1. Introduction

Ultra-short laser pulses have opened the way for many interesting applications during the last years. Especially few-cycle laser pulses, with their tight spatial and temporal localization of laser radiation in the focus, enabled studies on temporal and spatial scales not accessible before [1]. Ultrafast pump probe experiments enabled triggering and tracing of chemical reactions and studies of ultrafast relaxation processes as well [1].

The generation of high order harmonics (HHG) in noble gases allows for the generation of coherent laser like radiation with a wavelength of a few nanometers (XUV) [2]. Furthermore, when using carrier-envelop phase (CEP) stabilized few-cycle driving laser pulses, the generation of isolated attosecond pulses becomes feasible [3]. The shortest pulses generated so far have only 80 as [4] duration, and therefore, allow for observation of the electronic motion in atoms, plasmas and solids. A remaining challenge associated with HHG is the low photon flux in the XUV range, caused by the poor conversion efficiency of the HHG process itself. Hence, a high laser output power and repetition rate is desirable for many applications, improving the XUV-flux and the signal to noise ratio at the detector.

Another attractive application for coherent XUV radiation, generated via HHG of a high repetition rate laser source, is seeding of a free electron laser (FEL) [5] to improve the stability and brilliance as well as spectral and temporal properties of the emitted radiation.

Today, optical parametric chirped pulse amplifier technology (OPCPA) is a known concept to amplify ultra-short laser pulses up to highest pulse energies [6]. Due to the enormous gain bandwidth in noncollinear geometry, it supports amplification of few-cycle pulses to peak powers in the TW range [7,8]. Negligible absorption and absence of energy storage in the nonlinear medium lead to strongly reduced thermal load, compared to traditional Ti:Sa amplifiers, which makes the concept suitable for high average power operation.

Diode pumped Yb-doped lasers and amplifiers having a low quantum defect of only a few percent are suitable high power pump lasers. In particular the fiber geometry, due to its huge cooling surface, has proven average power scalability, relaxed thermo-optical management, while the waveguide structure conserves excellent beam quality at high power operation.

State-of-the-art fiber chirped pulse amplifiers (FCPA) deliverer sub-ps pulses with up to 1 mJ pulse energy [9] and up to 830 W of average power [10]. Therefore these powerful amplifiers are well suited as OPCPA pump. Laser average powers as high as 4.0 W [11] as well as sub-30 fs pulse durations [12] have been achieved at degeneracy. Recently, a laser system incorporating a few-cycle Ti:Sa oscillator, a high power FCPA pump laser and parametric amplification in BBO produced pulses as short as 7 fs with up to 35 µJ pulse energy at 60 kHz repetition rate [13]. The corresponding average power was as high as 2.0 W.

In this contribution we report on an update of this OPCPA laser system with significantly improved performance.

2. OPCPA laser system

The experimental setup of the laser system is shown in Fig. 1 . A broadband Ti:Sa oscillator, delivering ~2 nJ pulses with ~300 nm optical bandwidth (−10 dB) at 108 MHz repetition rate, is used to seed the OPCPA as well as the pump laser.

 

Fig. 1 Schematic experimental setup of the OPCPA system including Ti:Sa oscillator, DFG- difference frequency generator, CM- chirped mirror, BS- beam splitter, HNL-PCF- highly nonlinear photonic crystal fiber, IF-interference filter, stretcher fiber, two fiber preamplifiers, high power fiber CPA system, SHG- second harmonic generation, pulse shaper based on SLM- spatial light modulator, prism stretcher, two stage optical parametric amplifier and fused silica bulk compressor. (AOM- acusto-optical modulator, PKC- Pockels cell, OI- optical isolator).

Download Full Size | PDF

Pump laser

Soliton self frequency shift of the oscillator pulses in a 20 cm long highly nonlinear photonic crystal fiber (HNL-PCF, NL-PM 750) generates synchronized seed pulses [14] for the pump laser. A reasonable amount of 5 pJ pulse energy within the Yb gain region is selected by an interference filter (IF, 1030 nm ± 2.5 nm). A 100 m long fiber stretcher prevents the pulses from nonlinear distortions in the preamplifiers. A Pockels-cell (PKC) and an acusto-optical modulator (AOM) are used to reduce the repetition rate to 48 kHz or 96 kHz in the preamplifier section. The high power FCPA system which is applied as power amplifier for the pump pulses, delivering 910 fs pulses with 550 µJ pulse energy, is described in detail elsewhere [9]. Due to the excellent beam quality (M²<1.3) and the good pulse contrast (ASE < 1%, unwanted intermediate pulses < 1%) efficient frequency doubling is possible in a 1 mm long BBO, resulting in 780 fs long 330 µJ pulses for OPA pumping.

Dispersion management

To generate the beat signal for CEP stabilization of the oscillator, a 2 mm long PPLN crystal is inserted for difference frequency generation (DFG) at the oscillator output. Its group velocity dispersion is mainly compensated via 12 bounces on a chirped mirror pair (CM, −50 fs² / bounce). For efficient parametric amplification, the broadband pulses have to be stretched to ~500 fs duration and recompressed afterwards.

For precise control of the spectral phase, which is necessary to achieve few-cycle pulse durations, a prism (SF57) based phase shaper is implemented [15], utilizing a 640 pixel single-mask spatial light modulator (SLM). A fused silica Brewster prism stretcher is applied to compensate for the material dispersion of the SF57 prism and chirp the pulses negatively prior to optical parametric amplification. In consequence, a simple bulk compressor (10 mm long fused silica block) can be used as final pulse compressor with high throughput (measured 95%). Table 1 shows the dispersion at 800 nm central wavelength for every optical element within the setup up to the 6^th order. The last row contains the residual dispersion which is corrected by the pulse shaper.

Table 1. Dispersion of the involved optical components in the OPCPA setup

View Table

Optical parametric amplifier

To achieve high gain and good pump to signal conversion efficiency we apply two amplification stages [16]. Both BBO crystals are operated at ~100 GW/cm² pump intensity, which allows for very short crystal lengths. A first high gain stage, consisting of a 3 mm long BBO, provides a gain factor of ~5·10⁴. A second 2 mm long BBO crystal, driven by the remainder of the pump, is used to achieve high conversion efficiency. Indeed, the pulse energy is increased to 74 µJ, which corresponds to a gain factor of ~3 and a total pump to signal conversion efficiency as high as 22%. To achieve good beam quality careful mode matching of pump and signal in the crystals is necessary. The diameter of the signal beam is chosen to be slightly larger than the pump which partly compensates spatial gain narrowing within the OPA. In order to avoid parasitic SHG of the signal pulses we apply a tangential phase matching scheme at a noncollinear angle of 2.3 °. The amount of parasitic signal SHG is measured to be smaller than 1%. Slight angular detuning from perfect phase matching of the signal central wavelength allows us to amplify the full bandwidth of the seed oscillator, which has been theoretically predicted in [17].

3. Laser system performance

The achieved output power of the two stage OPA (measured after the compressor) is plotted versus the incident pump power in Fig. 2a . The maximum output power is as high as 6.7 W with a corresponding pulse energy of 70 µJ. Figure 2b shows the normalized amplified spectra measured after the first and second OPA. The amplified spectrum covers a bandwidth (−10 dB) as large as 320 nm.

 

Fig. 2 a) Compressed output power of the OPCPA versus incident pump power b) Measured amplified spectra with one (grey) and both (red) OPAs pumped.

Download Full Size | PDF

The Fourier limit of the measured spectrum is 5.6 fs pulse duration, which would result in 10.2 GW peak power. Characterization of the temporal pulse profile is performed with an interferometric autocorrelator supporting 5 fs pulse durations. The measured autocorrelation trace is presented in Fig. 3 and indicates a pulse duration of 8 fs. A larger scan range measurement indicates no significant pulse content outside the plotted ± 75 fs window, which is shown in Fig. 3. Experimentally, pulse compression is performed by compensating the calculated residual dispersion up to the 6^th order (tab. 1.) with the SLM phase shaper. Additional fine-tuning of the second and third order dispersion is applied to achieve the shortest pulse duration. Consequently, for comparison, we stretched the Fourier transform of the measured spectrum with 4^th order dispersion to 8.0 fs pulse duration. This corresponds to 6.5 GW peak power. Please note that within a ± 10 fs wide temporal window, containing only the main pulse, we found > 80% of the pulse energy. In consequence a pulse peak power > 6 GW can be assumed and is validated in the HHG experiment presented later.

 

Fig. 3 Measured interferometric autocorrelation trace indicating 8.0 fs pulse duration.

Download Full Size | PDF

The amount of superfluorescence is measured to be <15 mW (<20 mW) for the first (second) OPA, when the seed is blocked. Significant reduction of the fluorescence level is expected and numerically predicted [18] for the seeded OPA.

4. CEP stability

The full advantage of few-cycle lasers can only be exploited when stabilizing the carrier-envelop phase (CEP). For that purpose a DFG beat signal is generated in a PPLN, located at the oscillator output. The oscillator is stabilized with a Menlo Systems XPS800 stabilization electronics device, driving an AOM to modulate the oscillators pump power. The stabilization electronics are operated at one fifth of the laser repetition rate, meaning every 5^th pulse has the same CEP. In consequence, the pump laser repetition rate is chosen to amplify only pulses with the same CEP in the OPCPA system.

The CEP of the amplified pulses is measured with an f-2f interferometer (Menlo Systems APS800) incorporating white light generation in sapphire and frequency doubling in a thin BBO crystal. The spectral interference pattern is observed with a fast CCD spectrometer and shown in Fig. 4 a . For the CEP measurements presented here, the laser is operated at 48 kHz repetition rate and the spectrometer acquisition time is 21 µs, so each measurement point contains two laser pulses.

 

Fig. 4 a) Three successive f-2f interference patterns observed with the spectrometer (spectrometer acquisition time: 21 µs - containing 2 laser pulses) b) CEP measurement over a short measurement time of 1.0 s.

Download Full Size | PDF

Figure 4b shows the calculated CEP over a measurement time of 1.0 s. Due to the limited spectrometer readout speed and the necessary Fourier transform algorithm, the update rate of this measurement is only 60 Hz. We observe a standard deviation of 0.47 rad, which is satisfying, keeping in mind that the laser system emits about 50’000 laser pulses within this period. It has to be stressed that the laser is operated in a rough environment with plenty of acoustic and vibration noise sources as well as turbulent air flow caused by the air conditioning system. Certainly, CEP stable operation over a longer time scale would require a slow feedback loop, additional CEP servos, such as moveable fused silica wedges, and isolation of the entire laser systems from any mechanical and thermal noise sources.

5. High harmonic generation in Krypton gas jet

The far field beam profile and the intensity distribution of the laser radiation in the focal spot of a f=200 mm curved silver mirror (inset) are shown in Fig. 5a . This focal spot measuring 69 µm · 91 µm in diameter is placed in front of a Kr gas jet within a vacuum chamber in order to generate high order harmonics. The generated XUV radiation is characterized spectrally by a grazing incidence monochromator (McPherson 245/310G) and a channel electron multiplier. A 200 nm thick aluminum filter is used to block the fundamental laser light. The obtained harmonic spectrum is shown in Fig. 5b. The shortest wavelength we observe is 21.8 nm corresponding to the 33^th harmonic of the incident infrared laser light. An estimation of the peak intensity in the focal spot via the cut-of law [19] leads to 2.7·10¹⁴ W/cm², revealing a pulse peak power of 6.4 GW.

 

Fig. 5 a) Far field beam profile and intensity profile at the focal spot (inset) taken with a CCD camera. b) Measured harmonic spectrum and magnified spectrum at cut-off region (inset).

Download Full Size | PDF

Estimation of the generated pulse energy, at the strongest harmonic (~40 nm), leads to a value of some pJ. This value corresponds to typical conversion efficiencies that can be obtained in gas jet harmonic generation [20].

6. Conclusion and outlook

In this paper, we present a high power few-cycle OPCPA system delivering 8.0 fs pulses at 96 kHz repetition rate. The corresponding emitted average power is as high as 6.7 W, which is, to our knowledge, the highest average output power reported for a few-cycle laser so far. The pulse energy is increased to 70 µJ by implementing a carefully optimized two stage OPA scheme. The resulting pulse peak power, also confirmed by a HHG experiment, is more than 6 GW, which is the highest value ever reported for a fiber driven laser system.

In addition, the CEP of the amplified pulses is characterized. When stabilizing only the oscillator, the best CEP standard deviation measured is 0.47 rad over a short period of time. Certainly, CEP stable operation over a longer time scale would require a slow feedback loop and various precautions to isolate the laser system from the environmental noise.

Furthermore, we generate high order harmonics in a Kr gas jet. Wavelengths as short as 21.8 nm are detected. The pulse energy per harmonic is estimated to of the order of some pJ. Quasi-phase matching schemes have the potential to increase the HHG efficiency further, in order to fulfill the FEL seeding requirements (~nJ pulse energy/harmonic).

Additionally, an increased few-cycle pulse energy is desired. For this purpose a combination of multiple fiber amplifiers [21], as well as alternative laser concepts, such as inno-slab [22], have to be considered as pump laser in future.

The average output power of the system is determined by the pump laser. However, an 830 W average power femtosecond fiber laser has been demonstrated recently [10] and milijoule pulse energies at this average power level are in reach. By applying this kind of pump laser a 100 W OPCPA seams feasible in future.