With the development of chirped-pulse amplification (CPA) technology, numerous petawatt (PW)-class laser facilities have been constructed around the world^[1], such as the BELLA laser^[2], the J-KAREN-P laser^[3], the ELI beamlines^[4], and the Shanghai Superintense Ultrafast Laser Facility (SULF) beamlines^[5,6]. The PW-class laser can produce focused laser intensities up to ${10}^{22}–{10}^{23}\mathrm{W}/{\mathrm{cm}}^2$ for high-field physics applications^[7–9]. At such high intensity levels, the temporal contrast of laser pulses has become one of the most important issues for laser–matter interaction experiments^[10,11]. The temporal contrast is generally defined as the ratio of the intensity of the pedestal or subpeaks to the intensity of the main pulse^[12]. The pedestal and subpeaks mainly include: (i) amplified spontaneous emission (ASE) on the nanosecond (ns) time scale; (ii) the pedestal on the picosecond (ps) time scale, arising from spectral clipping or modulation; and (iii) the pre-pulses on ps/ns time scales, originating from surface reflection or scattering^[9,11]. The pedestal or subpeaks with intensities of ${10}^{10}–{10}^{11}\mathrm{W}/{\mathrm{cm}}^2$ prior to the main pulse would ionize solid targets in laser–matter interactions^[10,13].

Many techniques have been proposed and developed to improve the temporal contrast of high-power lasers, such as cross-polarized wave generation (XPW)^[14], optical parametric amplification (OPA)^[15], and plasma mirrors (PMs)^[16–20]. Several approaches have been demonstrated in the SULF-1PW beamline before, such as the XPW and OPA filter^[6]. Recently, a PM system was developed in the SULF-1PW beamline and tested for a proton acceleration experiment. For the PM, the substrate is usually a polished glass plate that is coated with a broadband antireflection coating^[19]. When laser is focused on the surface of the PM, the low intensity pedestal or prepulses pass through the PM substrate with low reflectivity. Then the leading edge of the main pulse would ionize the surface of the PM substrate and generate the high-density plasma^[21,22]. Therefore, the subsequent main pulse would be reflected by the high-density plasma surface^[23]. The PM provides an ultrafast, self-induced optical switch for temporal contrast enhancement with the unlimited energy and power of the input laser.

In this paper, the characterization of the laser contrast using the PM system in the SULF-1PW beamline is reported. This PM system improves the temporal contrasts on ps and ns time scales by about 2 orders of magnitude. To evaluate the performance of this system, we preliminarily performed proton acceleration contrast experiments. It shows that the proton beams accelerated by the high-contrast (HC, with PM system) laser have significantly improved cutoff energy and quantity compared to the low-contrast (LC, without PM system) case. In addition, simulation studies show that the target with a larger initial expansion scale (due to prepulse) had a lower cutoff energy of the proton beam. These results demonstrate that our PM system effectively improves the laser performance and can be well applied for laser-driven proton accelerations.

The experiments were carried out on the SULF-1PW beamline platform. The schematic of the PM system and the proton acceleration experiment is shown in Fig. 1. In experiments, the laser pulse duration was 27 fs at full width at half-maximum (FWHM), with an 800 nm center wavelength. The p-polarized laser beam was focused by using an f/4 off-axis parabola (OAP1) high-reflection mirror that had a 980 mm focal length with an off-axis angle of 20 deg. The PM substrate was clamped on a three-dimensional motorized translation stage placed in front of the focus position. It should be noted that the PM substrate used in our case was essentially a piece of fused silica glass whose surface was polished and coated with a broadband antireflection coating. The laser transmittance was greater than 99.5% when the PM was not excited. In order to avoid the effect of laser pulse reflected from the rear surface of the PM, the rear surface was also coated with the antireflection coating, and the wedge angle (3 deg) had been designed between the rear and front surfaces of the PM. In addition, the upper and lower parts of the PM were coated with reflective silver to facilitate alignment. After the PM, the divergent beam was recollimated into the parallel beam by OAP2, which had the same parameters as OAP1 and was placed symmetrically with OAP1. Subsequently, the temporal contrast-enhanced laser was reflected by mirror (R3) into the target chamber.

The proton acceleration experiments were carried out in the target chamber. As shown in Fig. 1, the laser beam was finally focused by OAP3, which had a 998 mm focal length with an off-axis angle of 26 deg. The 50 nm thickness hydrocarbon target used in the experiment was placed at the laser focus, and the laser incident angle was 30 deg. The radiochromatic film (RCF) stack was placed behind the target at a distance of 80 mm from the target to measure the spatial intensity distribution of the proton beams. It was wrapped with a layer of 14 µm aluminum film to block the X-ray signal. A Thomson spectrometer was placed 420 mm away from the target in the normal direction. It should be noted that the R1 would be removed and R3 would reflect laser directly to target chamber by moving and rotating for the case without the PM.

Before measuring the temporal contrast of the laser, it is necessary to find the optimal reflectivity efficiency of the PM used in our case, i.e., to measure the numerical relationship between the intensity of the focal spot on the surface of the PM and the reflectivity. The intensity of laser beam on the PM was varied from $1\times {10}^{15}$ to $9\times {10}^{15}\mathrm{W}/{\mathrm{cm}}^2$ by adjusting the spot size on the surface by changing the distance between the PM and OAP1 while the input laser energy was fixed at $\sim 13.5\mathrm{J}$. The energy ratio of the reflected and incident beams of the PM system was measured for different intensity on the PM, as shown in Fig. 2. This indicates that the reflectivity reaches about 75% when the intensity is between $2.5\times {10}^{15}\mathrm{W}/{\mathrm{cm}}^2$ and $8.9\times {10}^{15}\mathrm{W}/{\mathrm{cm}}^2$ in our case. In addition, when the intensity was greater than $2\times {10}^{16}\mathrm{W}/{\mathrm{cm}}^2$, the far-field distribution of the reflected laser pulse was distorted, indicating that beam cannot be properly focused. The worse ability to focus may be owing to the distortion of phase front by the damaged PM surface, which can be caused by the several-picosecond rising edge of the main pulse^[24] or by the prepulses, which were more than 10 ps prior to the main pulse^[25]. Therefore, considering the wider variation range of input laser energy for flexible application requirements, the middle intensity of $5\times {10}^{15}\mathrm{W}/{\mathrm{cm}}^2$ was chosen, meaning about 8 to 27 J of the input laser energy was allowed for our PM system.

The temporal contrasts on the ps time scale were measured by employing a high-dynamic-range third-order cross correlator (Sequoia, Amplitude Technologies), while the temporal contrasts on the ns time scale were measured by employing a fast photodiode (ET2030, Electro-Optics Technology) with neutral density filters. The measurement limits for the photodiode and Sequoia are ${10}^{-8}$ and ${10}^{-12}$, respectively. The measurement results of temporal contrasts with PM and without PM are shown in Fig. 3. It should be noted that the temporal contrasts with PM were measured after R6, while the temporal contrasts without PM were measured with R1 removed and laser reflected directly to the target chamber, as mentioned at the end of the previous section. As shown in Fig. 3(a), three prepulses at $-7.5$, $-11.3$, and $-15.7\mathrm{ns}$ were detected without using PM. Their temporal contrast ratios were $\sim 3.22\times {10}^{-8}$, $\sim 2.27\times {10}^{-8}$, and $\sim 4.07\times {10}^{-8}$, respectively. The prepulses were mainly generated from surface reflection or scattering in the amplifiers and compressor. By using PM, the prepulses were suppressed to an undetectable level, and the ASE pedestal was also suppressed effectively, as shown in Fig. 3(a). It could be estimated that the temporal contrast on the ns time scale was improved by about 2 orders of magnitude by the PM^[22]. For the measurement of the temporal contrast on the ps time scale, the third-order cross correlator works by scanning data acquisition, i.e., one data point can be collected for one laser pulse. It can work well at fully automatic mode for the measurement of repetitive laser pulses. When the PM was employed, the surface of the PM irradiated by the laser would be ionized, and then the PM should be moved to a fresh surface for a subsequent laser pulse. Consequently, the third-order cross correlator had to work at the manual mode, which would limit the number of data points. Finally, the 30 discrete data points were collected, as shown in Fig. 3(b). Comparing the temporal contrasts on the ps time scale with and without the PM, it is not difficult to find that the PM system could suppress the ps scale pedestal by about 2 orders of magnitude, i.e., the temporal contrast at $-2.8\mathrm{ps}$ was improved from $\sim 1.55\times {10}^{-6}$ to $\sim 2.03\times {10}^{-9}$. Therefore, the PM system can obviously improve the temporal contrast on both the ns and ps time scales, which has great significance for many physical experiments.

The comparative studies were made with and without the PM system. The laser energy was fixed to $\sim 9\mathrm{J}$ on the target with a focal spot radius of $\sim 3\mathrm{\mu m}$ (FWHM), corresponding to the intensity of $\sim 5\times {10}^{20}\mathrm{W}/{\mathrm{cm}}^2$. Figures 4(g) and 4(h) show the ion imaging trajectory collected by the energy spectrometer. The proton energy spectrum can be resolved, as shown in Fig. 4(i). This indicates that the cutoff energy of the proton beam obtained by the HC laser-driven acceleration is higher than in the LC case: the former reaches $\sim 9.2\mathrm{MeV}$, while the latter is $\sim 6.1\mathrm{MeV}$. In addition, the flux of the proton beam in the HC case, whose trajectories appeared saturated, as shown in Fig. 4(g), was significantly higher than that in the LC case, as shown in Fig. 4(h). This phenomenon was more apparent with RCF stacks, as shown in Figs. 4(a)–4(f). It also indicates that under the HC laser condition, the divergence of the proton beam was better than the LC laser results. The reason for the formation of an annular proton beam [see Figs. 4(d)–4(f)] may be the deformation and transverse nonuniform distribution of target caused by the prepulse^[26]. It should be noted that the results of experiments are repeatable in our case. Such results showed that the laser contrast played an important role in the process of proton acceleration in our experiments.

The prepulses can affect the interaction when the main pulse irradiates on targets, because prepulses arrive at the target prior to the main pulse and expand the target. A combination of hydrodynamic codes and particle-in-cell (PIC) simulation to handle the whole interaction driven by prepulses and main pulses has been researched in previous studies^[27]. However, for ultrathin solid targets with a thickness of tens of nanometers close to the molecular scale, it is not suitable for hydrodynamic simulations (at the ns scale) anymore^[28,29]. Then, a simple model was used to estimate the effect of prepulses in our cases. The interaction of the femtosecond laser with different expanded targets was simulated by using the two-dimensional PIC code (EPOCH^[30]).

The laser parameters in the PIC simulations were the same as those used in the experiments. The size of the simulation box is $24\mathrm{\mu m}(x)\times 24\mathrm{\mu m}(y)$ with $3840\times 1920\mathrm{cells}$, and each cell was filled with 20 particles. The incident angle of laser was 30 deg. The targets are modeled by cold plasmas composed of ${\mathrm{C}}^{4+}$, protons, and electrons, where the ratio of ${\mathrm{C}}^{4+}$ to proton was 1:2. In the ideal case, the thickness of the unexpanded target was 50 nm, and the proton number density was 94.6$n_c$ with uniform distribution, where $n_c=1.75\times {10}^{21}{\mathrm{cm}}^{-3}$ was the critical density for 800 nm of laser wavelength. However, due to effect of the prepulse, we assume that the expanded target density distribution was a Gaussian function before the main pulse arrives. It can be expressed as $n_p(l)=n_0\exp [-4\ln (2)l^2/L^2]$, where $l$ is the distance from the center of the target, $L$ is the thickness at FWHM, and $n_0$ is the maximum density. The total number of protons in the target was kept as constant for different simulations, i.e., $\int n_p(l)\mathrm{d}l=50\mathrm{nm}\times 94.6n_c$. Therefore, the $L=0.1\mathrm{\mu m}$ ($n_0=44.6n_c$), 0.5 µm ($n_0=8.92n_c$), 1.5 µm ($n_0=2.97n_c$), 5 µm ($n_0=0.892n_c$) cases, and an ideal unexpanded case were simulated, respectively.

The energy spectrum from the PIC simulations is shown in Fig. 5. When the initial expanded size of the target was small, the proton cutoff energy was around 14 MeV for the cases of $L=0.1\mathrm{\mu m}$, 0.5 µm expansion targets, and the ideal unexpanded target. However, while the expanded size of the target increased, the cutoff energy of protons became smaller. For example, the $L=1.5\mathrm{\mu m}$ and 5 µm expansions corresponded to cutoff energies of 10 and 5.6 MeV, respectively. Such results were close to the proton energy spectra from the HC and LC cases [as shown in Fig. 4(i)] in the experiments. Because the target parameters in PIC simulation were estimated by the simplified model, the simulation results could not be strictly used as the quantitative basis for the experimental results. However, it could be explained that the proton energy was reduced due to the larger expansion of the target in experiments. We note that under the conditions of our experiments and simulations, the proton was accelerated by the target normal sheath acceleration (TNSA) mechanism^[13]. The appropriate plasma gradient was beneficial to laser absorption of electrons, which would enhance the effect of TNSA^[31]. Therefore, the proton cutoff energy of the 0.1 and 0.5 µm targets with plasma gradient was higher than that of the 50 nm target without plasma gradient in PIC simulations. However, the larger expansion of the target would induce the smaller density gradient at the rear surface, resulting in a weaker sheath electric field, thereby reducing the acceleration effect of the TNSA^[32]. Therefore, the LC laser greatly increases the expansion of the target, detrimental to the TNSA, leading to the lower cutoff energy, which is consistent with the experimental results in Fig. 4.

In conclusion, this paper reports the characterization of the laser contrast using the PM system in the SULF-1PW beamline. This PM system enhances the temporal contrasts on ps and ns time scales by about 2 orders of magnitude. The experiments with proton acceleration also verified that the use of this system can significantly improve the cutoff energy and quantity of proton beams. The simulation can qualitatively predict the expansion of the target with HC and LC lasers. The LC prepulse would greatly increase the expansion of the target, which is bad for proton accelerations. To solve this issue, the PM system is designed to effectively improve the laser contrast in our case. With the progress of PM technology, experiments have shown that PM reflectivity can reach 95%^[33], effectively improving the laser energy on target. It is also expected to effectively alleviate the problem of high energy loss in the cascade use of multiple PMs, providing a robust and efficient tool for research on particle acceleration, solid high-order harmonics, and attosecond radiation.