1. Introduction

The optical parametric chirped pulse amplification (OPCPA) concept has become one of the common methods to generate high energetic pulses with durations of few optical cycles [1]. At repetition rates around 1 kHz, pulse energies up to several mJ and pulse durations down to 5.5 fs could be achieved [2]. Nevertheless, countless applications would benefit from higher repetition rates beyond 100 kHz to further increase signal-to-noise ratios or photon fluxes. Regarding the infrared spectral region around 3 μm, pulses with more than 1 μJ of energy and pulse durations of 96 fs at 100 kHz repetition rate have been reported [3, 4]. Around 800 nm, pulses with 37 μJ and 52 fs at 97 kHz were demonstrated using fiber chirped pulse amplification concepts [6]. At even higher repetition rates of 2 MHz, sub-20 fs, 860 nJ pulses have been demonstrated so far [5]. In both systems, the seed for the parametric process was produced beforehand by using narrowband pump sources around 1 µm and generating the seed via spectral broadening mechanisms in sapphire [5] or photonic crystal fibers [6], respectively. In contrast, another approach is to generate the pump pulse for the parametric process from a fraction of the seed spectrum by soliton frequency shifting in photonic crystal fibers to the near infrared [7]. With this, pulse energies up to 500 nJ and durations down to 15.6 fs at 2 MHz have been demonstrated [8].

The drawback of the intrinsically noisy Raman frequency shifting can be avoided by using a single low-noise broadband oscillator simultaneously seeding the parametric amplifier and the pump amplifier. Additionally, those oscillators can easily be carrier-envelope phase (CEP) stabilized with common stabilization techniques due to their broad emission spectrum [9]. With this scheme, pulses with up to 1 μJ were demonstrated using a rod-type fiber amplifier generating the pump radiation. Both, pump amplifier and parametric amplifier were seeded by a broadband Ti:sapphire oscillator. With an amplification bandwidth of 100 nm, the compressed pulse duration was 15 fs [10].

In this paper we present a compact OPCPA system based on a noncollinear phase matching geometry (NOPA) with compressed pulse energies of more than 1 μJ and pulse durations below 9 fs, which are the shortest OPCPA pulse durations ever obtained at repetition rates beyond 100 kHz to our knowledge. The pump of the parametric process is generated by a frequency-doubled thin-disk Yb:YAG regenerative amplifier. The NOPA itself and the regenerative amplifier are both seeded by a broadband Ti:sapphire oscillator. Thin-disk regenerative amplification concepts for OPCPA pumping have been already demonstrated at repetition rates around 5 kHz [7, 11]. These systems suffer from complex stretcher and compressor devices to prevent nonlinear pulse distortion as well as optical damage at pulse energies beyond 1 mJ. At higher repetition rates and lower energies, instead, this disadvantage can be circumvented by reducing the accumulated nonlinear phase simply by enlarging the beam diameters inside the regenerative amplifier cavity. Although this concept is well established [12], we present the first realization of a CPA-free thin-disk based regenerative amplifier pumping an OPCPA system. Regarding the high repetition rate of 143 kHz in combination with the resulting peak power of nearly 130 MW, this system forms a promising laser source for future table top strong-field experiments in atomic and molecular physics.

2. Experimental setup

An overview of the experimental setup is given in Fig. 1 . It can be divided into three different modules; the broadband Ti:sapphire oscillator, the regenerative thin-disk amplifier in combination with a fiber preamplifier and second harmonic generation (SHG) delivering the pump and finally the non-collinear optical parametric amplifier (NOPA) with subsequent pulse compression.

 

Fig. 1 Schematic setup of the non-collinear OPCPA. The abbreviations are explained in the text.

Download Full Size | PDF

The Ti:sapphire oscillator (Venteon|Pulse: one ub) delivers pulses with more than 3 nJ of energy at 70.5 MHz repetition rate covering a spectrum from 620 nm to 1160 nm (see Fig. 2 ). The infrared edge of the oscillator overlaps with the gain band of common Ytterbium doped laser materials at 1030 nm. Hence, this part of the spectrum can be seeded into the amplifier without any further frequency shifting. The separation is realized with an output coupling mirror primarily made for broadband Ti:sapphire oscillators. The left hand side of Fig. 2 shows the full spectrum and the leftover after reflection at the output coupler (dashed). Even though the energy of about 20 pJ at 1030 nm (Δλ ≈6 nm) showed to be enough to seed the regenerative amplifier directly, the pulses are preamplified in a fiber amplifier to ensure better ASE suppression and operation close to saturation.

 

Fig. 2 (a) Optical output spectrum of the Ti:sapphire oscillator in logarithmic scale. The solid line represents the complete spectrum, whereas the dashed line shows the spectrum after extraction of the seed for the pump laser. (b) Output spectrum of the fiber preamplifier.

Download Full Size | PDF

The amplification is realized using 25 cm of ytterbium doped fiber (Liekki Yb 1200-4/125, single mode) pumped with up to 500 mW at 976 nm. With this simple scheme, the pulse energy is enhanced to the nJ-level, and the regenerative amplifier is seeded with about 200 pJ of pulse energy inside the amplification bandwidth of Yb:YAG. Due to the dispersion of the fiber itself, the pulse duration of the amplified pulses increases to approximately 1 ps. The pulse train passes through an optical isolator to prevent the preamplifier from unwanted back reflections. The regenerative amplifier is based on an Yb:YAG thin-disk as active gain medium. It is pumped with up to 120 W at 940 nm. In a home-built pump chamber, the pump radiation passes 24 times through the gain medium (doping: 7%, 215 μm thick, wedge: 0.1°). The pump diameter on the disk is about 2.4 mm, the laser spot slightly smaller. The pulses are trapped inside the resonator using a β-barium-borat (BBO) Pockels cell in combination with a thin film polarizer and a switching arrangement analogue to [13]. By enlarging the beam diameter inside the Pockels cell to approximately 3 mm, the accumulated nonlinear phase is kept small enough to omit some stretcher and compressor arrangement. The repetition rate is set to 143 kHz, but in principle, the switching electronics allows for even higher values up to 1 MHz. In summary, the regenerative amplifier delivers pulses with up to 40 μJ at 143 kHz taking 84 round trips inside the cavity. The output of the regenerative amplifier is stable in time with an rms-noise below 1%. The autocorrelation is measured with a long-range intensity autocorrelator and leads to pulse durations of 1.56 ps assuming Gaussian pulse shapes. Regarding the pulse spectrum (see Fig. 3 ), no emerging self phase modulation was detected up to pulse energies of 40 μJ. At this point, the calculated B-Integral equals approximately π/10. Behind the amplifier, the pulses are frequency doubled inside a 2 mm long lithium borat (LBO) crystal with an efficiency of nearly 50%.

 

Fig. 3 Regenerative amplifier characteristics. (a) power spectrum. (b) measured background free intensity autocorrelation (dots). The Gaussian fit (solid line) reveals a pulse duration of 1.56 ps.

Download Full Size | PDF

To ensure optimum NOPA performance, the broadband seed from the Ti:sapphire oscillator is stretched in time to a significant fraction of the pump pulse. With comparatively short pump pulses just one 6.35 mm thick fused silica (FS) substrate and the ambient air plus two BK7 wedges for fine adjustment are sufficient to stretch the pulse before it enters the 5 mm long BBO crystal, where the parametric amplification takes place. The crystal is cut for type I phase-matching (Θ= 23°, Φ= 0°) with extraordinarily polarized pump and ordinary seed and idler. The seed is focused by a mirror with a radius of curvature of 200 mm down to a beam diameter of approximately 90 µm. The pump is focused slightly weaker (f = 500 mm) resulting in a beam diameter of about 120 µm. To maximize the amplification bandwidth, the internal angle between pump and seed is chosen to be 2.4°. The temporal overlap inside the crystal can by adjusted with a delay stage in the seed arm. Corresponding to the repetition rates of pump (f_regen= 143 kHz) and seed (f_seed= 70.5 MHz) pulse, every 493rd pulse of the Ti:sapphire pulse train is amplified. The pulses are compressed using broadband double chirped mirrors (Nanolayers Naneo|Chord). Each bounce on the mirror pair compensates the dispersion of 2.1 mm fused silica up to fourth order. Pulse characterization is realized with an interferometric autocorrelator specially designed for few-cycle pulses and a SPIDER, respectively.

3. Results and discussion

For the experiments, 14 μJ of energy at 515 nm were used to pump the NOPA resulting in a pump intensity of approximately 70 GW/cm². The average power increases by an amount of 215 mW leading to a pulse energy of 1.5 μJ at a repetition rate of 143 kHz and an energy contrast ratio between amplified pulses and unamplified background pulses (70.5 MHz, 1.5 nJ) of approximately 1000, which is also the gain factor of the system. The power stability is mainly determined by the noise of the used pump and seed source. Both rms-values are less than 1%, and the total rms-stability is approximately 2%.

The amplified spectrum (see Fig. 4 ) reveals a spectral width of more than 200 nm and supports Fourier limited pulse durations down to 6.0 fs. The structure on top of the amplified spectrum reproduces well the seed spectrum and the modulations are due to dispersive ripples inside the Ti:sapphire oscillator [9]. After amplification, the pulses are recompressed by thirteen bounces on double chirped mirror pairs with an efficiency of approximately 86% leading to pulse energies of 1.3 μJ. Taking the dispersion of the ambient air, the BBO-crystal and the fused silica substrate into account, this is in good agreement with the theoretical calculations.

 

Fig. 4 Normalized amplified (solid line) and seed spectrum (dotted line).

Download Full Size | PDF

In a final step, pulse characterization is done using an interferometric autocorrelator and SPIDER, respectively. To ensure not to measure the background pulses at 70.5 MHz, the autocorrelation is performed using lock-in detection with a reference signal from the regenerative amplifier at 143 kHz. In the SPIDER measurements, this is ascertained by vanishing of the SPIDER signal in case of blocking the pump beam.

Both measurements reveal pulse durations of 8.8 fs and a clean pulse shape (see Fig. 5 ). The mismatch between measured and Fourier limited pulse duration can be explained with uncompressed spectral components beyond 940 nm (see pulse phase on the left hand site of Fig. 5). The energy fraction inside the main pulse is higher than 85% leading to a peak power of nearly 130 MW. The results are confirmed by the generation of a white light continuum in a 3 mm thick sapphire plate. Even with weak focusing (f = 100 mm) only 200 nJ of pulse energy was necessary to generate a stable white light continuum covering the hole visible spectrum.

 

Fig. 5 (a) Amplified spectrum and retrieved SPIDER phase. (b) Reconstructed pulse shape revealing clean 8.8 fs pulses (FWHM).

Download Full Size | PDF

4. Conclusion and outlook

In conclusion, we reported on a high repetition rate OPCPA system delivering pulse durations below 9 fs, which have not been shown before for repetition rates beyond 100 kHz. The compressed pulse energy is 1.3 μJ resulting in a peak power of nearly 130 MW. The system consists of a broadband Ti:sapphire oscillator simultaneously seeding the parametric amplifier and a home-built thin-disk regenerative amplifier generating the pump. Using a regenerative amplifier to generate the pump of the OPCPA makes the repetition rate of our system in principle scalable.

By implementing a second NOPA stage, the energy is anticipated to be further increased for future applications. Apart from energy and repetition rate scaling, CEP-stabilization of the Ti:sapphire oscillator is in progress. With this, the high peak intensity in combination with the high repetition rate makes this system a promising candidate for the next generation of compact high repetition rate laser sources for applications in high field physics such as harmonic generation.