1. Introduction

Optical parametric chirped-pulse amplification (OPCPA) is known as an attractive technique to amplify ultra short laser pulses with durations of a few femtoseconds and an ultra high intensity [1–5]. Up to now OPCPAs were operated with repetition rates of up to 1 kHz together with pump pulse energies in the order of a few mJ [5]. The intensity of the amplified and recompressed pulses are well suited for high-harmonic or X-ray generation for numerous applications such as photoelectron spectroscopy [6] or combustion diagnostics [7]. However these applications strongly benefit from higher repetition rates of more than 10 kHz, which requires a suitable pump source.

For the pump laser in an OPCPA system usually Q-switched Nd:YAG laser with ns pulse duration have been utilized, which often cause a limited efficiency for pump to signal conversion [8, 9] and also limit the repetition rate. To achieve multi-kHz repetition rates regenerative amplifiers (RA) can be used [10–14]. RA are common to amplify ultra short pulses from laser oscillators by many orders of magnitude, up to tens of mJ and more than 1 kHz. Recently Metzger et al. [12] published the so far most powerful multi kHz RA based on a Yb:YAG thin-disk amplifier design with 3 kHz repetition rate and 75 W output power. RA based on bulk laser crystals as Nd:YVO₄ [13] or Nd:GdVO₄ [14] have lower pulse energy as well as average power but less complex cavities and pumping schemes. Other RA with some hundred ps pulse duration, some mJ pulse energy and sub 10 Hz repetition rate have been reported, seeded with shaped Q-switched pulses [15] and with stacked pulses from a titan sapphire laser oscillator [16]. The pulse duration as well as the pulse energy would be suitable for our desired application, but the repetition rate is more than thousand times to low.

We present a RA as a pump source for an OPCPA with 20 kHz repetition-rate based on a recently developed oscillator with adjustable pulse duration between 34 ps and 1 ns [17], a 888 nm cw diode-pumped high power Nd:YVO₄ crystal [18, 19] and a second harmonic generation (SHG) stage. The system allows for the generation of pulses at 532 nm with a repetition rate of 20 kHz, a high pulse energy above 1 mJ and a freely adjustable pulse-duration of some hundred picoseconds. The value of 20 kHz is a compromise between a pump impulse energy necessary for the desired OPCPA application on the one hand and the increase of repetition rate as well as optical efficiency of the pump laser on the other hand. Furthermore the adjustable pulse-durations permits for a good temporal overlap between pump and stretched broadband signal pulses in an OPCPA. Finally these properties in combination with the excellent beam quality allow for high efficient OPCPAs.

2. Experimental setup

The experimental setup of the RA and the external SHG generation is shown in Fig. 1. The seed pulses are provided by a recently developed active mode locked diode pumped Nd:YVO₄ master oscillator which provides pulses with an adjustable pulse-duration between 34 ps up to 1 ns and a repetition rate of 108 MHz [17]. In order to reduce the repetition rate to 20 kHz an electro optical pulse picker is used.

To separate the seed pulses from the amplified pulses a thin-film polarizer (TFP), a Faraday rotator and a half-wave plate were used. The following two mirrors and the lens are required to match the mode diameter of the seed and the amplified radiation. The TFP (M₂) of the RA, the quarter-wave plate and the β-barium borate Pockels cell (BBO-PC, aperture Ø 4 mm) are used to switch the pulses. The Nd:YVO₄ RA crystal has a dimension of 4×4×30 mm³ and is longitudinal pumped with 100 W from a fiber coupled diode laser at 888 nm. Due to the low repetition rate of 20 kHz the thermal lens is significantly stronger compared to cw operation. This is due to of the higher inversion and hence heating from energy-transfer upconversion [20, 21]. Furthermore the output power is reduced by spontaneous emission, due to the short fluorescence lifetime of 100 µs [22]. This is in the order of the separation of the amplified pulses. In order to compensate for the thermal lens the curvature of the spherical mirrors M₃ and M₄ is adapted to r=1000 mm. The mirrors M₅ and M₆ (r=-750 mm) are required in order to stretch the cavity of the RA to 2.3 m to ensure that the BBO-PC is completely switched between two passes of a pulse and also limits the maximal theoretically amplificable pulse duration in the RA to some ns.

 

Fig. 1. Schematic of the RA with external SHG generation.

Download Full Size | PDF

 

Fig. 2. (a) Output power (square) and beam quality factor M² (triangle) for 2 to 7 cycles in the RA with a repetition rate of 20 kHz and a seed pulse duration of 300 ps. (b) Contrast of the output pulse for 5 cycles in the RA. The insert shows the intracavity signal of the RA.

Download Full Size | PDF

Finally the amplified pulse is focused by a single lens into a critical phase matched LBO with the dimension of 3×3×15 mm³. The frequency doubled pulses are separated from the pump pulses by two dichroic mirrors.

3. Results

Figure 2(a) shows the output power and the average of the M²-value for x- and y-direction for different numbers of cycles of a pulse in the RA. All M²-values in this paper were measured with the second moment method. The difference of the M²-value for x- and y-direction for all these measurements was less than or equal to 0.02. It is seen that the obtained output power strongly increases with the number of cycles. This results from the high overall gain in the system. Hence only 5 cycles in the RA are required to amplify the 50 nJ of the seed pulses to the maximum pulse energy of well above 1.6 mJ. Between 2 to 6 cycles the M²-value increases only slightly from 1.2 to 1.3. With the 7th cycle the beam quality becomes noticeably inferior and the output power decreases by 37 % compared to the 5th cycle in the RA. The reasons are a reduced gain guiding in the RA crystal and nonlinear effects resulting from the high peak intensity circulating pulses. So it turns out that in this case 5 cycles in the RA are the best choice with respect to output power and beam quality. Therefore all further measurements were done with 5 cycles in the RA.

In general the high gain would allow for a similar amplification in a multipass amplifier setup. However as seen in Fig. 2(a) such an arrangement would require 8-10 passes (corresponding to 4-5 cycles), which is difficult to achieve with the long and small aperture Nd:YVO₄ crystal used. A lower number of passes would be made possible by a setup consisting of a preamplifier and a main amplifier. So such an arrangement has no benefits with respect to the complexity. Furthermore the modulator in the RA improves the contrast by defining a short window of amplification, which reduces the amplified spontaneous emission, which is a special problem in amplifiers with very high gain [23].

In Fig. 2(b) the output pulse of the RA is shown. No pre- or post-pulses can be seen and a contrast exceeding 1:750 for the amplified pulse compared to the background has been achieved. The insert in Fig. 2(b) shows the intracavity signal of the RA. The separation of 15.4 ns between two consecutive pulses is related to the round trip time. Even for the 5th and last cycle the large amplification from cycle to cycle is obvious. Finally, the complete extraction of the pulse after the 5th cycle is shown.

 

Fig. 3. (a) Output power of RA (triangle, up), SHG output power (triangle, down) and SHG conversion efficiency (cycle) for different seed pulse durations. (b) Pulse duration after RA (triangle, up) and SHG generation (triangle, down) for different seed pulse durations.

Download Full Size | PDF

Due to the very high gain in the RA and hence the low number of cycles even components with moderate losses are acceptable, the sensitivity in terms of misalignment is reduced and finally the long-term stability is improved.

In Fig. 3(a) the output power of the RA and of the SHG for different seed pulse durations is shown. To avoid a destruction of RA components no seed pulses with pulse-durations shorter than 200 ps were used. While the output power of the RA varies only slightly between 32.1 W and 33.7 Wfor an increasing seed pulse duration, the output power of the SHG stage decreases from 25.5 W to 19.0 W. The reduction of the SHG output power results from the reduced peak intensity for the longer seed pulse duration, since a constant beam diameter of 770 µm has been used in all experiments. This beam diameter was chosen as a compromise between a high conversion efficiency and a spatially undisturbed SHG beam. However, efficiencies similar to that obtained for the 205 ps seed pulse (79 %) should also be possible for longer pulses if a smaller beam diameter is used. For each pulse duration each average M²-value of x- and y-direction obtained in the SHG experiments was below 1.1, while each average M²-value for the amplified pulses was below 1.3. While for each pulse duration the M²-value for the RA in both directions was nearly identical, the M²-value after SHG was slightly worse in x-direction as in y-direction. We measured an average of 1.10 for the x-direction and 1.03 for the y-direction. Probably this is a result of a slight walk off of the pump- and signal-beam in the LBO crystal.

Figure 3(b) shows the pulse duration after the RA and SHG for different seed pulse durations. Data points above the dashed line indicate an elongation and below the dashed line a shortening of the pulses compared to the pulses directly obtained from the seed oscillator. Due to gain narrowing the amplified pulses are slightly longer compared to the seed pulses. In contrast the pulses behind the SHG stage become shorter. This is a result of the suppression of the leading and trailing edge of the pulse during the second-harmonic generation due to the reduced conversion efficiency in the pulse wings. The difference between the pulse duration of the seed pulses at 1064 nm and the frequency doubled pulses at 532 nm constantly increases. With the maximum pulse duration of 1020 ps from the master oscillator we obtained pulses with 822 ps pulse duration from the SHG.

To determine the energy stability of the system the photo-diode signal from the pulses after the RA and SHG was measured with an oscilloscope (bandwidth 250 MHz, sampling rate 5 GS/s). For the RA we measured for all pulse durations a standard deviation of the peak signal below 0.9 % with typical values of 0.6 % and for the SHG a standard deviation below 1.5 % with typical values of 1.0 %. A housing of the system should further improve the stability.

 

Fig. 4. (a) Normalized, superposed temporal profile of seed (solid), RA (dashed) and SHG (dotted) pulses. (b) Normalized, superposed spectral profile of seed (solid), RA (dashed) and SHG (dotted) pulses.

Download Full Size | PDF

Figure 4 shows the temporal and spectral properties of the pulses behind the master oscillator (solid line), the RA (dashed line) as well as the second harmonic stage (dotted line), respectively. The temporal intensity distributions were recorded by a fast photo-diode and a sampling oscilloscope. The measured response time of this setup was 19.3 ps. The spectra were measured with a scanning Fabry-Perot interferometer. The measured pulse durations at the output of the master oscillator, the RA and the SHG shown in Fig. 4(a) are 205 ps, 217 ps and 185 ps, respectively. The temporal intensity distribution has nearly a Gaussian shape for all pulses. Figure 4(b) shows the corresponding experimentally obtained spectrum. All spectra show a symmetrical shape with a bandwidth of 2.45 GHz, 2.42 GHz and 2.90 GHz for the seed pulses, for the amplified pulses and for the frequency doubled pulses, respectively. This results in a time bandwidth product of 0.50 for the seed laser, 0.53 for the RA and 0.54 for the SHG pulses, which is 18 % above the Fourier limit for Gaussian pulses. In the measured spectrum of the seed laser a modulation with a frequency of 108 MHz is observed, which is related to the repetition rate of 108 MHz. As expected the modulation vanishes in the measured spectrum for RA and SHG because of the much lower repetition rate of 20 kHz. Each reported spectral bandwidth was reduced according to Marzenell et al. [24] assuming Gaussian pulses. All corrections change the spectral bandwidth less than 4 %.

4. Conclusion

In conclusion we have demonstrated a 888 nm diode-pumped Nd:YVO₄ RA with 33.7 W output power at 1064 nm with a repetition-rate of 20 kHz and adjustable pulse durations between 200 ps and 1020 ps. To our knowledge this is the highest average power and pulse energy reported so far for a Nd:YVO₄ regenerative amplifier. Finally a high conversion efficiency of up to 79 % for second harmonic has been achieved, which is due to the excellent temporal and spatial profiles of the amplified pulses. These properties make this system an attractive pump source for an OPCPA.