1. Introduction

In the past decade, optical parametric chirped pulse amplification (OPCPA) systems have gained an important role in terms of generating highly intense, ultrashort laser pulses [1]. Currently, those laser systems deliver pulse energies reaching the mJ-level with pulse durations of only few optical cycles at repetition rates up to 100 kHz [2, 3]. They are auspicious optical sources regarding state of the art light-matter interaction experiments such as high-harmonic generation [3, 4]. Nevertheless, there is still a lack of high repetition rate μJ-OPCPA systems with more than 100 kHz repetition rate. Here, the relatively high repetition rates with their high photon fluxes can further reduce some processing time and improve the statistics, while the few-cycle μJ-pulses would still be sufficient for high-field experiments. Pushing the pulse durations down to the single cycle limit, the influence of the carrier envelope phase (CEP) of the pulses becomes apparent [5, 6]. Stabilizing the CEP of Ti:sapphire oscillators is meanwhile routinely realized with self referencing techniques and feedback loops to the lasers’ pump power [7]. While oscillators CEP is usually stabilized to a fraction of the oscillator repetition rate, stabilizing the CEP to zero can easily be achieved in amplifier systems by choosing the right picking ratio [8].

In this paper we present a compact and table top OPCPA system generating CEP stabilized, multi-μJ pulses in the two-cycle regime at high repetition rates between 100 and 500 kHz. Using an octave spanning Ti:sapphire oscillator simultaneously generating pump and seed radiation, neither frequency shifting nor preamplification is necessary to seed the pump amplifier. Also CEP stabilization of the Ti:sapphire oscillator is directly realized without any external spectral broadening. Compared to previous results [9], the energy is boosted from 1.3 μJ in the single stage setup to more than 3 μJ in a double stage configuration. The pulse duration is further reduced from 8.8 fs down to the two-cycle regime (5.7 fs) with pulse peak powers of more than 250 MW. Additionally, the amplified output pulses are stabilized with respect to their carrier envelope phase with a standard deviation of 210 mrad rms measured over more than 10⁶ consecutive pulses. The repetition rate of the amplified output pulses was varied between 100 and 500 kHz showing the versatility of the system especially for possible applications.

2. Experimental setup

The experimental setup of our OPCPA system consists of a broadband Ti:sapphire oscillator (VENTEON - Pulse one UB), a homebuilt CPA-free, Yb:YAG thin disk regenerative amplifier with subsequent second harmonic generation and the double stage parametric amplifier followed by the pulse compression (Fig. 1). The Ti:sapphire oscillator delivers an octave spanning spectrum from 580 to nearly 1200 nm with nJ pulse energies at 70.5 MHz repetition rate. The spectral wings around 600 and 1200 nm were separated and directly sent to an f-2f interferometer, measuring a beat note without any additional spectral shifting. With a common self referencing scheme and a feed back loop to the oscillator pump power via an AOM, every fourth output pulse coming from the oscillator is stabilized with respect to its carrier envelope phase. While the main spectral part from 600 to 1000 nm is used as signal beam for the parametric amplification, the infrared edge is amplified, afterwards frequency doubled and used as pump beam at 515 nm. The Yb:YAG regenerative amplifier is directly seeded without any preamplification stage, which is another benefit in terms of the compactness of the system. The energy amount of approximately 70 pJ at 1030 nm is well above the predicted minimum seed energy of 2 pJ ( [10], for Nd:YLF regenerative amplifier) and therefore sufficient for complete ASE suppression inside the amplifier. This is confirmed by SHG efficiencies of the output pulses of around 50 %. The measured pulse duration of the 1030 nm pulses is 1.6 ps and the pulse duration of the 515 nm pulses is therefore estimated to be slightly shorter.

 

Fig. 1 Schematic setup of the noncollinear OPCPA. The abbreviations are explained in the text.

Download Full Size | PDF

Subsequently, the generated green light is used to pump the two NOPA stages. The signal pulses coming from the oscillator are stretched in time using two BK7 wedges and one 6.35 mm long fused silica substrate. This in combination with the ambient air and the BBO crystal itself is sufficient for stretching the seed pulses to approximately 600 fs and ensure a good temporal overlap between the pump and the seed pulses. In the first stage, the parametric amplification is realized with a 3 mm long BBO crystal. The signal beam is focused with a mirror (radius of curvature: 250 mm) whereas the pump beam is focused using a lens with a focal length of 500 mm leading to a beam diameter of approximately 240 μm for the pump and a slightly smaller beam diameter for the seed beam. After the amplification, the remaining pump light is recollimated with a lens (focal length: 300 mm) and afterwards focused with a 200 mm lens in a 5 mm long BBO crystal which is used in the second amplification stage. The signal beam is recollimated after the first stage and focused down into the second BBO crystal with a radius of curvature of 200 mm. Inside the BBO both beam diameters equal approximately 200 μm. In both amplification stages the internal noncollinear angle between pump and signal is set to approximately 2.4° to enable broadband amplification. The two crystals are cut for Type I configuration, which means the polarization of the pump beam is extraordinary, whereas the signal and idler polarizations are ordinarily oriented with respect to the crystal optical axes. Regarding the horizontal plane confined by pump and seed beam, the polarization of the seed is perpendicular and that of the pump is parallel. The parametric amplification stages are set up in the tangential phase matching geometry to avoid unwanted frequency doubling of the amplified signal and therefore reduction of the total gain [11].

The compression of the amplified pulses is realized with double chirped mirrors (DCMs). In contrast to previous works [9], we are now using DCMs designed for a better suited spectral bandwidth from 600 to 1200 nm (previously 500 to 1000 nm). Fine adjustment of the dispersion can be managed with the BK7 wedges in front of the first amplification stage.

3. Results and discussion

3.1. Energy scaling and pulse characterization

In the double stage configuration, up to 17.5 μJ of energy at 515 nm with a 143 kHz repetition rate were used to pump the first NOPA stage. At this point the pump intensity equals approximately 20 GW/cm². The signal beam with pulse energies of about 0.5 nJ is amplified to 1.07 μJ, i.e. the average power increases from 34 mW to 187 mW. Since no picker is employed, the average power of the unamplified 70.5 MHz background pulses (34 mW) is subtracted, resulting in the above mentioned pulse energy. Subsequently, the remaining pump energy (75 % of total pump energy) is recollimated and used to pump the second amplification stage. Here, the pump intensity equals approximately 25 GW/cm² leading to amplified pulse energies of 4.1 μJ and a total optical-to-optical efficiency of 23 %. The total gain of the first stage is more than 1000 whereas the second amplification stage reveals a gain of 4.

Pumping only the first amplification stage and blocking the signal beam in front of the first amplifier leads to a negligible superfluorescence background measured after the second stage. The same behavior was observed in case of blocking the seed after the first stage and pumping the second one. Anyway, there is a superfluorescence background of approximately 15 % of the total power in case of blocking the seed in front of the first stage and pumping both stages. We interpret this to a weak superfluorescence background generated in the first stage which is further amplified in the second one. Nevertheless, as described in [12], the amount of superfluorescence in the unseeded case is predicted to be much higher compared to the seeded operation.

Following the amplification, the pulses are recompressed by means of a DCM pair. Especially due to the high dispersion amount of the two BBO crystals (total length: 8 mm), 32 bounces on the mirrors were necessary to compress the pulses down to their minimum duration. The compression efficiency was 80% resulting in compressed pulse energies of 3.3 μJ. For spectral and temporal characterization, we use SPIDER and interferometric autocorrelation apparatus (IAC), respectively. The spectrum reveals a width of nearly 400 nm (at −10 dB) and supports Fourier-limited pulse durations down to 5.2 fs. The measured pulse duration equals 5.7 fs, confirmed by SPIDER and IAC measurements. This is a reduction of nearly 35 % compared to previous results enabled by the use of optimized compression mirrors. Nevertheless, as can be clearly seen in Fig. 2(a), there is a notable remaining structure inside the measured spectral phase resulting in temporal satellites reducing the peak power of the amplified pulses. This remaining phase structure arises by reason of the total amount of 32 bounces on the double chirped mirrors and a residual aberration between the ideal mirror design and the measured GDD curves. Considering the actual temporal pulse shape with approximately 65 % of energy inside the main pulse, the peak power of the amplified pulses equals 0.25 GW. The pulse to pulse error of the amplified pulses was measured to be less than 2% and the far-field beam profile is nearly Gaussian like (Fig. 2(a)).

 

Fig. 2 Spectral and temporal characterization of the amplified pulses; (a) output power spectrum (black) and SPIDER pulse phase (red); the beam profile of the amplified pulses at 2 μJ is shown in the inset (b) reconstructed temporal pulse shape.

Download Full Size | PDF

3.2. CEP stabilization

Regarding possible applications such as high harmonic generation, the stabilization of the CEP is one major request for the used laser source. Since all phase fluctuations of the pump source leave the CEP of the signal unchanged, this can easily be realized by stabilizing the carrier envelope phase of the Ti:sapphire oscillator. Using standard CEP locking techniques, every fourth pulse within the oscillator pulse train has an identical electric field. During the parametric amplification process every 492^nd pulse is amplified. So in fact, every amplified pulse should reveal exactly the same electric field trace under its envelope. To verify this, an additional f-2f interferometer was set up. In a first step, the amplified and compressed pulses are focused into a 3 mm thin sapphire substrate to generate a white light continuum spanning from 500 to more than 1100 nm. In a second step, the infrared part of the spectrum is frequency doubled using a BBO crystal. With stabilized CEP, a clear spectral interference pattern between the fundamental and the frequency doubled spectral components around 510 nm was observed, immediately vanishing in case of switching off the phase lock. For detection, a spectrometer with an integration time of 4 ms (i.e. each interferogram contains more than 570 amplified pulses at 143 kHz repetition rate) was used. To study the CEP time stability, the spectrometer was read out every 280 ms. For a quantitative information about the phase stability, the rms phase error was calculated. The CEP remains stable over 10 s (i.e. more than 10⁶ pulses) with a phase error of 0.21 rad rms (Fig. 3(a)). In even longer timescales of 100 s, phase stability errors below 0.6 rad rms have been obtained (Fig. 3(b)), which are good values taking into account, that there is no slow feedback loop from the OPCPA output to the phase locking electronic. Since only 200 nJ of input energy were necessary to generate the white light continuum, the verification of the CEP-stability and a possible slow feed back loop could be realized only with a fraction of the total energy, whereas the main part can be used for experiments.

 

Fig. 3 Temporal evolution of the observed interference pattern and resulting phase error for a time interval of 10 s (a) and 100 s (b), respectively.

Download Full Size | PDF

3.3. Repetition rate scaling

So far, all experiments have been carried out at 143 kHz repetition rate. Using a regenerative amplifier generating the pump for the parametric amplification process, a change of the repetition rate can easily be realized by changing the repetition rate of the Pockels-cell gating windows. In the experiment, the repetition rate was scaled from 100 kHz up to 500 kHz in steps of 100 kHz. The repeating time can be varied without any further optical realignment of the amplifier; it is limited at 500 kHz by the driving electronics. Only the delay between pump and seed pulses has to be realigned within a range of less than 1 mm. With this, uncompressed two-cycle pulse energies up to 4.5 μJ at 100 kHz and maximum output powers up to 950 mW at 500 kHz have been achieved (Fig. 4). In all configurations, the optical-to-optical efficiency was between 20 and 25 %. The amplification spectrum remains unchanged, so the compressed pulse duration is in the sub-6 fs regime analogue to the measured pulse duration at 143 kHz repetition rate. For lower repetition rates, i.e. higher pulse energies, the accumulated B-Integral inside the regenerative amplifier becomes apparent because of the stretcher-free amplification scheme.

 

Fig. 4 Repetition rate scaling; the diagram shows the uncompressed output energy plotted against the pump energy for various repetition rates.

Download Full Size | PDF

4. Conclusion and outlook

In conclusion, we demonstrated a compact OPCPA system delivering compressed pulse energies of more than 3 μJ in combination with two-cycle pulse durations down to 5.7 fs and high repetition rates between 100 and 500 kHz. Additionally, the CEP of the Ti:sapphire oscillator was stabilized and also the phase of the amplified output pulse remains stable for more than 10 s with a calculated phase error of about 210 mrad rms. The setup is still very compact, consisting of the seed oscillator simultaneously seeding the parametric amplifier and the Yb:YAG regenerative amplifier generating the pump radiation at 515 nm, the two NOPA stages and finally the pulse compression. The pump pulses are amplified without the typical stretcher and compressor arrangement, which is not only a benefit for the compactness of the system, but also allows for comparative short seed pulses. Hence, pulse recompression can simply be done with several bounces on DCMs and avoid complex grating or prism compressors.

In future, further energy scaling could be realized by implementing a stretcher/compressor arrangement to keep the accumulated nonlinear phase inside the regenerative amplifier as small as possible. For example, using chirped volume Bragg gratings with their compact behavior in combination with the relatively easy alignment would be a proper solution for this. With this simple approach, pump energies up to 80 μJ are realistic for repetition rates higher than 100 kHz. With respect to the obtained optical efficiencies for the parametric amplification of more than 20 %, few-cycle pulses with up to 20 μJ of energy should be possible by marginally changing the existing OPCPA system. Regarding the CEP-stabilization, the performance could be further optimized by implementing a slow feed back loop from the amplifier output back to the phase stabilization unit ending up in phase errors around 0.2 rad over timescales of several minutes.

Considering the relatively easy and compact setup, the repetition rate scaling between several tens and hundreds of kHz in combination with the CEP-stabilized μJ-two-cycle-pulses, this compact table-top laser source is ideally suited for phase-sensitive spectroscopic experiments.