1. Introduction

Few-cycle ultrashort high-power lasers play a key role in many laser applications, such as generations of x-ray pulses and electron beams, and intense field-matter interaction [1]. Optical parametric chirped-pulse amplification (OPCPA) [2–6] offers a particularly promising route toward compact ultrashort ultrahigh-peak-power laser systems because of its advantages including ultrabroad gain bandwidth, high pulse contrast, good beam quality, negligible thermal load on a nonlinear crystal and so on. For efficient and stable operation of an OPCPA system, it is very important to synchronize accurately its signal and pumping beams. Conventionally, the seed and the pumping sources of OPCPA come from independent lasers. High-power few-cycle OPCPA tends to choose the signal seeds from ultrabroadband Ti:sapphire fs lasers and the pump pulses from high-power frequency-doubled Nd-doped lasers. A conventional electronic synchronization method can typically lock pump and signal pulses from independent lasers with timing jitters about 100 ps. In an effort to improve the accuracy of synchronization between independent lasers, a hybrid method has been invented by careful control of laser cavity length together with some phase-locked loop electronics. The timing jitter has been recently locked within about 2 ps between the 600 ps pump and the 300 ps signal pulses [9], and within 100fs between the 7 ps pump and the 3 ps signal pulses [10] from two independent lasers. However, it is still experimentally difficult to maintain the cavity lock of independent lasers with a satisfactory long-term stability. Recently, some optical synchronization schemes [12–15] have been demonstrated as robust methods to reduce timing jitter based on the basic idea that the pump and the seed pulses are provided by the same source. An OPCPA at 1064 nm with an output power up to 16.7 TW has been realized by operating a Ti:sapphire oscillator at the central wavelength of 1064 nm as the seeds of both the pump and signal, where a timing jitter of about 10 ps can be maintained during the pump pulses are boosted to high energies. However, the narrow gain bandwidth of Ti:sapphire oscillator at 1064 nm increases the signal pulse duration up to 120 fs. Another straightforward way for this purpose is to generate the pulses at 1064 nm as the seed of the pumping source for few-cycle OPCPA with nonlinear frequency conversion, which guarantees automatic synchronization with the ultra broadband signal around 800 nm. All-optical synchronization of the pump and signal pulses has been realized by transferring a fraction of a broadband seed pulse energy at 800 nm to 1064 nm in a photonic crystal fiber [14], and by generating background-free 1064 nm pulses through two-stage non-collinear optical parametric amplifications (NOPAs). In the two-stage NOPA scheme, the NOPA pumped by 400 nm intense laser pulses from a frequency-doubled Ti:sapphire regenerative amplifier is seeded with a CW Nd:GdVO₄ laser at the first stage, where a second-stage NOPA is used to get pulses at 1064 nm free of CW background [13].

In this paper, a CW He-Ne laser is used to replace the Nd:GdVO₄ laser [13] so that background-free 1064 nm pulses can be generated by only one-stage NOPA rather than two. Typically, the output of a CW He-Ne laser at 632.8 nm operating at TEM₀₀ ranges from several to several tens of milliwatt. Such a weak CW seed may be of comparable intensity with the parametric fluorescence. In order to enhance the CW seed power, a novel design, CW-seeded intracavity non-collinear parametric amplification (IC-NOPA), is presented to generate strong enough 1064 nm pulses which are synchronized accurately with the Ti:sapphire fs pulses. Such an all-optical synchronization scheme does not need any additional devices, not even an optical delay line. The nonlinear crystal of the IC-NOPA is set at one of waist positions of the He-Ne cavity in order to increase the intensity of the CW 632.8 nm signal thus control efficiently parametric fluorescence. Our result shows the single pulse energy of the background-free 1064 nm pulses is obtained as high as 6 nJ through only one-stage NOPA, which can be used as the seeding pulses of the laser amplification chain to produce the strong pumping pulses for few-cycle OPCPAs.

2. Experimental setup

As shown in Fig. 1, a 45 fs, 600 mW, 1-kHz pulse chain is provided by a Ti:sapphire regenerative amplifier (Spitfire, Spectra-Physics). The central wavelength of the output pulses is tuned around 794 nm with the bandwidth of 26 nm. After a half-wave plate, an afocal optical system, consisting of a concave mirror M2 (HR@800 nm) and a fused-quartz lens L1, is used to reduce the beam size from 8 to 3 mm. The dispersion induced by the wave plate and lens is pre-compensated by adjusting the grating pulse compressor of the Ti:sapphire laser system. About 35% second harmonic (SH) conversion efficiency is achieved with a 29.2°-cut 0.2-mm-thick type I phase-matched beta-barium borate (β-BBO) crystal (C1), corresponding to 210 mW SH average power. The 397 nm beam is then reflected by a dichronic mirror (DM) to another 29.2°-cut 2-mm-thick type I phase-matched β-BBO crystal (C2) as pumping beam for NOPA, while the fundamental beam is transmitted through the DM. A CW He-Ne laser at 632.8 nm is focused into the β-BBO crystal C2 as the signal beam. According to the energy conservation, the central wavelength of the idler pulses is at 1064 nm. These pulses are inherently synchronized with the 397 nm beam thus with the 794 nm broadband laser pulses.

The He-Ne laser used for the IC-NOPA is an improved version of the Model 1200A made by Shanghai Institute of Laser Technology. Its original configuration consists of one concave mirror coated HR at 632.8nm with a radius of curvature as 5.0 m (M₆), one output coupler with 2.5% transmission (M₇), and a tube filled with gain medium. The separation between M₆ and M₇ is about 1.45 m. Its output beam diameter is about 1.4 mm, which agrees with our theoretical estimation. For the sake of high efficient OPA, the design of the He-Ne laser resonator shall be changed so that there is one beam waist outside the tube. As shown inside the dotted box in Fig. 1, the new cavity includes concave mirrors M₄, M₅, and M₆ coated HR at 632.8 nm. In order to ensure the optimal output performance, the new resonator is designed to ensure that it has the same round-trip matrix as that of the original one, which implies

where L₁ stands for the separation between M₅ and the original position of M₇, and L₂ = R₄ + R₅/2, for that between M₅ and M₄. R₄, R₅ and R₆ are the curvature radii of M₄, M₅, and M₆, respectively. Figure 2 shows our measured output power behind the HR mirror M₄ (R≈99.9%) versus L₁ when R₆=5000 mm, R₅=500 mm, and R₄=500 mm. It is clear that the maximal output power occurs at L₁=375 mm, corresponding to L₁=0.75 R₅, which matches our theoretical prediction very well.

 

Fig. 1. Schematic of the experimental setup. λ/2: half wave plate at 800 nm; M1: mirror with 45° HR at 800 nm; M2: concave mirror with 0° HR at 800 nm; M3: mirror with 45° HR at 400 nm; M4, M5 and M6: mirrors with 0° HR at 632.8 nm; M7: output coupler of original He-Ne laser; DM: mirror with 45° HR at 400 nm and AR at 800 nm; L1, L2: lens; C1, C2: β-BBO crystal.

Download Full Size | PDF

The maximal output power of the original He-Ne laser is about 20 mW when the reflectance of M₇ is 97.5%. After our modification, the maximal output power measured behind the HR mirror M4 is about 0.72 mW, corresponding to an intracavity power P_in larger than 720 mW. The 29.2°-cut 2-mm-thick type I phase-matched β-BBO crystal (C2) for OPA is set at the waist position between M₅ to M₄ inside the cavity, both surfaces of which are coated with AR at 632.8, 400, and 1064 nm. The spot size of the CW beam at 632.8 nm is estimated at 0.15 mm. However, once the BBO crystal was insert into the cavity, the measured power behind the M₄ dropped down to about 0.20 mW, thus P_in ≈ 200 mW. The measured value is lower than that of our prediction. We think it may be caused by the large inserting loss of the C2 crystal due to the bad AR coating quality at 632.8 nm on both surfaces of the C2, which are measured about 1% from each surface.

 

Fig. 2. the output power of the modified He-Ne laser measured behind the HR mirror M₄ (R~99.9%) vs. the separation between M₅ and M₄

Download Full Size | PDF

3. Experimental measurements and discussions

The alignment for the IC-NOPA to get maximal 1064 nm idler pulses can be described as follows. Firstly, we use M₇ as an outcoupler with the transmission of 2.5% so that the He-Ne laser operates with its original configuration. The output power is measured about 20 mW. Then we align carefully the crystal C2, M₅, the pumping and the CW beams to optimize the OPA. Finally, we remove M₇ and align M₄ to get the optimal laser performance among M₄, M₅ and M₆ thereby the maximum power of the 1064 nm pulse chain can also be achieved.

Figure 3(a) shows the measured SH beam is around 396.8 nm with a broad bandwidth of about 6 nm. Figure 3(b) is the spectrum of the signal at 632.8 nm recorded behind M₄, which is almost the same as that directly from the He-Ne laser. Since the amplification occurs within a time scale of several hundred femtoseconds, which is much smaller than the temporal interval between adjacent pumping pulses, the recorded spectra are still dominated by the CW background. As shown in Fig. 3(c), the spectrum of the idler pulses is centred at 1064 nm with a bandwidth of about 14 nm. The corresponding SH pulses are around 532 nm with a bandwidth of 5 nm (see Fig. 3(d)). The existence of the SH beam indicates the idler pulses have very high peak power due to their short pulse duration. It is easy to understand that the broad bandwidth of the idler pulses results from the broad bandwidth of the pumping because of the requirement of energy conservation in OPA. The idler pulse energy in this CW-seeded IC-OPA is measured about 6 nJ. Based on the measured spectrum as shown in Figure 3(c), it can be concluded more than 0.4 nJ from 6 nJ pulse energy contains within the gain bandwidth of Nd:YAG. According to reference 14, about 2 pJ pulse energy is proved to be sufficient to obtain clean pulse amplification and suppress the nanosecond Q switched-pulse background. Obviously, our idler pulses energy are large enough for the seed pulses of a Nd:YAG amplifier. Here, broad bandwidth is unnecessary. However, the broad bandwidth is quite helpful to ensure there is enough energy at 1064 nm in the idler pulses in spite of the spectral instability of the Ti:sapphire regenerative amplifier. As shown in Fig. 4, the autocorrelation duration is measured about 60 fs for the pump pulses from the frequency-doubled Ti:sapphire laser, while 205 fs for the generated idler pulse. From the measured spectra in Figs. 3(a) and (c), it is obvious that the spectral narrowing broadens the pulse duration up to about 160 fs and the dispersion from the nonlinear crystal (C2) stretches it further to 205 fs.

The generated idler pulses show their elliptic spatial intensity distribution in our experiment. It is easy to understand that there exists some spatial chirp in the spatial intensity distribution of the idler beam. For effective optical parametric amplification pumped by boradband laser pulses and seeded by narrow bandwidth CW beam, the propagating directions for different spectral components of the idler pulses are dominated by the phase matching condition. In our setup, the dispersion angle of the idler pulses is estimated theoretically about 5.0×10^-4 radian/nm, which agrees with our experimental results very well. Assume that the waist radius of the idler is about 0.15 mm, the same as that of the signal there, the divergent angle of the idler can be calculated about 2.2×10^-3 radian, which is more than 4 times larger than the dispersion angle for a bandwidth of 1 nm. For an efficient OPCPA operation, it is necessary to stretch the pumping pulse duration long enough to overlap temporally with the chirped seed. In general, before or during the amplification of the 1064 nm beam, some steps, for example, spectral filtering [10] or inserting an appropriate etalon into the cavity of a regenerative amplifier [14], are taken to narrow the bandwidth thus widen the pulse duration up to hundreds of picoseconds. The resultant bandwidth shall be more than 0.02 nm for 100 ps pulses at 1064 nm. From above analyses, it can be concluded that the spatial chirp of the idler is so weak that it does not degrade the spatial intensity distribution of the 1064 nm beam if the pulses pass through a Nd-doped, e.g. Nd:YAG, amplifier. The generated idler pulses are CW background-free, which make them be qualified as the seeding pulses of a regenerative amplifier for upscaling pulse power, improving intensity stability and spectral/temporal shaping [15].

 

Fig. 3. The measured spectra of (a) the pump pulses from the SH of the Ti:sapphire fs laser, (b) the 632.8 nm beam, (c) the generated idler pulses and (d) their SH pulses of the CW-seeded IC-NOPA

Download Full Size | PDF

 

Fig. 4. The autocorrelation trace of (a) the pump pulses from the doubled-frequency Ti:sapphire fs laser and (b) the generated 1064 nm pulse after the CW-seeded IC-NOPA

Download Full Size | PDF

The impressive advantage of our design is the inherent synchronization between the 1064 nm and 794 nm pulses. This design can be used to develop efficient and stable few-cycle OPCPA systems. For instance, output beam from a Ti:sapphire laser oscillator operated at 794 nm with ultrabroad bandwidth can be split into two parts. One part is used as the ultra-broadband signal of the OPCPA system, and the other part is seeded into a Ti:sapphire fs regenerative amplifier to generate strong 397 nm pulses. The 397 nm pulses are used as the pump source of the CW-seeded IC-NOPA to achieve the 1064 nm idler pulses which are synchronized with the 397 nm pulses thus with the ultrabroad bandwidth 794 nm pulses.

The most important contribution to the time jitter in our system results from the cavity length drifts of the Ti:sapphire fs laser system, which was measured about 10 fs in a few seconds and less than 200 fs in one hour [16]. Since the time span of the optical pulses before and after our Ti:sapphire fs regenerative amplifier is about 350 ns, the timing jitter between the broadband OPCPA signal at 794 nm and the generated 1064 nm idler from the CW seeded NOPA can be estimated as ~10 fs. The short-term drift caused by mechanical instability and thermal expansion can be disregarded, because the period of conceivable vibrations and thermal expansion in a laboratory is much larger than 350ns. The timing jitter due to the spectral stability of He-Ne laser is also negligible [15]. The peak-to-peak average power stability of the idler pulses at 1064 nm is measured about ±8% within about one hour, which can be improved greatly after an Nd-doped regenerative amplifier.

4. Conclusion

In summary, we have demonstrated a novel method to synchronize the background-free 1064 nm pulses with broadband 794 nm pulses from Ti:sapphire laser through a CW-seeded IC-NOPA, which is very simple, robust and easy to maintain. The intracavity power of the CW 632.8 nm laser used as the signal beam is about 200 mW, and the corresponding single pulse energy of the idler is about 6 nJ. Their pulse duration can be stretched up to tens even hundreds of picoseconds by narrowing the bandwidth of the pulses. The spatial chirp of the 1064 nm pulses due to the phase matching condition is also discussed both theoretically and experimentally. After spectral shaping for pulse widening, the great spectral narrowing of the idler makes the spatial chirp be so week that it does not degrade the spatial intensity distribution of the 1064 nm beam if the pulses pass through an Nd-doped amplifier. After being boasted by the Nd-doped laser amplification chain, these pulses can be frequency-doubled and then used as the pump for optical parametric amplification seeded by the ultra-broadband chirped pulses from the same Ti:sapphire laser with very accurate synchronization.