can be used for better compression capability.
Spatial beam profile was near Guassian as shown in Fig. 5(a). It was monitored before the compressor telescope because of the aperture size limitation with our charge-coupled device, but there was no degrade of beam quality after the compressor. The focusing capability was quantified by measuring M2 value of the output pulse. Figure 5(b) shows that M2 values for the horizontal direction and the vertical direction are 3.6 and 5.3, respectively. It is believed that the main cause of the relatively high M2 value is a thermal lensing effect at the Ti:sapphire crystal. Actually, we observed that the beam divergence of unseeded regenerative amplifier was slightly higher than that of seeded one, which seemed to be due to the difference in thermal loading. Nevertheless, the thermal problem was not seriously considered in this work. Thus, further cooling along with cavity adjustment will reduce this parameter down to 2, which is a general requirement for many industrial lasers.
We demonstrated the downchirped regenerative amplification of femtosecond Ti:sapphire laser pulses at 100 kHz repetition rate. For higher energy and peak power, the acousto-optic regenerative amplifier was pumped by a pulsed green laser instead of a CW laser and the system dispersion was compensated for up to the third order term. The regenerative amplifier cavity was dispersion-compensated by broadband chirped mirrors. The DPA compressor allowed a compression efficiency of 95 % along with simple alignment. As a result, we obtained 28-μJ 39-fs (0.7-GW) pulses with a good beam quality.
Our 100 kHz high-power DPA laser system is a useful femtosecond light source for a variety of research issues such as high-order harmonic generation, carrier-envelope phase locking, and efficient micro-machining.
Note: A patent on this laser amplification system is pending.
This work was supported by the Ministry of Commerce, Industry and Energy of Korea through the Industrial Technology Infrastructure Building Program. The authors thank Dr. Nak-Hyun Seong for helping the installation of the FROG.
References and links
6. N. Zhavoronkov, H. Maekawa, H. Okuno, and K. Tominaga, “All-solid-state femtosecond multi-kilohertz laser system based on a new cavity-dumped oscillator design,” J. Opt. Soc. Am. B 22, 567 (2005). [CrossRef]
7. D. Strickland and G. Mourou, “Compression of amplified chirped optical pulses,” Opt. Commun. 55, 447 (1985). [CrossRef]
8. See the data sheet of RegA series of Coherent Incorporation (http://www.coherentinc.com).
9. J.-K. Rhee, T. S. Sosnowski, T. B. Norris, J. A. Arns, and W. S. Colburn, “Chirped-pulse amplification of 85-fs pulses at 250 kHz with third-order dispersion compensation by use of holographic transmission gratings,” Opt. Lett. 19, 1550 (1994). [CrossRef] [PubMed]
10. F. Lindner, G. G. Paulus, F. Grasbon, A. Dreischuh, and H. Walther, “Dispersion control in a 100-kHz-repetition-rate 35-fs Ti:sapphire regenerative amplifier system,” IEEE J. Quantum Electron. 38, 1465 (2002). [CrossRef]
11. F. Lindner, W. Stremme, M. G. Schatzel, F. Grasbon, G. G. Paulus, H. Walther, R. Hartmann, and L. Struder, “High-order harmonic generation at a repetition rate of 100 kHz,” Phys. Rev. A 68, 013814 (2003). [CrossRef]
12. D. M. Gaudiosi, A. L. Lytle, P. Kohl, M. M. Murnane, H. C. Kapteyn, and S. Backus, “11-W average power Ti:sapphire amplifier system using downchirped pulse amplification,” Opt. Lett. 29, 2665 (2004). [CrossRef] [PubMed]
13. M. Kakehata, H. Takada, Y. Kobayashi, and K. Torizuka, “Carrier-envelope-phase stabilized chirped-pulse amplification system scalable to higher pulse energies,” Opt. Express 12, 2070 (2004), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-12-10-2070 [CrossRef] [PubMed]
14. Y. S. Lee, J. H. Sung, C. H. Nam, T. J. Yu, and K.-H. Hong, “Novel method for carrier-envelope-phase stabilization of femtosecond laser pulses,” Opt. Express 13, 2969 (2005),http://www.opticsexpress.org/abstract.cfm?URI=OPEX-13-8-2969 [CrossRef] [PubMed]
15. K.-H. Hong, S. Kostritsa, T. J. Yu, J. H. Sung, Y.-C. Noh, D.-K. Ko, and J. Lee, “Downchirped regenerative amplification of femtosecond laser pulses at 100 kHz repetition rate,” UFO/HFSW 2005 (Nara, Japan, Sept. 25-30, 2005) F2–3.
16. L. M. Frantz and J. S. Nodvick, “Theory of pulse propagation in a laser amplifier,” J. Appl. Phys. 34, 2346 (1963). [CrossRef]
17. T. Imran, K.-H. Hong, T. J. Yu, and C. H. Nam, “Measurement of the group-delay dispersion of femtosecond optics using white-light interferometry,” Rev. Sci. Instrum. 75, 2266 (2004). [CrossRef]