We have developed an all-solid-state 5-kHz Ti:sapphire Laser System, which produces 22-fs, 0.2-TW pulses. An average power of 22.2 W is the highest ever obtained in ultrashort laser sources. The serious thermal lensing due to high power pumping in a small area of the Ti:sapphire crystal is controlled successfully by a stable quasi-cavity with two concave mirrors.
©1999 Optical Society of America
Progress in the developement of high-peak-power ultrashort pulse lasers has opened up novel areas of physics of materials under a high optical electric field. Among them the mechanism of high-order harmonics has been investigated extensively for several years and has allowed accessibility to wavelengths below the water window[1, 2].
Harmonics are attractive light sources for nonlinear, time-resolved spectroscopy in gases and solids in the XUV and soft X-ray wavelength region, because they inherit good spatial coherence and ultrashort time duration from the fundamental beam. In fact, two photon absorption in He at the 9th harmonic of a Ti:sapphire laser was observed and applied to the pulsewidth measurement in the XUV region. The success of this experiment apparently resulted from the focusability and high intensity of harmonics. It takes, however, a few tens of minutes to obtain one autocorrelation trace of harmonics at a 10-Hz repetition rate. We developed a kHz, 0.66-TW Ti:sapphire laser  with a pulsewidth of 21-fs which employs five green lasers pumped by arc-lamps. The repetition rate of 1 kHz is, however, not so high. Compared with synchrotron orbit radiation (SOR) as the light source at XUV wavelengths, further efforts to increase the repetition rate are required to shorten the data acquisition time. And diode-laser-pumped green lasers are obviously favorable for long time stability. Several works on chirped pulse amplification (CPA) systems at a high repetition rate were reported[6, 7, 8, 9] and 1.1-TW pulses were generated at 1-kHz repetition rate. However, the peak powers were 1.4 GW  at 10kHz and 2 GW  at 5 kHz, which are not high enough for harmonic generation. This restriction comes from thermal lensing increasing proportionally to the repetition rate when a constant pump fluence is kept and also from the lack of pumping lasers.
In this paper, we report on a Ti:sapphire amplifier system consisting of regenerative and multi-pass amplifiers with a peak power of 0.2 TW and a pulse width of 22 fs at 5 kHz. An average power of 22.2 W is the highest ever obtained, to our knowledge, in any ultrashort pulsed laser source at any repetition rate. Average powers of 19 W and 21 W were recorded in the past at 10-Hz and 1-kHz respectively. Thermal lensing is successfully controlled by using a stable quasi-cavity with two concave mirrors in a multipass amplifier. The pumping sources for the amplifiers are two diode-pumped, Q-switched YAG lasers newly developed which can produce 83 W at 5 kHz and 100 W at 10kHz in the second harmonic. This all-solid-state system ensures a long term stability.
2 System configuration and results
Fig. 1 shows a schematic diagram of the CPA system, which consists of a mode-locked oscillator, an Offner-type stretcher, a regenerative amplifier, a 4-pass amplifier, a 1-pass amplifier, and a compressor. The outlines for each part are almost the same as ref.  except for the two amplifiers.
2.1 Pumping laser
All Ti:sapphire crystals in the amplifier chain in this system are pumped by intracavity, frequency-doubled, diode-pumped, Q-switched Nd:YAG lasers (MITSUBISHI ELECTRIC CORP. MEL-Green100) with a novel pumping scheme; Close-coupled Internal Diffusive Exciting Reflector (CIDER). These lasers generate a green power of above 100 W at a 10kHz repetition rate. In the system, two green lasers are operated at 5 kHz with output powers of 56 W and 83 W respectively. The former is used for pumping the regenerative amplifier and the 4-pass amplifier while the latter is for a final single-pass amplifier.
The regenerative amplifier consists of a conventional 4-mirror cavity having two flat mirrors and two concave mirrors with radii of curvature of 200 mm (CC1) and 300 mm (CC2) respectively. A 12-mm long, 0.15-wt.% dope Ti:sapphire crystal is placed between the concave mirrors.
The cavity of the regenerative amplifier contains a Pockels cell driven by a triode-unit with a pre-bias module (MEDOX), which is cooled by water and can be operated at a 10-kHz repetition rate. It also contains a LAH64 (OHARA Optical Glass Inc.) prism pair for the compensation of high order dispersions as described in ref.  and thin film polarizer etalon for avoiding the gain narrowing. The position of the Ti:sapphire crystal relative to the concave mirrors are carefully adjusted to compensate thermal lensing. The focal length of thermal lens is estimated to be 18 mm when the Ti:sapphire crystal is pumped by 120-µm diameter, 17-W green beam . From the analysis of cavity stability by the ABCD matrices, the thermal lens is placed 112 mm apart from CC1 while two concave mirrors are separated by 251 mm.
The output power of the regenerative amplifier is 750mW at 17 W pumping. A Faraday rotator is placed to isolate the 4-pass and single pass amplifiers from the regenerative amplifier and the oscillator.
For multi-pass amplifiers, thermal lensing becomes more serious at increasing pump power[15, 16]. According to the analysis in ref. , the focal length of the thermal lens f therm is inversely proportional to P o/(P o:the average power of the absorbed beam, r o:the radius of the pumping region) which can be rewritten as fluence×repetition rate. Thus the thermal focusing at 5 kHz becomes 5 times more severe than that at 1 kHz if we maintain the same fluence so as to obtain the same gain of the amplifiers. Because we can treat a rod as a spherical lens as far as the parabolic approximation of the thermal aberration is valid, an appropriate configuration for the Gaussian beam can be designed in multi-pass amplification[4, 15]. In the 1-kHz system of ref. , a bow-tie configuration with flat turning mirrors was used at a 20-W pump level by taking the thermal lens in the crystal to reconstruct the beam for each pass. The length of one pass L in the bow-tie configuration is inversely proportional to a repetition rate when the same fluence is kept. Then the pass length at 5 kHz becomes about 0.1 m in a fluence of ref.  (2.5J/cm2). This is too short to overlap the beam paths. For this reason we adopt a nonconfocal geometry of two concave mirrors in this experiment. The configuration for multi-pass amplification is formed by two concave mirrors and turning mirrors including a thermal lens at the center of the cavity as shown in Fig. 2(a).
To reconstruct the Gaussian beam for each pass, this quasi-cavity has to be in a stable condition shown by the inequality as follows.
where α=L a/f cc, β=L cc/(2f cc) Δ=f cc/(2f therm) and L a:the distance between the concave mirror and the end mirror, f cc:the focal length of concave mirror, L cc:the distance between the concave mirrors.
As a special condition, we can easily see from this inequality that the cavity remains stable for increasing L cc from 2f cc to 2f cc+4f therm when L a is equal to f cc. This stable range ensures the control of the beam size by changing the separation of the concave mirrors for the variation of thermal focusing. Although the stable range of the L cc becomes smaller as L a increases from 2f cc, we still can find the stable condition around L cc≈2f cc. The radius of the amplified beam w 1 (the half width at 1/e2 intensity) should be matched to the pump beam in order to obtain a sufficient fluence. The radius of the amplified beam can be varied by a slight change of L cc without any geometrical difficulty of the configuration of the amplifier, even if f therm is not correctly known. In Fig. 3 the L cc versus f therm is shown for the several sizes of the amplified beam. In this calculation we set f therm=0.25 m and L a=0.6 m. From this figure, we see that w 1 can be adjusted with a change of L cc by several centimeters while f therm varies from 10mm to 60mm.
Based on this principle, the radius of curvature of the concave mirrors (CC3’s in Fig. 2)(a) in the stage of 4-pass and single-pass amplifiers are determined to 0.5 m (f cc=0.25 m) and L cc is adjusted from about 260mm to 300mm so as to find the good spatial mode and the maximum output power of the amplified beam. Because this quasi-cavity of the 4-pass amplifier is vertically folded, there are an additional two flat mirrors (FM in Fig. 2(a)) to ensure that the folding mirrors for each pass are placed correctly. We show the picture of this amplifier in Fig. 2(b).
The measured size of the pump beam w 0(the radius at 1/e2 intensity) is about 200 µm, while the beam size of the amplified beam w 1 in the Ti:sapphire crystal is always set to 120 µm. The change of thermal lensing is accomplished by adjusting L cc. The parabolic approximation of the thermally induced change of refraction index is valid in Ti:sapphire because w 0 is much larger than w 1. We use a 0.05-wt.% doped, 30-mm long, 5-mm diameter, Brewster-cut Ti:sapphire in a copper holder cooled to 10°C by circulating a mixture of water and ethylene glycol in the 4-pass amplifier.
The power of the amplified beam at each 4 pass is shown in Fig. 4. An absorbed power of the pump beam in Ti:sapphire at this stage is 32.5 W and the input power of the amplified beam is 580mW, which comes from the regenerative amplifier with insertion loss of the isolator. At the end of this stage, an output power of 12.8 W is obtained with an extraction efficiency of 38% after we adjusted the separation of the concave mirrors to 580mm (including the length of the Ti:sapphire). Although it seems in Fig. 4 that the gain saturation does not fully occur at the 4th pass, we did not add another pass because poor beam overlap may cause the degradation of beam quality and damage to the Ti:sapphire crystal. At the final single-pass amplifier, the output power increases almost linearly with the pump power, and reaches 37.7 W when a pump beam of 66.1 W is absorbed in Ti:sapphire, resulting a 38% extraction efficiency. This result is also shown in Fig. 4. Although the power slightly decreases in several hours, we can reproduce the power by the fine adjustment of the direction of the pumping beam. Shot to shot stability and the beam pointing of the pulses have not been measured.
The focus ability of the amplified beam was measured at this stage. The beam was attenuated by the Fresnel reflection of the wedge plate and the attenuator, and was softly focused, giving the waist about 200 µm. We obtained M2=1.3 in the tangential plane to concave mirrors and 1.2 in the sagittal plane.
2.3 Management of the dispersion, Beam quality
The output beam from the final amplifier is magnified to about 40mm in diameter by a telescope consisting of a convex and a concave mirrors and is then sent to the grating pair compressor. The mismatch of the incident angles to the gratings in the stretcher and in the compressor, combined with the prism pair in the regenerative amplifier generates the shortest pulses by balancing higher order dispersion. The temporal profile and phase of the output pulse are obtained from the inverse-Fourier transform of the measured spectrum. The dispersion was calculated by ray tracing as shown by the solid curve and the dotted curve of Fig. 5 respectively. The detailed analysis for this matter is described in ref. .
The compressed pulse was characterized by frequency resolved optical gating of second-harmonic generation (SHG-FROG) as also shown in Fig. 5. The retrieved intensity profile (open circles) and the phase (open squares) agree well with the calculated profiles except at the edges. Although the measured phase does not give the transform limit, a 22-fs pulse width (FWHM) with a 22.2-W average power is obtained after the compressor at 5 kHz, resulting a peak power of 0.2 TW.
Thermal load on the surface of the gratings may degrade the focus ability. The value of M2 after the compressor is 1.4, both in the s- and p-plane to the grating, and is larger than those before the compressor. The gratings used in the compressor are made by Richardson Grating Laboratory (35-73-17-360, S/N2917 series on Zerodur substrate).
We have developed an all-solid-state 5-kHz 0.2TW Ti:sapphire laser system. To compensate for thermal lensing in Ti:sapphire, a non-confocal geometry of two concave mirrors has been successfully adopted. A 22.2-W average power is the highest ever obtained at any repetition rate in ultrashort pulse amplifiers. A 0.2-TW peak power is also the highest at any repetition rate higher than 1 kHz. This laser will be soon applied to harmonic generation and research on solid state spectroscopy in the XUV region.
References and links
1. M. Schnürer, Ch. Spielmann, P. Wobrauschek, C. Streli, N. H. Burnett, C. Kan, K. Ferencz, R. Koppitsch, Z. Cheng, T. Brabec, and F. Krausz, “Coherent 0.5-keV X-Ray Emission from Helium Driven by a Sub-10-fs Laser,” Phys. Rev. Lett. 80, 3236–3239 (1998) [CrossRef]
2. M. Schnürer, Z. Cheng, M. Hentschel, G. Tempea, P. Kálmán, T. Brabec, and F. Krausz, “Absorption-Limited Generation of Coherent Ultrashort Soft-X-Ray Pulses,” Phys Rev. Lett. 83, 722–725 (1999) [CrossRef]
3. Y. Kobayashi, T. Sekikawa, Y. Nabekawa, and S. Watanabe, “27-fs extreme ultraviolet pulse generation by high-order harmonics,” Opt. Lett. 23, 64–66 (1998) [CrossRef]
4. Y. Nabekawa, Y. Kuramoto, T. Togashi, T. Sekikawa, and S. Watanabe, “Generation of 0.66-TW pulses at 1 kHz by a Ti:sapphire laser,” Opt. Lett. 23, 1384–1386 (1998) [CrossRef]
5. D. Strickland and G. Mourou, “Compression of amplified chirped optical pulses,” Opt. Commun. 56, 219–221 (1985) [CrossRef]
6. Y. Bagnoud and F. Salin, “1.1 Terawatt, kilohertz femtosecond laser,” in Conference on Laser and Electro Optics of 1999 OSA Techenical Digest Series (Optical Society of America, Wasington, D. C.,1999), CTuD3, 71–72
10. K. Yamakawa, M. Aoyama, S. Matsuoka, T. Kase, Y. Akahane, and H. Takuma, “100-TW sub-20-fs Ti:sapphire laser system operating at a 10-Hz repetition rate,” Opt. Lett. 23, 1468–1470 (1998) [CrossRef]
12. S. Konno, S. Fujikawa, T. Kojima, and K. Yasui, “High brightness 127W green beam generation by intracavity-frequency-doubling of diode pumped Nd:YAG laser,” Conference on Laser and Electro Optics of 1999 OSA Techenical Digest Series (Optical Society of America, Wasington, D. C.,1999), CWG1, 301
13. C. P. J. Barty, G. Korn, F. Laksi, C. Rose-Petruck, J. Squier, A. -C. Tien, K. R. Wilson, V. V. Yakovlev, and K. Yamakawa, “Regenerative pulse shaping and amplification of ultrabroadban optical pulses,” Opt. Lett. 21, 219–221 (1996) [CrossRef] [PubMed]
14. M. E. Innocenizi, H. T. Yura, C. L. Fincher, and R. A. Fields, “Thermal modeling of continuous-wave end-pumped solid-state lasers,” Appl. Phys. Lett. 56, 1831–1833 (1990) [CrossRef]
15. F. Salin, C. L. Blanc, J. Squier, and C. Barty, “Thermal eigenmode amplifiers for diffraction-limited amplification of ultrashort pulses,” Opt. Lett. 23, 718–720 (1998) [CrossRef]
17. K. W. DeLong, Rick Trebino, J. Hunter, and W. E. White, “Frequency-resolved optical gating with the use of second-harmonic generation,” J. Opt. Soc. Am. B112206–2215 (1994)