2 Laser setup

The laser system is based on the double CPA scheme^{[Reference Kalashnikov, Risse, Schönnagel and Sandner12]} which includes a 1 kHz Ti:sapphire CPA laser, an OPA–SHG temporal pulse cleaning device, an Offner stretcher, an OPCPA pre-amplifier, two Nd:glass rod amplifiers, and a Treacy compressor, as shown in Figure 1.

Figure 1. Schematic overview of the double CPA laser system.

As shown in Figure 2, the commercial Ti:sapphire CPA laser (Spitfire, Spectra-Physics) outputs 800 nm, 40 fs, 3.3 mJ pulses with a repetition rate of 1 kHz. It is used as the source and pump laser of the pulse cleaning device. In the pulse cleaning device, a small part of energy of the 800 nm laser is focused in a sapphire plate to generate white-light continuum pulses. In the first OPA stage, the 1290 nm components of the white-light continuum pulses are used as the signals to generate 2106 nm idlers with the 800 nm pump. After two following OPA stages, the 2106 nm pulses are further amplified to about $450~\unicode[STIX]{x03BC}\text{J}$ . Finally, the pulses pass through a $\unicode[STIX]{x1D6FD}$ -barium borate (BBO) crystal for frequency doubling to generate $220~\unicode[STIX]{x03BC}\text{J}$ , 1053 nm cleaning pulses.

Figure 2. Layout of the OPA–SHG pulse cleaning device. BS1–BS3: beam splitter; ATT: attenuator; L1–L6: lens; SAPH: sapphire plate; $\text{M}_{\text{Ag}}1$ , $\text{M}_{\text{Ag}}2$ : silver mirror; DM1–DM6: dichroic mirror; M1–M15: 800 nm HR mirror; m1, m2: 2106 nm HR mirror.

The spectrum is shown in Figure 3(a). Most of the energy is concentrated in a 20 nm bandwidth range near the gain peak of Nd:glass (1053 nm). The autocorrelation curve of the pulse is shown in Figure 3(b), which is corresponding to 156 fs (assuming Gaussian shape). The scanning cross correlation measurements (Sequoia-800 and Sequoia-1000, Amplitude Technologies) are shown in Figure 4. The measured contrast of the cleaned pulse is limited (about $10^{-11}$ ) by the dynamic range of the measurement device. The satellite pulse pair at $\pm 10~\text{ps}$ disappears in the contrast curve of 1053 nm pulses, which shows the contrast is improved at least by 8 orders. There is a prepulse with intensity of about $10^{-7}$ at 51 ps before the main pulse. It may be generated by surface reflections and be amplified in OPAs.

Figure 3. (a) Measured spectrum and (b) autocorrelation trace of the OPA–SHG cleaning pulses.

Figure 4. Third-order correlation contrast measurements of the 800 nm pump pulses and the 1053 nm OPA–SHG cleaning pulses.

The temporal contrast of the amplified pulse in a CPA system depends not only on the contrast of the seed pulses but also on the energy of the seed^{[Reference Ivanov, Maksimchuk and Mourou13, Reference Bagnoud, Zuegel, Forget and Le Blanc14]}. In the case of ps-OPA, limited by the energy and pulse duration of the pump pulses, the energy of the cleaned pulse is usually at the level of $\unicode[STIX]{x03BC}\text{J}$ or tens of $\unicode[STIX]{x03BC}\text{J}$ . Compared with ps-OPAs, our fs-OPA–SHG cleaned pulses have two main advantages: firstly, the energy of the cleaned pulse is higher because the shorter pump duration is used which leads to higher efficiency; secondly, higher contrast could be achieved because the doubling frequency process improves the contrast further after three OPA stages^{[Reference Huang, Zhang, Xu, Li, Leng, Li and Xu15]}.

In the 1053 nm CPA stage, the cleaned fs pulses are stretched to about 0.95 ns by a two-pass Offner triplet stretcher as shown in Figure 5(a). The stretcher consists of a $1740~\text{grooves}~\text{mm}^{-1}$ multilayer-dielectric grating, a concave spherical mirror ( $R=-1000~\text{mm}$ ), and a convex spherical mirror ( $R=500~\text{mm}$ ). The incident angle on the grating is about $71.7^{\circ }$ and the average distance between the grating and the concave mirror is about 750 mm. A prism is inserted at the exit of the first pass to reflect the pulses back for a second pass in the stretcher. The total transmission efficiency of the stretcher is 25%, indicating an output energy of $55~\unicode[STIX]{x03BC}\text{J}$ . The full spectrum width of the stretched pulse is about 18 nm as shown in Figure 5(b).

Figure 5. (a) Pulse durations; (b) spectra of the amplified pulse.

After the stretcher, the laser pulse is amplified in a two-stage OPCPA pre-amplifier^{[Reference Lu, Peng, Li, Guo, Leng, Sui, Xu and Wang10]}. The OPCPA pre-amplifier includes a 20-mm-thick BBO crystal in the first OPCPA stage and an 8-mm-thick BBO crystal in the second OPCPA stage. Both of them are cut at $22.85^{\circ }$ for type I phase matching. The two OPCPA stages are pumped in a cascade way by a single longitudinal mode $Q$ -switched Nd:YAG laser (Quanta-Ray lab-150, Spectra-Physics) which can generate 5.5 ns, 532 nm pulses with a repetition of 10 Hz. In the first OPCPA stage, the signal pulse is overlapped by the pump pulse completely while there is 1.5 ns additional time delay between the pump pulse and the signal pulse considering the second OPCPA stage. The energy of the pump laser is adjusted to about 170 mJ. The top-hat pump beam is relay imaged and down-collimated to 4 mm on the first OPCPA crystal. The 1053 nm signal pulse is amplified to about 15 mJ corresponding to a total gain of 300 in the two-stage OPCPA amplifier.

Both the characters of the amplified pulse (Figure 5) and the depletion of pump energy (Figure 6) indicate that the OPCPA is operating in a nonsaturation state. By blocking or opening the seed pulses of the OPCPA, we can measure that the intensity ratio between the parametric fluorescence and the amplified pulse is about $3\times 10^{-7}$ using a photodiode and a set of calibrated neutral density filters. Multiplying by a compression factor of about $2\times 10^{-4}$ , the relative intensity of the parametric fluorescence after pulse compression is about $6\times 10^{-11}$ . Considering the suppression of the signal injection, the real intensity of the parametric fluorescence will be lower.

Figure 6. Time profiles of the pump pulse in the OPCPA pre-amplifier.

A nonsaturated OPCPA is beneficial for high-contrast amplification^{[Reference Kiriyama, Mori, Nakai, Shimomura, Tanoue, Akutsu, Okada, Motomura, Kondo and Kanazawa16]}, but it is adverse to obtain the homogeneous beam. As shown in Figure 7(c), the beam profile of the OPCPA amplified pulse is seriously affected by the poor pump beam profile sensitively. To improve the homogeneity of the signal pulse beam profile, the signal pulse beam is expanded by a Galileo expander, and then a $\unicode[STIX]{x1D719}4~\text{mm}$ diameter signal pulse beam with good profile is picked out from the expanded beam by a serrated-tooth apodizer. After filtering off the high-frequency components, the signal pulse beam is homogenized as shown in Figure 7(d), and 2.1 mJ of energy is remained.

Figure 7. Beam profiles in the OPCPA pre-amplifier: (a) seed beam profile; (b) pump beam profile; (c) OPCPA amplified beam profile; (d) picked-out homogenized beam profile.

Figure 8. Layout of the two Nd:glass rod amplifiers. TFP1, TFP2: thin film polarizer; QWP1, QWP2: quarter wave plate; LM1, LM2: single-flashlamp-pumped Nd:glass rod module; FR1, FR2: Faraday Rotator; FI: Faraday Isolator; LM3, LM4: double-flashlamp-pumped Nd:glass rod module.

After the OPCPA pre-amplifier, the pulse is amplified by following two Nd:glass amplifiers, which are upgraded from the previous 100-mJ-level amplifier^{[Reference Lu, Peng, Li, Guo, Leng, Sui, Xu and Wang10]}. As shown in Figure 8, the first amplifier consists of a quarter wave plate, two $\unicode[STIX]{x1D719}7~\text{mm}$ single-flashlamp-pumped Nd:glass rod modules, a $90^{\circ }$ quartz rotator, a Faraday Rotator, a thin film polarizer, a 1:1 telescope system. The telescope makes the principle planes of the two rods image each other. The imaging system and the $90^{\circ }$ quartz rotator exchange the tangential polarization states of the two rods to compensate thermal-induced birefringence. A pinhole filter is set on the focus plane to filter off the high-frequency components of the beam. The quarter wave plate is inserted before the first glass module to generate the circular polarization for reducing the B-integral in the amplifier. The pulse is amplified twice in this amplifier stage, leading to an output of 45 mJ. The repetition rate is set at 0.5 Hz.

Before injected into the following $\unicode[STIX]{x1D719}12~\text{mm}$ Nd:glass rod amplifier, the laser beam is expanded to $\unicode[STIX]{x1D719}10~\text{mm}$ . The second amplifier has the similar structure as the first one, and the imaging system is set in a vacuum environment to avoid laser-induced breakdown in the focus plane (Figure 8). The pulse energy is boosted to 1.9 J at the repetition rate of 0.5 Hz. The total gain of the two Nd:glass rod amplification stages is about 900.

The spectra and the time profiles of the amplified pulse are shown in Figure 5. Because of the gain narrowing effect in the Nd:glass amplifiers, the spectrum of the amplified pulse reduces to about 4.7 nm (FWHM), and the duration also reduces to about 0.65 ns (FWHM).

The near-field beam profile of the amplified pulse is shown in Figure 9. The strong area in the beam profile limits further energy improvement of the amplified pulse. This strong area in the vertical direction is mainly caused by the single-flashlamp-pumped rod modules in the first Nd:glass rod amplifier. The uniform pumped modules will be used to replace the present modules in the future.

Figure 9. Near-field beam profile of the amplified pulse.

The sampling pulse of the amplified pulse is sent to the compressor. The compressor consisted of two $1740~\text{grooves}~\text{mm}^{-1}$ golden gratings and a roof mirror. The autocorrelation measurement shows that the duration of the compressed pulse is about 328 fs (Figure 10, assuming Gaussian shape), while the Fourier transform limit of the amplified pulse (Figure 5) is 295 fs.

Figure 10. Autocorrelation measurement and Fourier-transform-limited pulse duration.

3 Contrast measurement

The third-order correlation scanning contrast measurement of the compressed pulse is shown in Figure 11. Limited by the low operating repetition and the low long-time stability of the laser system, we choose a scanning step of 2 ps except the range near the main pulse In the measurement, every point is averaged by 4 times. Therefore, it takes about one hour to complete a contrast measurement. Here, the prepulse at 51 ps before the main pulse is missed due to 2-ps step and mixed in the noise (about $10^{-8}$ ).

Figure 11. Third-order correlation contrast measurement of the compressed pulse.

It shows that the contrast is about $10^{-8}$ in the tens of ps range before the main pulse. The measurement also shows that a slow rising slope at the level of $10^{-10}$ starts from about 200 ps before the main pulse. The ASE noise is the main contrast degradation origin for most CPA lasers in which the pulses from oscillators are used as the seeds directly. Here, as the seed pulses for the Nd:glass laser system are high contrast and high energy, ASE noise can be suppressed efficiently. As shown in our previous report^{[Reference Lu, Peng, Li, Guo, Leng, Sui, Xu and Wang10]}, the contrast pedestal comes from the stretching–compressing process mainly. The theoretical analysis of Dorrer et al. indicated the high-frequency spectral phase noise in stretcher or compressor could impact the contrast^{[Reference Dorrer and Bromage17]}. It was also indicated that the convex mirror in Offner triplet stretcher has an important impact on the contrast degradation^{[Reference Bromage, Dorrer and Jungquist18]}. In the future, we will optimize our stretcher design to improve the contrast further.