1. Introduction

The development of high-power, few-cycle light sources is of great interest for a number of applications such as the generation of reproducible mono-energetic electron beams or generation of intense single attosecond pulses from solid density plasmas. Electron acceleration in the bubble regime has already been demonstrated [1, 2, 3] using pulses of 30–50 fs in duration but the reproducibility of the electron bunch parameters remains to be improved. In the few-cycle regime a much better reproducibility is expected because the parametric instabilities in the plasma will be strongly reduced [4]. Recent theoretical investigations of Tsakiris et.al [5] predict the emergence of an intense single attosecond pulse from solid density plasma surface exposed to a multi-terawatt few-cycle driving laser pulse with cosine-shaped electrical field.

Since its invention, the noncollinear optical parametric chirped pulse amplification (OPCPA) technique [6, 7] has been identified as a promising alternative to conventional Ti:sapphire-based amplifiers (The comprehensive review of the OPCPA technique progress is given in a Dubeitis et.all review paper[8]. In the spectral range of the Ti:sapphire amplifiers, peak powers up to 200 TW pulses have been demonstrated so far [9]. This technique allowed the generation of the first terawatt-scale few-cycle pulses [10, 11].

In this paper we present a performance of OPCPA system that amplifies 10 fs pulses till 90 mJ energy. The amplifier design is described in Section 2 and pulse compression, typical output characteristics and performance of the amplifier are discussed in the Section 3.

2. The amplifier

In this experiments we used the same OPCPA setup as presented in Ref. [12]. Basically the amplifier consists of three amplification stages which are designed to support the amplification of pulses comprising three optical cycles.

A key component of our system is the broadband Ti:Sapphire oscillator. The central wavelength λ_c of the oscillator is ~850 nm with the mode-locked spectrum coupling a range of~450 nm (1/e² bandwidth) and an output energy of 4 nJ. The typical oscillator spectrum is shown in Fig. 1(a) as shaded contour, implying a Fourier-limited pulse duration of 4.5 fs. Experimentally we succeeded to compress the oscillator output till ~5.5 fs.

We tested the possibility to stabilize the carrier-envelope phase offset of this oscillator using the monolithic scheme proposed by T. Fuji et al. [13]. We tightly focus few-cycle pulses from a Ti:sapphire oscillator into a highly nonlinear magnesium-oxide-doped periodically poled lithium niobate (PP-MgO:LN) crystal to induce self-phase modulation and difference frequency generation. Due to the enhanced nonlinear interaction and an improved spatial overlap between the two waves (due to the absence of walk-off effects), the interferometric beat signal emerging at f_CEO in the region of spectral overlap is strong enough (65 dB) for reliable CE-phase locking over several hours.

 

Fig. 1. Oscillator spectra obtained with the standard output coupler (shaded contour) and with a more broadband output coupler (dotted line)(a); Soliton-shift in a photonic crystal fiber for the pump amplifier seeding (b).

Download Full Size | PDF

To simplify the overall OPCPA scheme we investigated two possibilities of optical synchronization of the pump and seed pulses: direct seeding from the femtosecond oscillator [14] and soliton-based synchronization [15]. The Ti:sapphire oscillator currently in use (with standard 10% output coupler [16](Femtolasers)) has an energy output of 4 nJ (Fig. 1(a) shaded contour), and 20 fJ of seeding energy in the spectral acceptance region of the Nd:YAG amplifier. We also used a specially designed, more broadband output coupler which has the same transmission. With this output coupler oscillator output energy was only 2.8 nJ (Fig. 1(a) dotted line) and at 1064 nm it delivers 145 fJ of energy. In both cases the seed energy turned out to be unsufficient for stable operation of the Nd:YAG regenerative amplifier free from excessive amplified spontaneous emission. Therefore we resorted to the soliton-based synchronization. To this end we focused part (1.6 nJ) of the recompressed oscillator output into a photonic crystal fiber (PCF). Due to efficient soliton-based frequency shifting several pJ of seed energy is obtained for Nd:YAG amplifier seeding (see Fig. 1(b)). This energy is high enough for reliable operation of the regenerative amplifier.

We inserted inside the regenerative amplifier cavity two 2.1 mm and 4.1 mm thick glass etalons to reduce the seed bandwidth and thus obtain a pump pulse duration sufficiently long for avoiding excessive self phase modulation in the amplifiers chain. The measured pump pulse duration set by this etalon pair is 100 ps at 1064 nm. It is important to mention here, that pump pulse duration could be easily changed, when different thickness etalons is used. The pump pulses from the regenerative amplifier are then amplified up to ~1.4 J in three Nd:YAG amplifiers rods of increasing diameter (8 mm, 12 mm, 18 mm) at 10 Hz repetition rate. After frequency doubling in a Type-II DKDP crystal, the pump laser delivers about 750 mJ pulses at 532 nm with an 8^th order supergaussian spatial profile and a nearly Gaussian temporal profile. The ns-duration background is some 5 orders of magnitude less intense than the ps pump pulse (@ 532 nm), the energy of which is stable to within 1.5% rms.

The remaining 60% of the oscillator output (2.4 nJ) is first temporally stretched to 50 ps in a negative dispersion stretcher [10] and then used to seed the noncollinear OPCPA. In the four-pass, grating-based stretcher the seed beam is dispersed by a 900 groves/mm transmission grating (Wasatch Photonics) and collimated by a parabolic lens (f=800 mm). At Fourier plane, in front of the returning mirror the microfabricated fused silica plate is placed. This plate is used for the higher-order spectral phase correction (mainly 3rd-order dispersion caused by the bulk compressor). Due to the limited acceptance bandwidth (675–1050 nm) and optical losses of the stretcher, the stretched pulse carries an energy of 0.65 nJ. Fine tuning of the spectral phase of the amplified pulses was performed using an acousto-optical dispersive filter (Dazzler, Fastlite). The material used in the Dazzler is TeO₂ which compresses the negatively chirped pulses to ~40 ps. The diffraction efficiency of the Dazzler with phase correction loaded is ~5%, resulting in output pulses of an energy of 10–20 pJ from the Dazzler depending on the amount of dispersion to be compensated. The output of the Dazzler is fed into the OPCPA. The seed pulse duration of ~40 ps is chosen such that the pump intensity variation is negligible across the entire chirped seed pulse what facilitates amplification of the whole seed bandwidth. A larger ratio between durations of the pump and seed pulses can be used to facilitate the synchronization. Consequently this will decrease signal amplification efficiency and will strongly enhance amplification of parametric superfluorescence, particularly if at least one OPCPA stage is operated close to the amplification saturation.

The amplification takes place in three 5-mm-thick antireflection-coated BBO-crystals (θ=24°, ϕ=90°, type I phase matching) with a noncollinear angle of 2.3° in single-pass geometry. Typically the signal pulses are amplified to an energy level of ~1µJ in the first stage, to ~2mJ in the second stage and to ~110 mJ in the third stage. After the amplification, the beam is expanded to a FWHM diameter of ~100 mm and is sent through a compressor consisting of 160 mm of SF57 glass (Schott) and 100 mm of fused silica (Heraeus). After the bulk compressor the beam was downcollimated to ~50 mm diameter. Pulse compression is finished using a set of four positive-dispersion chirped mirrors (~100 fs²/bounce). Beam expansion and step-wise compression is used to reduce pulse self action inside the bulk material, and for the fully compressed 110 mJ pulse the B-integral value is estimated to be below 0.7. The total throughput of the pulse compressor is close to 95%. The residual spectral phase of the compressed pulses is measured directly with a Spider interferometer and used as a feedback signal for the Dazzler for further phase correction.

3. Amplifier performance

 

Fig. 2. Measurement of the amplified signal pulse duration. Retrieved group delay (solid line), maximal spectral bandwidth supported by amplifier (dotted line) and compressed pulse spectrum (shaded contour)(a); Measured pulse duration of 10 fs FWHM (b).

Download Full Size | PDF

The main goal of the present investigation is to improve compression of the parametrically amplified pulse as compared to the previously obtained 14 fs [12]. For that purpose we carefully investigated possible origins of the poor pulse compressability. According to our SPIDER measurement results the spectral phase of infrared part of amplified spectrum is always strongly modulated. In our opinion, this modulation can be attributed to the several phenomena. The spectral divergence of the infrared part of seed spectrum somewhat bigger than around 800 nm due to the frequency dependent mode size of the ultra-broadband Ti:Sapphire oscillator [17]. This will cause that each spectral component after the parabolic lens inside the stretcher will be divergent instead of being parallel. In this situation the calculated profile of the microfabricated correction plate will be wrong for the exact compensation of the higher order dispersion. Another possible explanation is the spectral interference induced by multiple reflections between protective glass covers of the transmission grating. To keep maximally possible amount of the amplified energy inside the compressible part of the seed spectrum we slightly detuned the third stage crystal angle from optimal orientation (Fig.2(a), dotted line, transform-limit-7.6 fs at FWHM) to enhance the amplification of the blue, compressible part of the spectrum (Fig.2(a), shaded contour, transform-limit-9 fs at FWHM). This allowed to obtain more exact SPIDER measurement and a better spectral phase feedback to the DAZZLER. The best pulse compression result obtained in this way is presented in (Fig.2(b)). The FWHM duration of the main peak is 10 fs, but a broad pedestal is clearly visible. Numerical integration of temporal structure of the compressed pulse revealed that only ~55% of the compressed pulse energy is enclosed in the ±10 fs temporal window. Further improvements of the pulse compression using a newly designed pulse stretcher are under way.

Our previous experimantal [18] and theoretical [12] investigations showed that the contrast ratio between compressed pulse peak and superfluorescence background depends strongly on the seed energy and signal and on the gain of the OPCPA stages. Due to increased load to the DAZZLER the signal energy at the first stage crystal is decreased two times. To minimize the amount of generated superfluorescence we decreased gain in the first stage and increased gain in the second one to reach amplification saturation in the third stage by using the same pump beam diameter. This allowed to obtain at the system output 110 mJ pulses with the 20 mJ contained in the superfluorescence background. Furthermore saturation suppressed the shot-to-shot pulse energy fluctuation to 4.5% rms and enhanced amplification of the edges of the signal spectrum. It is evident that for the present OPCPA design the simplest and most effective route to improving the contrast is to increase the energy of the seed pulses at least to several nJ level.

 

Fig. 3. (a) Beam profile of the amplified signal after 3^rd stage; (b) Focus from a f=1.5 m concave mirror of the output beam after wavefront correction.

Download Full Size | PDF

For high-field application the spatial properties of the amplified beam are also very important. The typical spatial intensity distribution of the amplified beam at the system output is depicted in Fig. 3(a). The measured maximal spatial modulation is 35% from peak to valley. The wavefront of the beam is measured using a Shack Hartmann sensor and is evaluated to be 0.3 λ_c. This result was achieved using a homemade deformable mirror for the compensation of astigmatism induced by beam expansion telescopes. Figure 3(b) depicts the intensity distribution of the beam in the focal plane, indicating a beam diameter (46.1 µm FWHM) that corresponds to 1.1 × diffraction limit.

4. Conclusions

We have reported the generation of 10-fs, 90 mJ laser pulses at a central wavelength of 850 nm and a repetition rate of 10 Hz from a broadband OPCPA system. Due to the incomplete compression of the pulse only ~50 mJ is contained in ±10 fs temporal window. The pulses are delivered in a near-diffraction-limited beam with an energy stability of 4.5% rms. We are confident that it will become an important workhorse for high-field science.