Ultrahigh intensity lasers with peak powers over hundreds of terawatt (TW) and even over 1 petawatt (PW), based on the chirped pulse amplification (CPA) technique, have been actively developed for exploring relativistic laser-matter interactions [1–5]. We have operated two ultraintense beamlines with outputs of 1.0 and 1.5 PW at 30 fs since 2012 for research on laser-driven particle acceleration [6,7]. Currently, multi-PW lasers are being constructed or planned in a number of institutes around the world for the investigations of high field science [8–10]. In order to develop multi-PW lasers, the reduction of pulse duration is required as much as the increase of pulse energy in order to save the cost and space.

There have been steady efforts to shorten pulse durations in PW class lasers. In general, the pulse duration can be shortened by broadening the spectral width and by flattening the spectral phase of amplified laser pulses. The flattening of the spectral phase was demonstrated using a spectral shaping device in a 100 TW laser [11]. To date, however, the spectral width has not been broadened enough to support sub-20 fs pulse duration in PW level lasers. Recently, 24 fs pulse duration was obtained at PW level by broadening a seed pulse spectrum with a cross-polarized wave generation (XPW) technique and by compensating partially for gain narrowing and gain depletion effects with a spectral shaping device [12]. In this case, the spectrum after the XPW stage was broad enough to support sub-15 fs pulses, but became narrowed after subsequent amplification, increasing the final pulse duration over 20 fs. Thus, to shorten the pulse duration below 20 fs, the spectral narrowing during amplification should be minimized.

In order to obtain a broad amplified laser spectrum, two spectrum narrowing effects—gain narrowing and gain depletion—have to be handled appropriately. The gain narrowing effect due to high gain amplification stage can be overcome by replacing a high gain preamplifier with an optical parametric chirped-pulse amplification (OPCPA) amplifier. The OPCPA amplifier has been widely employed as a preamplifier in PW level lasers to generate a broadband spectrum without the gain narrowing [9,13,14]. In addition, the spectral narrowing due to the gain depletion effect, occurring while extracting the maximum energy available at the subsequent amplification stages in the CPA scheme, cannot be avoided. To overcome this spectral narrowing problem, a spectral filter can be used at the input of power amplifiers [15]. The spectral filter, however, decreases the input pulse energy and has limitation in active spectral control since it is a passive device. Thus, we modulated the output spectral amplitude of the OPCPA preamplifier to compensate for the gain depletion effect in power amplifiers by controlling the input spectral amplitude and the spectral gain of the OPCPA.

In this Letter, we describe the development of a 20 fs, 4.2 PW Ti:sapphire laser. We first reduced the pulse duration of the laser and then boosted the output energy. For the reduction of the pulse duration, the spectral width was broadened by adopting the XPW and the OPCPA techniques, and the final spectral width was maximized by compensating for the gain depletion effect. The output energy was boosted by adding a high-energy booster amplifier, and the temporal profile was optimized by minimizing the spectral phase error with a spectral shaping device. With all the modifications, we achieved the final compressed laser energy of 83 J and the pulse duration of 19.4 fs, producing the 4.2 PW laser pulses at the repetition rate of 0.1 Hz with the low energy fluctuation of 1.5% rms.

For the power upgrade to 4.2 PW, the existing 1.5 PW beamline has been significantly modified. The 0.1 Hz 1.5 PW Ti:sapphire laser was developed using the CPA technique in 2012 [3]. In this beamline, a stretched pulse with a stretching factor of 15.6 ps/nm was amplified to a pulse of 60 J using a series of CPA amplifiers, which were compressed to 45 J at 30 fs. Figure 1 shows the block diagram of the 4.2 PW laser. The red blocks represent the major parts added or modified for the 4.2 PW upgrade. The 4.2 PW Ti:sapphire laser was designed to shorten the pulse duration by employing the XPW technique and the OPCPA, as shown in Fig. 1. The XPW and OPCPA stages were installed after the frontend amplifier and the grating stretcher, respectively, to obtain a high-contrast broadband laser pulse. In order to mitigate the effect of the B-integral, resulting in hot spots, the laser pulse was stretched with a larger stretching factor of 18.7 ps/nm before the OPCPA stage and the beam expanders among subsequent amplifiers consisted of reflective optics. An additional booster amplifier was installed to increase the pulse energy over 110 J. The previous grating compressor, consisting of four gratings, was totally replaced with a new grating set for a larger beam size and a wider spectral width.

 

Fig. 1. Schematics of the 0.1 Hz, 4.2 PW CPA Ti:sapphire laser. The black and the thick red blocks represent previously existing parts and the newly added or modified parts for the 4.2 PW upgrade, respectively.

Download Full Size | PDF

The XPW technique was applied after the frontend amplifier in order to broaden the laser spectrum and to enhance the temporal contrast [16,17]. The XPW stage consisted of a 20 cm long hollow-core fiber, a 3 mm thick ${\mathrm{BaF}}_2$ crystal and a Glan-laser analyzer. A 30 fs, 3 mJ laser pulse from the new frontend amplifier (Femtopower HE, Femtolasers GmbH) was sent through the XPW stage, resulting in the increase in its spectral width and the temporal contrast by a factor of two and by 4 orders of magnitude, respectively. A detailed description of the temporal contrast is given later. As a result, the XPW output pulse had a spectral width of 107 nm (FWHM), a temporal contrast ratio of about ${10}^{-12}$, and energy of 0.5 mJ.

To maintain the broad spectrum, an OPCPA amplifier was adopted after the stretcher instead of using the previous Ti:sapphire preamplifier, additionally enabling the enhancement of the temporal contrast [18]. The OPCPA amplifier, consisting of two Type I BBO optical parametric amplification (OPA) stages in non-collinear geometry, was pumped with 450 mJ, 532 nm laser pulses from a Nd:YAG laser (Continuum, Inc.) at a 5 Hz repetition rate. The two BBO crystals, cut at 23.8°, with 15 and 4 mm in length were used for OPA. The spectral width of the signal beam was reduced before the OPCPA amplifier due to the limited high-reflection bandwidths of 45° tunable laser mirrors (TLM2, CVI), resulting in the spectral range from 730 to 875 nm. The signal beam was amplified from 20 μJ to 50 mJ through the OPCPA amplifier, and its spectrum was well maintained without any narrowing of the spectrum.

A final two-pass booster amplifier was added to increase the final output energy. The laser pulse was expanded from 60 to 80 mm in diameter before entering the final booster amplifier. The booster amplifier was pumped with a green $Q$-switched Nd:glass laser. The Nd:glass laser (Continuum, Inc.) produced 12 beams at 527 nm with 15 ns pulse duration and 0.1 Hz repetition rate, delivering a total energy of 170 J. Each pump beam with an almost flattop spatial profile and a 24 mm diameter was expanded to 80 mm diameter on a large-aperture Ti:sapphire crystal, resulting in the pump fluence of $1.7\mathrm{J}/{\mathrm{cm}}^2$ per face. The Ti:sapphire crystal in the final booster amplifier had a 100 mm diameter, a 30 mm thickness, and an absorption coefficient of $0.94{\mathrm{cm}}^{-1}$, which helped to reduce a parasitic lasing in comparison with the previous 1.5 PW laser using a 25 mm thickness Ti:sapphire crystal. A transverse parasitic lasing, which deteriorates the output spatial quality and decreases the amplified energy, was suppressed by using an index-matched absorbing cladding material and by precisely controlling the time delay between the pump pulses and a seed pulse [19,20]. Here all the pump beams arrived at the crystal at the same time. After passing through the Ti:sapphire crystal twice, the laser pulse energy was amplified to 112 J, giving an amplification efficiency of 47%, as shown in Fig. 2(a). Figure 2(b) shows the spatial profile of the amplified laser beam. The shot-to-shot energy fluctuation of the 0.1 Hz amplified laser pulses was as low as 1.5% in rms value for 50 consecutive laser pulses, as shown in Fig. 3.

 

Fig. 2. (a) Measured output energy as a function of pump energy in the final booster amplifier. (b) Spatial beam profile after the final booster amplifier.

Download Full Size | PDF

 

Fig. 3. Output energies measured at 0.1 Hz for 170 J of pump energy in the final booster amplifier.

Download Full Size | PDF

The final spectral width was maximized by shaping the spectral amplitude of the output pulse in the OPCPA amplifier. The OPCPA output spectrum was modified by controlling the input spectral amplitude with an acousto-optic programmable dispersive filter (AOPDF, Fastlite) and by adjusting the phase matching condition for the spectral gain control. To compensate for the gain depletion effect induced during the amplification processes, the long wavelength component of the OPCPA output spectrum was suppressed strongly, as shown in Fig. 4. In result, a broadband spectrum with a bandwidth of 84 nm (FWHM), supporting the Fourier-transform-limited pulse duration of 18.9 fs, was obtained after the final booster amplifier.

 

Fig. 4. Laser spectra after the OPCPA amplifier (thin solid line), the second power amplifier (dotted line), the first booster amplifier (dashed line), and the second booster amplifier (thick solid line).

Download Full Size | PDF

The amplified laser pulse was recompressed in the grating compressor. The amplified laser pulse from the final booster amplifier was expanded to 300 mm in diameter with a reflective beam expander located prior to the compressor. The compressor consisted of four 1480 groove/mm gold-coated holographic gratings with the size of the first and fourth grating being $360\times 565\mathrm{mm}$, and that of the second and third grating being $417\times 807\mathrm{mm}$. The pulse duration was optimized by minimizing the spectral phase distortion through a feedback loop between the AOPDF and a real-time spectral phase measurement device (Wizzler, Fastlite) [21]. After the feedback loop, the temporal profile of the final pulse became close to that of the Fourier-transform-limited pulse, as shown in Fig. 5. The compressor throughput efficiency was measured to be about 74%. As a result, the laser pulse had 19.4 fs pulse duration and an 83 J of energy, corresponding to a peak power of 4.2 PW.

 

Fig. 5. Reconstructed temporal profile of the 4.2 PW laser pulse (solid line) and calculated temporal profile of the Fourier-transform-limited pulse (dotted line). The inset shows the final spectrum (solid line) and the final spectral phase (dotted line).

Download Full Size | PDF

The temporal contrast of the laser pulse was enhanced by the XPW stage. The temporal contrast ratio was measured with a third-order cross-correlator (Sequoia, amplitude). The amplified spontaneous emission level at 100 ps before the main pulse was $3\times {10}^{-12}$, as shown in Fig. 6. The prepulses at 298, 239, and 221 ps before the main pulse were ghost pulses originating from the third-order correlation between the main pulse and the postpulses. The postpulses were generated by double reflection in Faraday isolators and a polarizing beam splitter. On the contrary, the prepulses at 41 and 20 ps before the main pulse were real pulses, originating from the nonlinear coupling between the main pulse and the postpulses generated in a waveplate and a window [22]. As the frontend amplifier had a temporal contrast ratio of $2\times {10}^{-8}$, it can be inferred that the temporal contrast was enhanced by about 4 orders of magnitude by employing the XPW.

 

Fig. 6. Temporal contrast of the compressed laser pulse without the pumping of two booster amplifiers.

Download Full Size | PDF

The wavefront of the laser pulse was corrected with two adaptive optics (AO) systems before and after the pulse compressor. Each AO system is composed of a wavefront sensor, a deformable mirror (DM), and a feedback loop software. The first DM with the diameter of 100 mm was installed between the final booster amplifier and the final beam expander. The image of the laser beam before the beam expander was relayed to the wavefront sensor with demagnification. After the wavefront correction, the rms value for the residual wavefront aberration was less than 0.04 μm. The second large aperture DM with the diameter of 320 mm was installed after the pulse compressor. From the preliminary wavefront correction using both AO systems, the energy concentration over 60% inside the Airy disk was achieved.

In conclusion, we developed a high-contrast 4.2 PW Ti:sapphire laser with a 0.1 Hz repetition rate. To achieve the 4.2 PW peak power, a high-energy booster amplifier was added while adopting XPW and OPCPA techniques to broaden the spectral width of the seed laser pulse to amplifiers. The temporal contrast was enhanced by implementing the XPW and the OPCPA stages. The final spectral width was maximized by shaping the spectrum of the OPCPA output for the compensation of the gain depletion effect during amplification. Furthermore, the final spectral phase was optimized with the AOPDF, resulting in 19.4-fs duration, which is the shortest duration in multi-PW Ti:sapphire lasers to the best of our knowledge. This 0.1 Hz, 4.2 PW laser will boost greatly the progress in the fields of laser-driven particle acceleration and high field science.