1. Introduction

Thin disk lasers have attracted much interest for years [1] owing to their easy and efficient heat management which promises a high beam quality. Especially in ultra-short pulse operation, thin disk regenerative amplifiers are promising tools for a variety of applications e.g. micromachining and waveguide writing in transparent materials. As active media in such systems Ytterbium doped monoclinic potassium double tungstates turned out to support ultra short pulses while offering high absorption and emission cross sections [2], partly due to the strong anisotropy of these biaxial crystals. The KLu(WO₄)₂ (KLuW) host is actually predestined for doping with Yb because of the close ionic radii and masses of Yb and Lu. This allows high Yb-doping levels with low defect formation probability and without substantial fluorescence quenching. Moreover, the Yb-dopant affects only weakly the thermal conductivity of the host [3].

A regenerative Yb:KLuW thin disk amplifier was presented recently, producing sub 100 fs pulses [4] by using intracavity self-phase-modulation to broaden the spectrum and thus counteracting the gain narrowing effect [5]. However less work has been done to investigate the performance of KLuW thin disks in a chirped pulse amplification (CPA) system [6] which would have the advantage that the achievable pulse duration is decoupled from nonlinear effects.

In this paper we present a regenerative Yb:KLuW thin disk amplifier based on the CPA concept operating in the wavelength region around 1025 nm. At a repetition rate of 20 kHz an output energy of 571 µJ corresponding to an average power of 11.42 W has been achieved. These pulses could be dechirped to a pulse duration of 197 fs assuming a deconvolution factor of 2.16. Owing to the cavity design no limitation due to bifurcation was observed despite the high extracted pulse energy [7].

2. Experimental setup

The experimental setup is shown in Fig. 1 and consisted of a fiber based seed system, a regenerative amplifier cavity designed for 12 passes through the thin disk per roundtrip and a transmissive grating-prism compressor (GRISM) [8]. As the setup was similar to the one shown in Ref. 9 and 10 just the differences are described in the following. In the present experiment, the seed pulses had an energy of 4 nJ and a pulse duration of 43 ps. Dechirping resulted in a pulse duration of 170 fs. The central wavelength of the seed source was 1026 nm. As active medium an N_p oriented Yb:KLuW thin disk with a doping concentration of 10 at % was used. It had a thickness of 103 µm and a diameter of 7 mm. The thin disc was pumped with a fiber coupled pump diode which had a maximum output power of 78 W. Contrary to the design in Refs. [9] and [10] the radiation emitted by the thin disc was only parallel to s-polarisation. The thin film polarizer (TFP) restricted the polarisation in the cavity to s. A quarterwave plate (QWP) was placed in front of the electrooptical switch instead of replacing the QWP by a static quarterwave phase shift.

 

Fig. 1 Experimental setup of thin disk based regenerative chirped pulse amplifier.

Download Full Size | PDF

3. Experimental results

The output energy of the regenerative amplifier versus seed energy is shown in Fig. 2 (a) . For this measurement the amplifier was operated at a constant pump power of 70 W and a fixed roundtrip number of 25. By varying the seed energy from 0.1 to 4 nJ a linear increase of the output energy could be observed with a maximum value of 48 µJ, measured behind the GRISM compressor.

 

Fig. 2 (a) Energy of compressed pulses vs. seed pulse energy at 25 roundtrips and 70 W pump power. (b) Energy (left axis) and pulse duration (right axis) of compressed pulses vs. roundtrip number at 75 W pump power.

Download Full Size | PDF

A further scaling of the output energy can be performed by increasing the cavity roundtrip number. As shown in Fig. 2 (b) a variation from 28 to 44 roundtrips resulted in an almost linear increase of the amplification ratio from 1.25 x 10⁴ to 10⁵ corresponding to a pulse energy change from 50 µJ to 400 µJ. At the repetition rate of 20 kHz no bifurcation was observed over the entire roundtrip number range. The output pulse energy was limited by the damage threshold of the thin disk to 571 µJ, measured in front of the compressor. Although the high power output beam was not attenuated, no thermal problems in the GRISM compressor occurred. For this reason the M² value, that was measured to be better than 1.1 at the laser output, remained unchanged at the GRISM output. As the overall efficiency of the GRISM compressor was 70%, the resulting dechirped pulses had an energy of 400 µJ, corresponding to an average power of 8 W.

On the right ordinate of Fig. 2 (b) the dechirped pulse duration is depicted, determined from the measured autocorrelation traces and the corresponding deconvolution factors. These factors were calculated from FWHM values of the Fourier limited autocorrelation traces assuming a flat phase and the FWHM of the Fourier-transformed power spectra. As can be seen the pulse duration remained nearly unchanged in a range between 190 and 214 fs.

The normalized power spectrum of the output pulses at a maximum pulse energy of 400 µJ is shown in Fig. 3 (a) . It has a FWHM of 2.9 nm at a center wavelength of 1022.7 nm.

 

Fig. 3 (a) Measured power spectrum of 400 µJ pulses after compression. (b) Measured intensity autocorrelation of dechirped 400 µJ pulses (solid) and Fourier-limited autocorrelation function assuming a zero phase (dashed).

Download Full Size | PDF

Measured and calculated Fourier-limited autocorrelation (AC) traces at 400 µJ pulse energy are shown in Fig. 3 (b). The measured AC had a FWHM of 426 fs, whereas the Fourier-limited one assuming a zero phase had a width of 334 fs (FWHM), pointing out that the pulse was compressed to 21% over the Fourier-limit. A deconvolution factor of 2.16 was calculated as described above resulting in a pulse duration of 197 fs. Applying a long range autocorrelator covering a time scale of 150 ps we found satellite pulses at delays of ± 16 ps as depicted in Fig. 4 (a) . These satellite pulses had their origin in an etalon occurring between the planar surfaces of the intra cavity used QWP which exactly had a thickness of 1.6 mm. The satellite pulses reached less then 3% of the intensity in relation to the main pulse. Additional artifacts were found at ± 9.6 ps that were caused by a 1 mm thick beam splitter inside the commercially available autocorrelator, as was checked with the manufacturer [11].

 

Fig. 4 (a) Autocorrelation trace of compressed 400 µJ pulses showing satellite pulses at ± 16 ps and ± 9 ps. (b) Pulse duration versus repetition rate at a fixed energy of 100 µJ behind the compressor.

Download Full Size | PDF

The potential of working at high repetition rates and high average output power was investigated by varying the repetition rate between 20 kHz and 125 kHz at a constant pump power of 75 W. To keep the energy of the compressed pulses constant at 100 µJ during this measurement, the roundtrip number was adjusted between 32 and 75. The resulting pulse duration was measured to be in a range from 194 fs to 255 fs, as shown in Fig. 4 (b). The maximum extracted average power was 17.9 W, measured in front of the compressor at 125 kHz and a roundtrip number of 75. As the maximum extractable power was reached [12], a further increase of the number of roundtrips resulted in a decrease of the average power. Over the whole variation range of the repetition rate no bifurcation was observed.

4. Conclusion

An ultrafast high power thin disk regenerative, chirped pulse amplifier based on the infrequently employed laser host Yb:KLuW has been presented. The amplified pulse energy showed a linear relation to the seeded energy in a range from 0.1 nJ to 4 nJ. The dechirped pulse duration remained nearly unchanged over the entire energy scaling range. Amplification at a repetition rate of 20 kHz lead to a maximum output pulse energy of 571 µJ. Dechirping with a transmission GRISM arrangement resulted in a pulse energy of 400 µJ, and a pulse duration of 197 fs. Scaling the repetition rate was demonstrated from 20 kHz to 125 kHz at a constant pulse energy of 100 µJ whereas the pulse duration was in a range between 194 fs and 255 fs.