Commissioning of the 1 PW experimental area at ELI-NP using a short focal parabolic mirror for proton acceleration

High-power laser systems (HPLS) have opened new frontiers in scientific research^1–3 and have revolutionized various scientific fields, offering unprecedented capabilities for understanding fundamental physics, ranging, for example, from the exploration of extreme states of matter, to particle acceleration⁴ and to laboratory astrophysics.⁵ For instance, these systems facilitate studies that replicate conditions similar to those found in stellar cores, thereby advancing our understanding of nuclear fusion processes.^6,7 Furthermore, HPLS may be integral to the development of compact particle accelerators with significant applications in medical imaging and cancer therapies.⁸ 

Worldwide, established high-power laser facilities, such as the CLF at the Rutherford Appleton Laboratory in the United Kingdom, LULI in France, QST in Japan, GIST in South Korea, GSI and HZDR in Germany, and LLNL in the United States, have set benchmarks in laser-driven research.^5,9–16 These lasers have not only enabled us to deepen our understanding of the universe, but have also paved the way for new technologies and applications across various fields. The efforts pursued in the past two decades on improvements in laser parameters and configuration, on diagnostics, and on target development have also allowed optimization of laser–plasma interaction with respect to hybrid acceleration mechanisms.^{10,11,17–20}

Extreme Light Infrastructure–Nuclear Physics (ELI-NP) in Romania is a leading facility in this field, equipped with one of the world’s most advanced HPLS.^21,22 The ELI-NP HPLS includes two 1 PW beams dedicated to exploring laser–matter interactions at extreme intensities.^23,24 The ultimate goal of the ELI-NP facility is to achieve groundbreaking results in laser–plasma physics and nuclear physics,²⁴ leveraging the capabilities of its state-of-the-art laser system.

For the commissioning phase of the ELI-NP 1 PW experimental area and the two laser beams, utilizing a short focal parabolic mirror, we have focused our efforts on a rigorous optimization of laser-driven ion acceleration, setting the stage for future experiments. This optimization has entailed meticulous adjustments of the laser’s spatial and temporal pulse parameters, alongside advancements in targetry involving thin and ultrathin films. Efforts have been concentrated on exploiting and optimizing the target normal sheath acceleration (TNSA) mechanism, since this is widely recognized for its robustness and efficacy in generating high-energy ion beams and can be used as a standard owing to the vast literature available on the subject.^1–4 

This work provides a comprehensive report on the current status of the 1 PW outputs of the HPLS of ELI-NP and presents the findings of the experimental campaign run for the commissioning of the 1 PW area using a short focal parabolic mirror.

The remainder of the paper is structured as follows: Sec. II describes the 1 PW laser beam parameters; Sec. III gives an overview of the experimental area, the setup, and the diagnostics utilized during the experimental campaign; Sec. IV presents the experimental results, providing insights into the potential of the ELI-NP laser system for advanced scientific research; Sec. V discusses the correlation between the experimental results and simulations, and then draws conclusions on the performance of the HPLS. The detailed investigation and the results of the commissioning underscore the significant role of the ELI-NP facility in advancing high-power laser applications and contributing to the global scientific landscape.

The HPLS has been developed by the Thales Group for the ELI-NP facility at the National Institute for Physics and Nuclear Engineering (IFIN-HH) located in Bucharest-Măgurele, Romania.²² The high-power laser is a dual-arm system (arms A and B) capable of delivering two laser beams simultaneously and with three different power outputs: 0.1, 1, and 10 PW, respectively. The HPLS is installed in a clean room area of ∼2800 m² (see Fig. 1). During the commissioning campaign, the two 1 PW beams were tested successively. The maximum pulse energy of the 1 PW beam is 35 J (measured before the temporal compressor) and can be compressed down to 25 ± 2 fs FWHM and fired at 1 Hz repetition rate. The laser spectrum extends from 750 to 840 nm, with a central wavelength of 810 nm. The typical near-field, pulse duration, and spectrum of the 1 PW laser arm B are illustrated in Fig. 2. The laser output is operated normally at an energy of 31 ± 1 J per pulse before compression with a typical stability better than 1% rms over 1 h of continuous operation. Before each output, the HPLS benefits from a mechanical deformable mirror (from Imagine Optic, France) placed before the compressor and coupled with a wavefront sensor diagnostic placed after the compressor.²¹ 

The adaptive optics system is used to control the laser beam wavefront at the output of the 1 PW temporal compressor, giving a Strehl ratio better than 0.85. The laser beam at the output of the compressor is reflected off a leaky mirror that transmits a small fraction of the full aperture beam to a diagnostic bench. Part of the transmitted beam is picked up by a beam splitter and directed to a Wizzler (from Fastlite) for pulse duration characterization and phase retrieval. The phase retrieved from the Wizzler is then compensated by setting a compensation phase into a Dazzler (also from Fastlite) located in the stretching system of the HPLS. This loop allows a pulse duration of around 23 fs to be obtained, approaching the Fourier transform limit (FTL) [Fig. 2(c)]. The beam transmitted to the diagnostic bench is reduced in size to around 3 mm diameter via an optical system made of two telescopes composed of spherical and off-axis parabolic mirrors (OAP). This minimizes aberrations in the imaging diagnostics.

The laser energy, beam near-field, far-field, and spectrum were measured at the output of the temporal compressor on the HPLS diagnostic bench at each full-power shot. The pulse duration and temporal contrast were measured on demand, on the HPLS diagnostic bench.

The temporal contrast on a picosecond time scale ranging from −400 to 100 ps was measured using a third-order autocorrelator (Tundra, from UltraFast Innovations GmbH) with ∼10¹¹ dynamic range (see Fig. 3), The contrast measurement performed on 1 PW arm B showed a cluster of four pre-pulses in the range of −175 to −150 ps. These prepulses were generated by the nonlinear interaction of the main pulse with the post-pulses introduced by reflections between the Amp 1.2 Ti:sapphire²¹ crystal faces.²⁵ Another cluster of pre-pulses was also identified closer to the main pulse, in the range of −80 to −50 ps, originating from defects in the Amp 1.1 crystal.²¹ 

The temporal contrast measured on both arms A and B is shown in Fig. 4, where one can also observe multiple peaks in the arm A trace. Regarding that trace, the pulse at −9 ps is an artifact generated by a beam splitter placed before the Tundra device, and as such it is matched by a lower post-pulse. The pulse at +33 ps is a post-pulse generated by the sampling optics and creates a ghost pre-pulse as a result of the third-order cross-correlation measurement method that was employed. These artifacts are not present in the arm B trace, because the setup was modified to eliminate them before the contrast measurement of arm B.

Arm A of the HPLS is characterized by a temporal contrast with a profile similar to that of arm B up to ∼35 ps before the main pulse. Compared with an intensity contrast of ∼10⁻⁶ at 15 ps before the main pulse for arm B, the ramping pedestal of arm A shows a better intensity contrast of nearly 4 × 10⁻⁸. This can be explained by a slight difference in roughness between the two convex mirrors of the respective stretchers or by a slight difference in the divergence of the input beam within the stretcher. The latter leads to different focused beam sizes on the respective convex mirrors of the stretchers, generating higher or lower coupling of the high-frequency diffusion from the convex mirror in the main beam.^25–27 

During the early stage of the commissioning phase, we deployed a tunable delay probe beam to identify through shadowgraphy the real pre-pulses in the laser temporal profile and eliminate them. An example of a probing outcome is shown in Fig. 5, where the formation of a plasma plume can be seen soon after the pre-pulse interaction. By the end of the commissioning campaign, we identified and eliminated all the picosecond pre-pulses.

A Pockels cell and a Faraday isolator are placed in the first part of the amplifying chain, after the front-end, to remove amplified spontaneous emission or pre-pulses on the ns time scale as well as to provide protection from back-reflected (BR) pulses.

After compression, the fully amplified and collimated laser beam of 190 mm diameter is diverted by two dielectric flat mirrors and propagated nearly 30 m across the laser beam transport system (LBTS) before reaching the target area.

Commissioning of the 1 PW experimental area was carried out for both laser arms, A and B. The experience gained with the commissioning run allowed us to further enhance the temporal contrast of the HPLS to the green curve in Fig. 4, ∼10⁻¹⁰ at −20 ps (i.e., nearly two orders of magnitude better). The contrast upgrade of the HPLS will be described in a separate work and took into consideration two improvements. One consisted of the replacement of the convex mirror in the stretcher with an ultralow-roughness mirror, and the other, giving the results presented in Fig. 4, consisted of the replacement of the Offner-type stretcher with a reflective Martinez-type.

The ELI-NP 1 PW experimental area hosts three vacuum chambers, C1, C2, and C3, as shown in Fig. 6. Each of these chambers is designed to accommodate an experimental setup that can be flexibly adjusted. Each laser arm is routed to the main interaction chamber, C1, via the C2 or C3 chamber. Generally, the HPLS output arm A is routed from C2 to C1, while the output of arm B is routed from C3 to C1. The laser arm A enters the experimental area at a height of 1150 mm from the floor, while the arm B optical axis is at 1510 mm from the floor.

The commissioning of the short focal configuration in the E5 area was carried out by investigating proton acceleration using various metallic and diamond-like carbon (DLC) flat foil targets. Arm B was first commissioned in the C3 chamber, followed by arm A in C1. Figure 7 illustrates the setups used for the commissioning of arms A and B. The two setups were similar, except that for arm A, the laser was focused directly on target without using a folding plasma mirror between the parabolic mirror and the target. The optical tables in the interaction chambers were anchored to the floor and decoupled from the chamber body, stabilizing the setup from air to vacuum. The arm B laser beam entered the C3 chamber with its optical axis at about 870 mm above the chamber optical table. To reduce any potential vibration arising from the use of very long optical posts, several optical breadboards were used to create a segmented optical table located about 400 mm below the laser optical axis. In the C1 chamber, the optical table was designed to be at a distance of 400 mm from the laser optical axis, allowing for a quick exchange of the optical setups used in C3.

Inside the C3 interaction chamber an f/3.7 off-axis parabolic mirror (OAP) with an off-axis angle of 45° (manufactured by QED Optics⁴³) was used to focus the laser beam. After the focusing mirror, the focal spot and beam wavefront were imaged via a dedicated setup, also operating under vacuum. The beam wavefront was corrected using the HPLS deformable mirror to compensate for the residual wavefront errors accumulated through the beam transport and focusing OAP.

Following the optimization, a beam with a residual wavefront error of 75 ± 5 nm rms was obtained, resulting in a focal spot size of about 3.8 ± 0.2 μm at FWHM [see Figs. 8(a) and 8(b)], with an ∼50% encircled energy. The encircled energy was calculated from the results of multiple measurements over a high dynamic range of four orders of magnitude using calibrated stacks of neutral densiy (ND) filters. The images were processed (for background noise subtraction and scaling) and analyzed using a custom-developed reconstruction algorithm that combines multiple dynamic level measurements and estimates the encircled energy. A 3D profile sample of the reconstructed focal spot energy distribution is depicted in Fig. 8(c) on a log scale. During the focal spot correction and optimization, the HPLS was operated at nominal output energy and pulse compression to maintain stable thermal lensing of the amplifiers. Accordingly, to avoid damage to the focus imaging system, the beam energy was attenuated with reflective optics before the compressor. This was the typical method for delivering low-energy pulsed laser beams.

Calibration measurements were undertaken to estimate the actual energy delivered in the interaction chamber. The energy of a low-power pulsed beam was measured before the compressor and at the output of the LBTS. With account taken of other losses due to the routing mirrors, an antireflex (AR) pellicle beam splitter, the off-axis parabola, and an AR protection pellicle in front of it, the maximum energy achieved in the focused beam was about 21.5 J, excluding the plasma mirror.

Proton acceleration with the arm B laser was first investigated with the intrinsic laser temporal contrast by using a folding gold-coated sacrificial mirror between the OAP and the target. The sacrificial mirror slab, with dimensions of 50 × 200 mm², could accommodate ∼70 shots.

In the second phase, to improve the laser contrast, the sacrificial mirror was replaced with a single plasma mirror consisting of an AR-coated slab with a flatness of λ/10 rms and a reflectivity of about 0.3%–0.5% at 800 nm wavelength (see Fig. 7). For each shot, the position of the plasma mirror was pre-aligned daily, in vacuum, using the HPLS beam at low power to allow for the laser spot to be positioned always at the target chamber center (TCC).

For the laser–target interaction diagnosis, the following diagnostics were used: specular light at 1ω and 2ω for qualitative assessment of the laser contrast characteristic; stacks of radiochromic films (RCF) and CR-39 nuclear track detector placed on a two-axis motorized frame to diagnose the entire spatial and energy distribution of the protons; a Thomson parabola spectrometer (TPS) paired with imaging plates (IPs) to diagnose the ions emitted along target normal; and 1ω optical shadowgraphy to quantify the pre-plasma formation due to pre-pulses (see Fig. 7).

Far-field pointing, near-field beam profile, and pulse duration were measured on each shot, at full energy inside the 1 PW interaction area with the aim of monitoring the laser beam parameters and stability. The near-field profile was recorded by imaging the reflection of the full-aperture main beam from a thin beam sampler onto a quasi-Lambertian PTFE screen [see Figs. 2(b) and 7(b)]. For far-field pointing and pulse duration measurements (using a second-order cross-correlator from FemtoEasy), the main beam was sampled through a pick-off mirror. The pointing stability of the laser spot, evaluated over 100 sequential shots grabbed at 1 Hz, is illustrated in Fig. 8(d). For this measurement, the laser was fully amplified and attenuated with wedges before the pulse compressor. To confidently retrieve the pulse duration, the measurement was carried out in the experimental area using simultaneously two different devices, a FROG and a Wizzler. The measurements showed a similar pulse duration, with an accuracy of ±10%.

The p-polarized laser beam was incident on the target at 25° ± 1° for the arm B commissioning and at a larger angle of 45° ± 1° for arm A, to avoid direct coupling of potential back-reflected pulses into the laser transport system. A positioning system composed of a three-axis linear stage was used to align the target. Two types of target frames were used: stalks on an additional rotating motorized wheel and a raster plate (Fig. 9).

The stalks were designed with a C-shape frame, on which the target was mounted. This configuration provides the capability for optical probing of both rear and front target surfaces simultaneously, allowing the shadowgraphy to look very close to the target interaction point for investigating the pre-plasma. Afterward, to increase the shooting rate and facilitate the alignment procedure, the targets were glued on the raster plate with holes of about 1 mm in diameter and 6 mm apart. The optical diagnostic system used to image the laser focal spot at TCC was also employed to align the target by focusing the rear side. For this purpose, an LED of the same wavelength as the laser fundamental was used to illuminate the back side of each target from ∼400 mm away. The targets were individually pre-aligned in vacuum and the coordinates of each target were recorded.

For the thicker targets, commercial aluminum foils with thicknesses ranging from 0.8 to 10 μm were used, while for thinner targets, in-house fabricated diamond-like carbon (DLC) foils of 0.4 and 0.6 μm were used. As ultrathin foils, the ELI-NP target laboratory has manufactured freestanding gold targets with thicknesses from 0.4 μm down to 0.05 μm. These gold films were produced using a fishing method in which they were deposited by radio-frequency sputtering on a water-soluble substrate (sodium chloride) at room temperature in an ultrahigh-vacuum cluster deposition system.²⁸ X-ray reflectometry and cross-section scanning electron microscopy were used to determine, ex situ, the thickness of the films, while in situ monitoring was performed with a quartz sensor. The obtained films were subsequently transferred to the target frame after substrate dissolution in water. An optimized fishing procedure led to a flat surface of the free-standing film, with a roughness influenced only by the roughness of the substrate. The free-standing aluminum targets from commercial foils were glued directly to the target frame (see Fig. 9).

A screen was placed along the laser–target specular axis to record the profile of the laser beam reflected from the target. It allowed investigation of the target morphology at the time of the main laser pulse interaction, giving qualitative information on temporal contrast^29–31 (see Fig. 11 in Sec. IV). Two cameras were employed to image the scattered light from the screen at the 1ω and 2ω wavelengths. In addition, the laser energy back-reflected (BR) from the target was constantly monitored by a permanent diagnostic, using a photodiode with less than 50 ps rise time and a spectrometer. The setup was absolutely calibrated and had an energy threshold of the order of 1 mJ. This diagnostic helped to corroborate the data obtained with the specular reflection diagnostic.

The Thomson parabola spectrometer was designed to give good resolution of protons with energy up to 50 MeV and C⁶⁺ ions with energy up to 20 MeV/u. Spatial filtering of the beam was realized by using a pinhole (200–500 μm diameter). The magnetic deflector was a magnetic dipole made of a pair of permanent magnets with surface area of 50 × 50 mm² (L × H) mounted in a magnetic yoke with a gap of 5 mm and an average B-field of 0.45 T. The electrostatic deflector was made of copper plates with a surface area of 120 × 100 mm² (L × H) and a gap of 13 mm. Usually, an electric field of about 12 kV/cm was applied.

The detector consisted of a matrix of 16 imaging plates (IPs) (4 lines × 4 columns) mounted on an XY motorized linear stage. To decrease the cumulative background level on the IP detectors, an aluminum shield was mounted in front of the IP matrix with a square window the size of a single IP (60 × 60 mm²) to prevent or considerably reduce undesired exposure of the IP matrix to ionizing radiation. Such a TPS configuration with a 500 μm pinhole gives an energy resolution of about 10% at ion energies of 40 and 10 MeV/u for protons and C⁶⁺, respectively.

To measure the full angular and energy distribution of the proton beam, a stack made of RCF and CR-39 was used. Multiple stack configurations were designed with various numbers of layers to be able to measure a proton beam with a cutoff energy up to 60 MeV. For the maximum cutoff energy of 60 MeV, the stack consisted of 19 layers of RCF, 2 layers of CR-39, and filters of 25 × 25 mm² area. Multiple stacks were placed in a frame with a 25 × 25 mm² rectangular hole and were located at a distance of 25 mm from the target rear surface. A motorized XY linear stage was used to move across the ten RCF stacks after each shot. The two CR39 detectors corresponded to proton cutoff energies of ∼48 and 58 MeV. The CR39 detectors were placed at the high-energy side of the proton spectrum to measure the end tail and cutoff of the spectrum with high sensitivity. For some of the stacks, a millimeter-scale hole was drilled in the center of the stack and aligned with the TPS to measure the ion spectrum with high resolution while also getting information on the proton angular and energy distribution by the stack. Calibrations were performed for absolute particle yield of both the RCF and IP films using conventional linear accelerators. Additionally, the IP films were calibrated on shot against the CR39 detectors, and all results were checked with the literature for consistency.

A prototype system to monitor and study the electromagnetic pulses (EMPs) produced during laser–matter interaction was developed and implemented for the commissioning experiments. The aim was to assess the EMPs produced inside the vacuum chambers and verify the shielding effectiveness of the implemented EMP barrier for electronic equipment used in the experiment. Electric and magnetic fields were measured simultaneously using both commercially available derivative sensors and in-house custom-designed sensors. To accurately cover the large frequency range from 100 kHz to 6 GHz (the upper limit was imposed by the bandwidth of the oscilloscopes used), a diverse set of field sensors was employed. A mobile sensor array probing EMPs within the vacuum chambers and attached to a flange, as shown in Fig. 10, consisted of the following components:

• two large, low-frequency B-dot sensors (Montena SFM2G);

• one high-frequency B-dot sensor (Prodyn B-24);

• one medium-frequency D-dot sensor (Montena SFE3-5G);

• one high-frequency D-dot sensor (Montena SGE10G).

The system included sensors, baluns, attenuators, cable assembles, an electronics rack, a delay generator, and oscilloscopes. This could capture up to nine EMP field waveforms for monitoring and later offline analysis. The sensors were fitted in the vacuum chamber on a detachable vacuum flange to measure vertical and transverse electric and magnetic fields (see Fig. 10). They were located at 0.65 m above the optical table and ∼1 m from the target. The system could be set up to capture waveforms when triggered by signals coming from the laser at 1 Hz or on the user’s demand.

Typically, poor contrast can alter the target surface before the main pulse interaction. This may cause the laser beam to scatter heavily off the pre-expanded target plasma^30,31 [see Fig. 11(a)]. As the pre-pulses were identified and removed, a brighter and sharper specular reflected (SR) beam profile with features clearly resembling the laser near-field was measured, as shown in Fig. 11(b).

A BR energy exceeding 35% of the incoming laser energy was also recorded in the presence of the multiple pre-pulses (see Fig. 3) that generated the scattered specular reflection [Fig. 11(a)]. After removal of the pre-pulses, BR pulses with less than 1% of the main laser energy were routinely recorded. With the help of the plasma mirror even lower BR values were obtained. A thorough analysis of BR suppression using different plasma mirror configurations at 1 PW will be the subject of a separate study.

During the commissioning shots, for both laser arms, the maximum available HPLS energy was typically pursued, and the data reported here were taken over multiple days. The dependence of the cutoff energy of the proton beam, its spectrum, and spatial distribution with the target thickness was recorded using RCF stack films and TPS.

In the setup utilizing arm B, a sacrificial mirror, either gold- or AR-coated, was placed between the OAP and the target. The commercial aluminum foils and thinner DLC foils were used as targets for arm B commissioning. The energy loss by the main laser pulse (i.e., the reflectivity at full power) for both gold and AR plasma mirrors was characterized. The laser pulse was compressed down to a minimum of 26 ± 2 fs. The AR mirror had an absolute reflectivity at full power of about 75% and an extinction ratio at low power in the range of 200–300. The gold plasma mirror exhibited a 10%–15% higher reflectivity than the AR mirror. The gold mirror was used in order to maintain the intrinsic temporal contrast and almost fully preserve the energy of the laser beam (see the red trace in Fig. 4). Peak intensities of the order of 3 × 10²¹ W/cm² were obtained on target when a gold plasma mirror was used and routinely of around 2.6 × 10²¹ W/cm² in the configuration with an AR plasma mirror.

For the gold sacrificial mirror configuration, the dependence of the proton cutoff energy on the target thickness is shown in Fig. 12. A proton beam with maximum cutoff energies of 31 MeV was reached with aluminum targets of around 3 μm thicknesses. The vertical bars represent the range of cutoff values recorded for the same thickness by each diagnostic, while the points represent the average values.

Subsequently, by replacing the gold sacrificial mirror with an AR plasma mirror, the laser temporal contrast of arm B was improved.

In this case, the proton maximum cutoff energy shifted toward thinner targets. The maximum proton energy was boosted to 38 MeV using a DLC foil with a thickness of about 0.4 μm, as shown in Fig. 13. For the micrometer-thick targets, the CR39 detectors placed at the end of the RCF stack showed ∼10⁷ particles/sr for a cutoff energy of around 42 MeV. The typical spectrum of the accelerated proton bunch is shown in Fig. 14 and is discussed in detail at the end of this section.

Additionally, the second laser output of the HPLS, arm A, was characterized using the configuration shown in Fig. 7. The dependence of the proton cutoff energy on the target thickness is plotted in Fig. 15. As with arm B, commercial aluminum foils were used as thick targets, while gold foils manufactured at ELI-NP were used as thin targets. Laser pulses with 19 ± 1 J energy, compressed down to 35 ± 5 fs, were delivered directly on target, that is, without employing any sacrificial mirror. In this configuration, peak intensities around 2.5 × 10²¹ W/cm² were obtained on target.

The maximum proton energy of 35 MeV was obtained from 1.5 μm-thick aluminum targets. Compared with the results obtained with arm B when using a gold sacrificial mirror, there is a shift of the highest proton energy toward a thinner target thickness, despite arm A being shot at a slightly higher laser energy (namely, around 19 J instead of 16.6 J on target). This can reasonably be attributed to the better temporal contrast profile of the arm A beam. In fact, the arm A intrinsic laser temporal profile from −30 to −10 ps is nearly an order of magnitude lower than that of arm B. A more detailed analysis is given in Sec. IV E.

By further improving the laser temporal contrast (toward the latest one depicted in Fig. 4), it was possible to shift the maximum proton energy to thicknesses below 1 μm, consistently with the data reported in the literature.²⁰ 

The two diagnostics used for ion detection have some advantages and disadvantages. The TPS has a high sensitivity when a detector like the IP is used, and can discriminate among the spectra of different ion species. However, to achieve a high resolution, the TPS has to be used as a pinhole camera, therefore totally losing information on the spatial distribution of the ion beam. On the contrary, the RCF stack allows full resolution of the spatial and spectral energy distribution of the proton beam, although the other ions cannot be discriminated from the protons and the response around the energy cutoff is generally much lower than that obtained with the TPS (using CR39 detectors can help to improve the detection threshold, as shown above).

Two examples of proton spectra inferred by using the RCF stack and the TPS are presented in Figs. 14 and 16.

These figures show a clear difference in the reconstruction of the energy spectra. The TPS spectra were obtained from an aperture with a very small subtended angle, of the order of 10⁻⁶ sr, while the RCF stack collected the entire proton beam, which has an angular opening of up to 1 sr. Two hole diameters were used for the drilled RCF stacks. The spectra in Fig. 14 were obtained using a stack with a hole diameter twice that of the stack used for the spectra in Fig. 16. The latter stack was preferred in order to maximize the recorded surface area of the RCF films. However, this configuration demanded a more precise alignment. The differences in the reconstructed spectra arise on a shot basis from several reasons: the alignment of the RCF stack hole can differ from the line of sight of the TPS; the proton beam axis can be subject to pointing fluctuations; and the proton density distribution can exhibit large variations over the beam profile. In particular, the two spectra of Fig. 14 are similar up to their ends, where the spatial size of the dose becomes comparable to that of the RCF stack hole (see the insert of Fig. 14). In this case, the proton beam is very well centered with the RCF stack hole such that the signal toward the end of the spectrum is not captured, and consequently the reconstructed cutoffs differ by about 14% (see also Fig. 13).

In Fig. 16, the proton beam is very well centered with respect to RCF stack hole for low-energy protons, but it goes clearly off-center at higher energies. Furthermore, the proton beam size remains consistently large until the last RCF layer of the stack. Second, the proton density distribution exhibits strong nonuniformities that are clearly visible in the first layers: a low dose around the hole location and higher outside it, with a difference of up to an order of magnitude. All these features, together with the uncertainties in the precise positioning of the RCF stack hole with the TPS line of sight, generally produce large differences between the RCF and TPS spectra, as can be seen in Fig. 16.

The key to optimizing the characteristics of proton and ion beams is understanding and fine-tuning the laser–plasma interaction parameters.^11,16,17 The major laser parameters affecting the characteristics of laser-driven ion beams are definitely the energy, focal spot size, pulse duration, and temporal contrast. Among these, the temporal profile is of paramount importance to the interaction. Therefore, special attention was given finally to the investigation of the effects of the laser temporal pulse shape on particle acceleration. Following recent results that demonstrated the enhancement of the proton acceleration performance by the manipulation of the laser sub-ps temporal profile,¹⁷ we also investigated here the effects of the temporal profile with different laser contrasts (i.e., the nominal contrast and that improved by the use of a plasma mirror) and different target thicknesses, using arm B of the HPLS.

During the optimization process, the spectral phase group delay dispersion (GDD) and third-order dispersion (TOD) were set to achieve a near Fourier transform-limited (FTL) pulse duration (i.e., best compression). Starting from the best-compressed laser pulse of 25 ± 2 fs (i.e., given values for the GDD and TOD), the TOD term alone was manually varied using the acousto-optic programmable dispersive filter (Fastlite Dazzler) already installed in the HPLS. Laser pulse profile measurements showed that positive TOD increments generated an asymmetrical and longer pulse with a longer rising edge and a shifted post-pulse distribution. This pulse shape was characterized with the FROG in the experimental area and with the Wizzler on the HPLS diagnostic bench to remove the time-reversal ambiguity of the FROG.

The TOD was varied in steps of 10 000 fs³ in the range from ΔTOD = −20 000 fs³ to ΔTOD = 30 000 fs³, with ΔTOD = 0 representing the FTL pulse. The TOD scans were performed both with the plasma mirror and with the intrinsic laser temporal contrast, on aluminum targets of 0.8 and 3 μm thickness.

Figure 17(a) shows that the proton cutoff energy increases with increasing TOD for both cases, with intrinsic or improved temporal contrast. The pulse shape is depicted in Fig. 17(b) for the FTL pulse (ΔTOD = 0) and increased TOD (ΔTOD = 20 000 fs³). For the maximum TOD increase that was set, the laser pulse was still compressed, albeit over a longer duration. The pulse durations measured with the FROG were used to calculate the peak intensity on target.

In agreement with previous work,¹⁷ we also found that optimization of the TOD compared with an FTL pulse can lead to an enhancement of more than 30% in the proton cutoff energy. It is worth noting that a lower laser energy or a longer pulse duration, that is, a lower laser peak intensity, will trigger the plasma mirror differently. A plasma mirror calibration performed with our HPLS showed a reflectivity of about 75% for an FTL pulse at a laser peak intensity on plasma mirror of about 3 × 10¹⁵ W/cm² and a reflectivity of about 50% at an order of magnitude lower intensity, similar to previously reported data.^32–34 This proves the effect of the TOD to be even more relevant, since for the largest ΔTOD = 30 000 fs³, the pulse duration was around 50 fs FWHM, yielding a laser peak intensity on the plasma mirror in the region of 5 × 10¹⁴ W/cm², which corresponds to an intensity on target about three times lower than that for ΔTOD = 0. Despite the lower intensity on target, a ΔTOD of +30 000 fs³ boosted the proton cutoff energy by ∼30%.

To demonstrate that the increase in proton cutoff energy is generated by the TOD adjustment and not only by the pulse duration, we studied the proton acceleration when the laser intensity on target was varied by changing the laser energy, pulse duration, and focus position while keeping the ΔTOD = 0. The results are illustrated in Fig. 17(c), where the proton cutoff energy clearly drops when the laser intensity decreases for all situations. The experimental data obtained during the laser energy scan show that the proton cutoff energy scales as $∼Ipeak0.72$⁠, where I_peak is the laser intensity, which follows the typical scaling laws of TNSA.^9,35 For the focal spot scan, which also indicated a laser intensity drop by approximately an order of magnitude, the proton energy cutoff was reduced by nearly 65%. This rather small drop in proton energy can be explained by the larger laser spot on target, which will generate a higher population of hot electrons,³⁶ and also by the relative improvement in the temporal contrast, which can sustain the acceleration even at a lower laser intensity. A detailed analysis of this phenomenon is outside the scope of this paper and will be investigated separately.

In Fig. 18, we present typical results for the EMP fields best represented at the head of the sensors used, retrieved from the wave functions recorded on the oscilloscope in the form of voltage vs time signals. All EMP results correspond to a shot using ∼14 J laser energy on a 3 μm-thick aluminum target, taken during experiments with arm B.

The EMP signals were corrected for attenuation in the baluns, oscilloscope protection, and insertion losses in the long cables used. Figures 18(a) and 18(c) display EMP signals in the time domain at the sensor head for a low-frequency B-dot sensor and a medium-frequency D-dot sensor, respectively.

The B-dot sensor, oriented to measure the horizontal magnetic field, was connected to a 1 GHz bandwidth oscilloscope with a 6.25 GS/s sampling rate. The D-dot sensor, oriented to measure the vertical electric field, was connected to a 6 GHz bandwidth and 25 GS/s sampling rate. The temporal traces of the field amplitudes show that the EMP rings transient for about 400 ns. The measurements indicate short rise times, which correspond to high equivalent frequencies that can result in substantial electromagnetic induction.

Figures 18(b) and 18(d) show the frequency-domain representations of the EMP signals. The multiple frequency peaks measured within the vacuum chamber are attributed to resonances of both the chamber itself and the components inside it. The frequency scale in Fig. 18(b) is limited to 1 GHz by the bandwidth of the oscilloscope, which is lower than the 2 GHz bandwidth of the B-dot sensor. In the case of Fig. 18(d), the frequency scale is restricted by the D-dot sensor response, which decreased by 3 dB at 3.5 GHz, leading to a significant reduction in its sensitivity beyond this value. The peak-to-peak EMP in the time domain measured during this experimental campaign reached values in the tens of kV/m range. Specifically, for the case presented in Figs. 18(a) and 18(c), the peak-to-peak magnetic field recorded at the sensor head was 70 A/m, while the peak-to-peak electric field was 75 kV/m. These results are typical for EMPs produced by PW class lasers with fs pulse durations and in line with similar findings reported in the literature.³ The generated EMP did not require specific shielding methods for the electronics operating around the interaction chamber. Typically, for machine safety reasons, the in-vacuum motorized stages were decoupled from the controllers during the high-energy shots, while tests were also undertaken to assess the operation on-line. No conclusive correlation could be derived for the effects of the EMP on on-line electronics, but there are plans for more data to be recorded in the future. Parasitic measurements performed with differential passive detectors (aluminum filters and IPs) inside the vacuum chamber reveal, however, a large angular X-ray emission extending to energies of few tens of keV, which can pose high risks to the electronics. The EMP monitoring system is now a permanent setup for recording and characterizing the EMPs generated during the experiments in the E5 area.

To understand and explain the experimental results in relation to the ELI-NP 1 PW laser characteristics, among which the temporal profile plays the major role, simulations were performed using the 1D radiation-hydrodynamic code HELIOS.^36–39 We investigated the hydrodynamic expansion of targets irradiated by the different laser pedestals of arms A (intrinsic contrast) and B (intrinsic and AR plasma mirror improved contrast). The different target thicknesses that generated the highest proton energies were simulated. For each of the laser arms A and B, the measured temporal contrast (reported in Fig. 4) was used as input for the hydrodynamic code. The temporal contrast was simulated up to a few picoseconds (denoted by t_end in the following analysis) before the main pulse arrival on target. For the case with the plasma mirror, the improvement in temporal contrast was computed by evaluating the measured intrinsic temporal contrast, the measured low-field reflectivity of the plasma mirror, the peak intensity of the laser pulse on the plasma mirror, and the ionization threshold.^32–34,40 For the configuration utilizing the AR plasma mirror with an attenuation factor of 200, the temporal contrast was estimated to be ∼10⁻⁸ at 10 ps before the arrival of the main pulse.

Considering the experimental results obtained and presented in Figs. 12, 13, and 15, we simulated the expansion of 0.4 μm-thick DLC and aluminum, and 1.5 μm-thick aluminum targets. The simulated results for the three laser temporal contrasts interacting with 1.5 μm-thick aluminum targets are plotted in Fig. 19, in which the initial target profile is depicted by a light blue shaded area. Figure 19 also shows the plasma critical density n_c (black dashed line) and the average proton cutoff energy that was experimentally obtained for each laser contrast and individual target.

With the improved contrast achieved with the help of the plasma mirror or using arm A of the HPLS, which has a lower intrinsic pedestal, the back side of the target remains intact, because the shock does not have sufficient time to propagate (see the red and blue curves in Fig. 19).

The shock that reaches the back side of the target influences the ion acceleration, generating lower proton cutoff energies,⁴¹ as were also experimentally measured. This can be explained by the increase in the overcritical target thickness by up to six times its initial thickness. Further, the propagation of the shock and expansion of the rear side of the target also affects the layer of contaminants that is responsible for the proton acceleration.^38,39

The proton energies were consistently increased while a sharp density gradient was preserved on the backside of the target with the help of improved contrast (arm B with plasma mirror, Fig. 21), even when a target as thin as 0.4 μm was used. We can reasonably associate this increase with the improved contrast and shift of the optimum acceleration regime toward thinner targets.²⁰ 

In line with the previous data, for the lower temporal contrast of arm A and without a plasma mirror, the 0.4 μm target is strongly expanded, as can be seen in the simulation plotted in Fig. 21. This in turn generates a lower proton cutoff energy, as expected and as experimentally measured (see Fig. 15).

We underline that the simulation results here do not represent precisely the plasma density profiles of the experimental laser–target interactions, since the simulations were performed in a 1D geometry, which is limited in representing all the interaction conditions. Moreover, the domain of laser mid-intensities (close to the relativistic regime) coupling into the plasma was not simulated. This topic is currently the focus of much research effort,⁴² and in particular is being investigated by ELI-NP through multidimensional hydrodynamic and particle-in-cell simulations.

Nevertheless, the present approach of adopting a hydrodynamic evaluation of target conditions under different laser contrast irradiations has allowed us to infer more realistic plasma conditions and understand the limitations of the commissioning experiments. As we are continuously developing and using this approach for understanding and simulating laser–target interaction, it also represents a helpful tool for determining the optimum conditions for experiments by other users.

The ELI-NP 1 PW experimental area has been successfully commissioned, with both laser beams delivering the designed parameters. In this paper, we have reported on relevant issues typically encountered in laser–solid target experiments such as those that can be performed at ELI-NP. We started the commissioning by characterizing and improving the spatial and temporal profiles of the laser pulse and testing different diagnostics. Finally, we demonstrated that accelerated proton beams with cutoff energies of around 30 MeV can be generated from thick targets using the ELI-NP laser beams with intrinsic contrast. With a cleaner temporal contrast, with the help of a single plasma mirror, ∼38 MeV maximum energy was obtained from thin targets. Additional improvements in the intrinsic HPLS temporal contrast performed after the commissioning campaign allowed higher proton energies to be obtained, and this will be reported in forthcoming publications. The effect of the laser pulse rising edge was also investigated and an enhancement of the laser–plasma coupling was revealed, together with a boost in the accelerated proton energies, in good agreement with results reported at similar facilities.¹⁷ After commissioning, the 1 PW experimental area has been opened to user beam-time access, with many experiments already demonstrating successful delivery of particle and photon beams. All the different optical and targetry systems, and the diagnostics that were employed and successfully tested, are now available to users. In addition, continuous work is being done to improve the laser beam parameters, to develop novel targetry and targets that allow shooting at high repetition rate, and to implement new diagnostics for online detection of plasma and of particle and photon beams.

This work was supported by the Extreme Light Infrastructure–Nuclear Physics (ELI-NP) Phase II, a project co-financed by the Romanian Government and the European Union through the European Regional Development Fund, by the Romanian Ministry of Education and Research CNCS-UEFISCDI (Project No. PN-IIIP4-IDPCCF-2016-0164) and Nucleu Projects (Grant No. PN 23210105 and 19060105). The Romanian Government also supports ELI-NP through IOSIN funds as a Facility of National Interest.

The authors have no conflicts to disclose.

M. O. Cernaianu: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Resources (lead); Software (equal); Supervision (lead); Validation (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead). P. Ghenuche: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (equal); Methodology (equal); Resources (equal); Software (equal); Supervision (lead); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). F. Rotaru: Conceptualization (equal); Data curation (lead); Formal analysis (lead); Investigation (equal); Methodology (equal); Resources (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). L. Tudor: Conceptualization (equal); Investigation (supporting); Supervision (equal). O. Chalus: Investigation (supporting); Methodology (supporting); Writing – original draft (supporting). C. Gheorghiu: Investigation (supporting); Methodology (supporting); Resources (supporting); Writing – original draft (supporting). D. C. Popescu: Data curation (supporting); Formal analysis (supporting); Methodology (supporting); Resources (supporting); Software (equal). M. Gugiu: Data curation (equal); Formal analysis (equal); Investigation (equal); Resources (equal); Software (equal); Writing – original draft (equal). S. Balascuta: Data curation (equal); Formal analysis (equal); Investigation (equal); Resources (equal). A. Magureanu: Data curation (supporting); Formal analysis (supporting); Investigation (equal); Methodology (supporting). M. Tataru: Methodology (supporting); Resources (supporting). V. Horny: Data curation (supporting); Investigation (supporting); Methodology (supporting); Resources (equal); Software (equal). B. Corobean: Data curation (supporting); Investigation (supporting); Methodology (supporting); Resources (equal); Software (equal). I. Dancus: Data curation (supporting); Formal analysis (supporting); Investigation (equal); Methodology (equal); Resources (equal); Software (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (supporting). A. Alincutei: Investigation (supporting); Methodology (supporting). T. Asavei: Investigation (supporting). B. Diaconescu: Investigation (supporting); Resources (supporting). L. Dinca: Resources (supporting). D. B. Dreghici: Data curation (supporting); Investigation (supporting); Methodology (supporting); Resources (equal); Software (equal). D. G. Ghita: Investigation (supporting). C. Jalba: Resources (supporting). V. Leca: Resources (supporting); Writing – original draft (supporting). A. M. Lupu: Investigation (supporting). V. Nastasa: Formal analysis (supporting); Investigation (equal). F. Negoita: Conceptualization (supporting). M. Patrascoiu: Investigation (supporting). F. Schimbeschi: Investigation (supporting). D. Stutman: Conceptualization (supporting). C. Ticos: Formal analysis (supporting); Investigation (supporting); Methodology (supporting). D. Ursescu: Methodology (supporting). A. Arefiev: Resources (supporting); Software (supporting). P. Tomassini: Investigation (equal); Methodology (equal); Resources (equal); Software (equal). C. A. Ur: Conceptualization (supporting); Funding acquisition (lead); Project administration (lead); Resources (lead). V. Malka: Conceptualization (supporting); Data curation (supporting); Writing – review & editing (supporting). S. Gales: Methodology (supporting). K. A. Tanaka: Data curation (supporting); Investigation (supporting); Methodology (supporting); Supervision (supporting). D. Doria: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Resources (lead); Software (equal); Supervision (lead); Validation (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.