Thin films of hydrocarbon plasma polymers were deposited on BN substrates (1.5 mm thick, Fondazione Bruno Kessler FBK) using plasma-assisted vapor phase deposition of poly(ethylene) (PE, Sigma-Aldrich, M_n = 2 × 10⁴ g/mol). The generic scheme of the process is shown in Figure 1A. Before deposition, granules of PE were loaded into a crucible which was positioned above a graphite electrode. The deposition chamber was pumped to a base pressure of 10⁻⁴ Pa using a tandem of rotary and diffusion pumps, and then Ar was introduced at a flow rate of 10 sccm, setting the working pressure at 5.2 Pa. The crucible was heated to achieve a release of oligoethylenes, which was monitored by Quartz Crystal Microbalance (QCM). After a pre-heating stage, the temperature of the crucible was set at 300°C to maintain the constant deposition rate of oligoethylenes of 16 nm/min. In the case of experiments with plasma, an r.f. discharge (13.56 MHz) was initiated by delivering power from an r.f. power supply (R600 Kurt J. Lesker) to the graphite covered electrode through an automatic matching unit (R600 m KJL). Then substrates were introduced into the chamber using a loadlock and the depositions were performed.

The thickness of thin films was measured by spectroscopic ellipsometry (Woolam M-2000DI) on samples deposited on polished silicon substrates to avoid difficulties related with the surface roughness. The measurements were done at three angles ranging from 50° to 70° and in the wavelength range of 192–1690 nm. The ellipsometry data were fitted using Complete EASE software (J.A. Woollam) according to a Cauchy model for silicon substrate with 1.5 nm of native oxide. For the calculations of the film density, the samples of known thickness were deposited on Al foils with known surface area. The density of the films was measured gravimetrically by taking the Al foil weight difference before and after deposition (micro-scale Sartorius, CP225D-0CE, 0.01 mg) and dividing it by the film volume.

The surface topography of ppC:H thin films on BN was measured using Atomic Force Microscopy AFM and Scanning Electric Microscope equipped with Energy Dispersive X-ray detector SEM-EDX. The AFM measurements were performed in an intermittent contact mode (Ntegra, NT-MDT) using standard silicon cantilevers (Multi75-Al-G, BudgetSensors) in air. Images of 256 × 256 data points were acquired with different scan sizes. Second-order polynomials were subtracted from the raw data to remove the sample tilt. The SEM/EDX measurements (JSM-7200F SEM with JED-2300 EDX, Jeol) were carried out in a side-view orientation with a 10 kV acceleration voltage after the sample metallization with a 10 nm gold layer.

The chemical composition of ppC:H thin films was studied by Fourier-Transform Infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), Rutherford Back-scattering spectrometry (RBS) and Elastic recoil detection analysis (ERDA). The FTIR measurements (Bruker Equinox 55) were performed in a reflectance−absorbance mode on samples deposited on gold-mirror-covered silicon with a resolution of 4 cm^-1 and with 200 scans. The XPS spectra (XPS, Phoibos 150, Specs) were acquired using an Al Kα X-ray source (1486.6 eV, Specs). Wide and high-resolution spectra were acquired with a pass energy of 40 eV and 10 eV, respectively. The spectra were analyzed using CasaXPS software and charge-referenced to aliphatic carbon at 285.0 eV. RBS/ERDA was performed using a tandem accelerator, Tandetron MC 4130. For RBS, He⁺ ions were used at an energy of 3.07 MeV for the identification of O, whereas H⁺ ions were used at an energy of 1.74 MeV for the detection of C using non-Rutherford cross section for elastic scattering advantageously offering enhancement of light element sensitivity detection. The scattered beam ions were detected by an ORTEC ULTRA-series detector with an active area of 50 mm² and a thickness of the depletion layer of 300 μm. The detector was placed at a backscattering angle of 170° out of a plane (the Cornell geometry). In the ERDA method, He⁺ ions were used at an energy of 2.5 MeV for the depth profiling of H up to the analytical depth at about 500 nm. The detector was placed at a backscattering angle of 30° in a plane (the IBM geometry). All samples were measured at three different spots to avoid any damage during measurements and the final spectrum was provided as a sum of articular spectra from different beam spots. The RBS/ERDA recorded spectrum analysis was done using a SIMNRA code [38].

Calibrated CR-39 (polyallyl diglycol carbonate) is a solid-state nuclear track detector able to detect and distinguish alpha particles emitted from laser-driven pB fusion with respect to other particles from the plasma emission [39]. Energetic nuclei disrupt chemical bonds when passing through the polymer, leaving tracks of broken polymer structure. These tracks can be exposed by removing the damaged material in alkali solutions. Irradiation of CR-39 with known sources of energetic nuclei allows for a spectroscopic calibration, which can be subsequently used to discriminate the traces from alpha particles obtained in pB experiments.

Experiments here described were performed focusing the laser beam tightly to the target surface. In such a scheme, the pB reaction occurs in the plasma environment and consequently, the generated alpha particles are affected by the plasma conditions. This scheme is called “in-target” configuration. The mm-thick target allows the detection of ions only in the front side of the target (for instance the same hemisphere of the laser direction). Two different experimental setups were used as depicted in Figures 1B, C, respectively for the TARANIS laser at Queen’s University Belfast (United Kingdom) and for the PERLA B laser at HiLASE, Dolni Brezany (Czech Republic). The TARANIS laser pulse has a wavelength of 1.053 μm, temporal width of 800 fs, energy of 8 J/pulse and can reach intensities around 2 × 10 19   W / c m 2 on target. The laser pulse was focused onto the ppC:H/BN target at an incidence angle of 15° with respect to the target normal. The CR-39 detector was placed at angle of 85° with respect to the target normal and a distance of 50 cm. A large detection angle relative to the target normal was used to reduce the flux of other backwards accelerated ions (e.g., carbon, boron, nitrogen, etc.) on the CR-39 detector. Three consecutive shots were fired in the same target conditions to improve the statistics of traces on the CR-39 detector with a repetition rate of 1 shot per 20 min. The BN target used at the TARANIS laser system had a 1 μm ppC:H layer deposition on the front surface. The PERLA B laser pulse has a wavelength of 1.030 μm, temporal width of 1.6 ps, energy of 17 mJ/pulse and can reach intensities around 2 × 10 16   W / c m 2 on target. The incident angle was set to 45° with respect to the ppC:H/BN target normal and several hundreds of shots were made per each target at a repetition rate of 1 Hz. In this case the ppC:H layer was 680 nm in thickness. The CR-39 detector was positioned on the front of the target in the laser backward direction, covering large angle from −10°–58° as shown in Figure 1C. In both cases, diamond detectors working in a Time of Flight (TOF) geometry were installed to check the interaction shot-to-shot. Al foils of different thicknesses were placed as filters in front of the CR-39 detector sensitive surface to study the alpha particles energy distribution based on the cutoff energy determined by the Al foil thickness, and also to avoid saturation of the detector by undesired low energy plasma particles. The cutoff energy of the Al filters was calculated using LISE++ [40] and the results are shown in Table 1.

The CR-39 detectors were etched in 6.25 M NaOH at 70°C for 1 h (TARANIS) and 3 h (PERLA B). The calibration for alpha particles and protons was performed in the range of energies from 0.3 to 5.0 MeV, according to the described in [41]. Detailed analysis of the calculation of the total alpha particles yield can be found in [29].

The laser plasma interaction can be modeled using a combination of 2D radiative-hydrodynamic and particle-in-cell (PIC) simulations. The PERLA B laser system has a 1.5 picosecond prepulse arriving 14 ns before the main pulse with a contrast of ∼10⁻³. The prepulse interaction and subsequent plasma expansion was modelled using a 2D radiative-hydrodynamic simulation carried out with the FLASH code [42]. A particle-in-cell simulation based on the SMILEI PIC code was used to model the interaction with the main laser pulse thanks to the output of the FLASH simulation. A detailed description of the simulations and full results are presented in [33].

Similarly, the interaction between the TARANIS laser and the ppC:H/BN targets was modelled using a combination of the FLASH 2D radiative-hydrodynamic and WarpX ([43] PIC simulations codes respectively).

The TARANIS laser system has a ns pedestal with a contrast of ∼10⁻⁷. Therefore, the effect of the pedestal on the thin ppC:H deposition layer needs to be considered. The interaction between a 1.5 ns pedestal with a ppC:H layer at 15° incidence to the target normal was modelled using the FLASH code in 2D Cartesian geometry for three ppC:H densities of 600, 900 and 1,650 kg/m³. An initial ionization fraction of ∼4% was used to ensure that the target was not transparent to incident radiation in FLASH. The FLASH code uses adaptive mesh refinement to increase resolution in regions with large temperature or density gradients. In these simulations the minimum grid cell size was 100 nm × 100 nm.

The output of the FLASH simulation was used to set the initial conditions of the PIC simulation. Using the WarpX [43] PIC code, the interaction between the target and the main laser pulse (8 J, 800 fs FWHM, 6.5 μm FWHM), and incident on the target at 15° to target normal, was modeled. A simulation box of approximately 50 μm × 80 μm and cell size of 8 nm × 8 nm was used. The number of macroparticles in the cell was set such that the number of particles per cell (ppc) was comparable to the number density in units of critical density (n_c) for C, H, and electrons. To minimize the computational cost, a reduced number of ppc was used for the BN layer behind the ppC:H set to 25 ppc for B and 16 ppc for N. The electron ppc in this region was also capped to 81 ppc. WarpX implements a nuclear collisional algorithm outlined in [44] to enable the calculation of fusion reactions, including the pB reaction.

The concept of macroscopic kinetics of plasma polymerization considers the energy delivered per precursor molecule during its residence in the plasma [45]. At low specific energy, precursors undergo little fragmentation and integrate into the growing film, mostly preserving their chemical motif. At high specific energy, significant fragmentation occurs, and the film grows from random molecular fragments, losing the chemical identity of the precursor. Plasma-assisted evaporation of polymers obeys the same macroscopic concept, with the only difference that precursors are not low-molar-mass molecules, but larger oligomers. Vacuum thermal decomposition of poly(ethylene) releases a flux of oligoethylenes with a mean molar mass of 1,400 g/mol, which corresponds to 100 CH₂ units in the macromolecular chain [28]. An essential part of the fragmentation of hydrocarbons in plasma is the cleavage of C-H bonds with loss of atomic hydrogen, which leads to a deficit of hydrogen in the resultant coating. Since hydrogen is one of the only two reactants in pB fusion, special care should be taken to maximize its retention in plasma polymers.

Three types of ppC:H samples were prepared with 15 and 50 W plasma and without plasma, and their elemental content was analyzed by XPS and RBS/ERDA (Table 2). Although XPS is a surface-sensitive technique (the analysis depth is up to 10 nm) and RBS/ERDA probes the bulk of the film (the analysis depth is micrometers), both methods provide remarkably similar results, confirming that only hydrocarbons are present in the films. Simultaneously, XPS is not capable of detecting H, which is done by ERDA. No contamination from other elements is detected except for oxygen, which becomes incorporated into the films deposited with the plasma assistance because of post-deposition oxidation. The aging of plasma polymers is well documented; it is attributed to reactions of atmospheric oxygen with unquenched carbon radicals present in such films, with the higher plasma power leading to more drastic oxidation [46–48]. It is worth noting that in our case the concentration of oxygen does not exceed 3 at%, even for the highest power of 50 W used, which means that the depositions are performed in a power-deficient regime, with good retention of the chemical pattern of the precursor. Indeed, the RBS/ERDA data indicate that the H/C ratio is close to that of PE (H/C = 2.0) for all types of films. The H/C ratio decreases with plasma power, pointing to a cleavage of C-H bonds that occurs in oligomers due to electron impact in the gas phase and in the growing film due to ion bombardment, and hence to a partial loss of hydrogen. As an illustrative example, we present an RBS spectrum of the 15 W ppC:H sample showing high signal of C in non-Rutherford scattering which is measured using H⁺ ions at 1.7 MeV resonant energy in Figure 2A. We see homogeneous C distribution accompanied with H distribution measured by ERDA in Figure 2.b. However, its absolute concentration is still more than 60 at%, even in the 50 W film, which also supports the idea that these films belong to a class of soft plasma polymers as opposed to hard hydrogenated carbon coatings [49]. The ppC:H can be customized and adapted on the specific laser-driven pB experiments by simply changing the power delivered to the plasma.

In more detail, the chemical composition of the ppC:H films was analyzed by high-resolution XPS, NMR, and FTIR spectroscopy. The C 1s XPS spectra (Figure 3A) are symmetric and perfectly overlap for all three films, confirming that the plasma assistance followed by the subtle oxidation does not influence the aliphatic character of the plasma polymers. Our previous NMR studies (Figure 3B) showed that the ¹³C CP/MAS spectra are dominated by the CH₂ groups with a sharp resonance at 32.6 ppm accompanied by a weaker shoulder at 30.3 ppm [50]. The dominant peak and the shoulder belong to the crystalline and amorphous phases of oligoethylenes [51]. The NMR detection of the crystalline order is also in line with our earlier XRD studies that found a high degree of crystallinity and specified an orthorhombic structure in such films with lattice parameters matching those of poly(ethylene) [52]. The contribution from the amorphous phase is enhanced in the plasma polymerized film, which reflects the deterioration of the short- and long-range order. Nevertheless, the amorphization is not complete in contrast to an entirely amorphous character of typical plasma polymers.

The NMR spectra are also characterized by the presence of a minor signal at 14.6 ppm from the CH₃ groups, which serve as end- or side-groups in branched poly(ethylenes). The resolution of the NMR spectra does not allow us to distinguish whether the contribution from the CH₃ groups changes with plasma power; however, complementary FTIR measurements provide such information (Figure 3C). Expectedly, the dominant FTIR peaks at 2,905 cm⁻¹ and 2,847 cm⁻¹ belong to the asymmetric and symmetric stretching vibrations of the CH₂ groups, whereas weaker stretching and deformation vibrations of the CH₃ groups are manifested at 2,957 cm⁻¹ and 1464 cm⁻¹ [53]. With increasing plasma power, both types of vibrations are enhanced, with stretching vibrations submerging into those of the CH₂ groups. Furthermore, a broad band develops in the midspectral range, which is contributed by the stretching vibrations of the C=O (1,715 cm⁻¹) and C=C (1,655 cm⁻¹) functionalities. The appearance of these unprotonated moieties agrees with the observed oxidation and deprotonation of the ppC:H films, while the enhancement of the CH₃ groups points to an enhanced macromolecular branching and to a decrease in order.

Interestingly, subtle changes in the chemical landscape of the ppC:H films contrast with a substantial deviation in the film density (Table 2). The film deposited without plasma shows a density of 620 kg/m³, which is significantly lower than ∼900 kg/m³ of conventional poly(ethylene). Vapor-deposited oligoethylenes are known to undergo self-organization phenomena at a solid interface, forming multi-terrace islands with a large portion of unoccupied space in between [52]. Although the porosity of thick coatings has not been studied, it can be hypothesized that they also contain a significant number of voids that decrease the net density of the material. The use of plasma during deposition brings a competing contribution of the active species from the gas phase. Radicals participating in the recombination reactions and ions bombarding the growing film intervene with the island growth by inducing cross-linking and densification of the film. Thus, the films prepared with plasma of 15 W and, to a larger extent, with plasma of 50 W are characterized by a higher cross-link density, shorter molecular chains between junctions, and a higher net density of the coatings, which reaches 1650 kg/m³ for the 50 W ppC:H.

For the preparation of targets for pB fusion, the same ppC:H films were deposited using BN as substrates, and their morphology was investigated. Figure 4A shows the AFM height images and compares the 15 W ppC:H 800 nm-thick film deposited on BN and on polished Si. The surface of the ppC:H/BN sample is very rough, with the surface profile fluctuating on a micrometer scale, as can be seen from the z-axis of the image. The same film deposited on Si is much smoother and its surface profile fluctuates in the tens of nanometers scale. Thus, the intrinsic morphology of the ppC:H film should not significantly affect the overall morphology of the targets, which is predominantly given by the surface morphology of BN substrates.

This finding is further supported by the dependence of the RMS roughness of the ppC:H/BN samples measured for different AFM scan sizes (Figure 4B). No significant difference is found between the films deposited at different powers, evidencing that the roughness is given by BN. For the sake of argument, it can be observed that the dependence is linear in log-log coordinates, giving a notable value of the roughness exponent α = 1.0, which points to a self-affine character of the BN surface [54].

The structure of the ppC:H/BN samples was also studied by SEM/EDX. A cross-sectional SEM image of the 15 W ppC:H sample reveals a highly porous structure of BN, which is a result of the hot-press sintering of BN powder used for its production (Figure 4C, other samples are not shown as they provide similar results). The high roughness of the surface does not allow us to distinguish the ppC:H layer and investigate its possible penetration into the pores. However, the EDX analysis helps to solve this issue. Figure 4D shows the EDX images of a cross-sectional area coded in different colors for individual elements of boron, nitrogen, and carbon, and the overlay of all the colors. The black voids in the images correspond to the out-of-focus areas. Boron and nitrogen are found to overlap homogeneously in the same regions, confirming that the material is boron nitride. The carbon signal is strongly enhanced in the surface area, and its thickness corresponds to a 1 μm thickness of the ppC:H film. No noticeable penetration of the plasma polymer into deeper regions can be seen, which confirms that ppC:H grows on the surface, replicating the morphology of the underlying BN substrate.

As pointed out in Section 2.3, three types of ppC:H/BN samples were used in the experiment at TARANIS and PERLA B lasers. Figure 5A shows an example of optical microscope image taken on the CR-39 detector, which was irradiated by charged particles generated during the interaction of the 15 W ppC:H target with the TARANIS laser. After the irradiation, the CR-39 were etched in a solution of NaOH at 70°C as described in [55, 56]. The exposed tracks match the size of the tracks in the calibration of the alpha particles pointing to an efficient generation of pB fusion events triggered by the TARANIS laser. The same effect was achieved using the PERLA B laser for which the results have already been presented in [33].

The alpha particles flux recorded during the TARANIS campaign was 7 × 10⁷, 1 × 10⁸ and 1 × 10⁷ α/sr/shot using 0, 15 and 50 W ppC:H, respectively. However, during the PERLA B campaign values of 8 × 10², 6 × 10³ and 5 × 10² α/sr/shot were measured using the same types of targets. Expectedly, the alpha particles flux was found to be many orders of magnitude lower for the 15-mJ PERLA B laser as compared to the 8-J TARANIS laser (Figure 5B). However, the difference becomes less drastic if the alpha particles flux is normalized by the laser energy (Figure 5C). For example, the 15 W ppC:H/BN sample gives 40 times difference in pB fusion efficiency between the two laser systems. This difference can be tentatively explained by considering the cross section of the pB fusion. Indeed, the PERLA B laser triggers pB fusion through the first resonant peak of 165 keV [57] with a moderate number of protons (the proton cutoff energy is only slightly above this value), whereas the TARANIS laser produces a high proton flux around the main resonant peak at 670 keV (proton cutoff energies are well above this value). Considerations about the efficiency of the experiments with the two different lasers will be discussed in the simulation section. The normalized alpha particles flux for 15 W is 10⁷ α/sr/J, i.e., in line with the energy conversion values obtained in the recent years as summarized in [22].

A somewhat surprising relationship can be found considering the dependence of the alpha particle yield on the properties of the ppC:H thin films. While the expected result would be a monotonous dependence on the plasma power, our study shows that the 15 W ppC:H sample is superior to the other two samples, and the trend is strikingly similar for both laser systems used. Our characterization of the targets reveals that the morphology and roughness of the surface are governed by the structure of BN, which is similar for all three samples. The hydrogen content decreases with power, but gradually and to a small extent, so it can hardly explain the observed kink in the 15 W ppC:H sample. The film density was found to be the only parameter that changes drastically with power, although its dependence is also monotonous.

Using the FLASH code, the interaction between the 1.5 ns pedestal, with a ns contrast of 10⁻⁷, of the TARANIS laser with the target surface was modelled. The target was composed of a 1 μm ppC:H film, on top of a 5 μm BN layer. Note that both the BN substrate and ppC:H are modeled as homogeneous slabs without any voids within the simulation region. This is rationalized by the surface roughness measurements presented in Figures 4A, B showing that the variation of deposition thickness is on the order of tens of nm, which will be quickly homogenized by the pedestal. The same batch of BN was used for all depositions, therefore, the surface variation is considerable the same for all targets. A simulation was performed for each ppC:H density (600 kg/m³, 900 kg/m³ and 1,650 kg/m³). The density of CH₂ along the target normal for each ppC:H deposition is shown in Figure 6A. The density profiles show that there has been some ablation and compression of the target surface by the pedestal. However, for all three deposition densities, despite the relatively low ns contrast of the TARANIS laser, following a 1.5 ns pedestal, there remains a dense region of CH₂ on the target surface with which the main pulse will interact. There appears to be minimal variation between the simulations for the three different targets in the underdense preplasma extending beyond 5 μm in front of the target surface. The output of the FLASH simulations was used as initial conditions for particle-in-cell simulations to model the interaction with the main laser pulse using the WarpX PIC code. The boron and nitrogen ions were given an initial ionization level of 3⁺ and 5⁺, respectively. The hydrogen and carbon ions were fully ionized during the hydrodynamic simulation.

Figure 6B shows the energy distribution of protons that are within the high boron density region (i.e., > 1 μm into the target) 1.7 ps after the main pulse. The proton distribution has a cutoff of above 2 MeV. Therefore, using the TARANIS laser, it is possible to accelerate protons into the BN target with energies in excess of the main cross-section resonance (1.4 b [57] at 670 keV in order to drive fusion reactions in this regime. The pair-wise fusion model was enabled in the WarpX simulation, allowing the generation of alpha particles through the pB fusion reaction to occur within the PIC simulation. Figure 6C displays the normalized density distribution of alpha particles plotted over the number density of the other ions, 600 fs after the peak laser intensity. This plot shows that most of the alpha particle generation occurs within the first 10 μm of the target surface, driven primarily by protons accelerated forward into the BN target. Considering the proximity of alpha particle generation to the target surface and the reduction of electronic stopping in an ionized plasma, which is discussed in [23], many of the alpha particles generated within the BN target will escape the target and reach the CR-39 detectors. It has not yet been possible to understand the trend in alpha particle yield, the deposition density and morphology that is observed experimentally.

In comparison, the simulations performed in [33], modelling the interaction between the PERLA B laser and the 680 nm ppC:H deposition on a BN target suggest that the energy cutoff of the protons accelerated forward is only 30 keV, whereas the backward accelerate proton cutoff extends to 300 keV. Therefore, it is suggested that when using the PERLA B laser, pB fusion occurs in the preplasma driven by protons accelerated to energies above the lower, secondary resonance peak of 0.1 b at 165 keV. The simulations performed to model the interaction of the TARANIS and PERLA B lasers and ppC:H-BN targets show a clear variation in the acceleration mechanism responsible for driving the pB fusion reactions. For future implementations of ppC:H-BN targets it will be important to consider the density, morphology, and thickness of the deposition to optimize the acceleration processes and the resulting fusion yield.

As discussed in the experimental results, the alpha particle yield normalized by laser energy is reduced when the PERLA B laser is used compared to the TARANIS laser. Considering the PIC simulations, the TARANIS laser can accelerate protons to energies above the main cross section resonance (670 keV), where the cross section is both over an order of magnitude higher, and covers a significantly broader energy range, than the secondary cross-section resonance. Secondly, in the PERLA B setup [33], the density of boron ions in the preplasma, obtained in simulations, was only 5 × 10 19 cm^-3 [33]. However, the simulations performed to benchmark the TARANIS experimental results show that fusion occurs in the solid BN target, where the boron ion density is much higher at 5 × 10 22 cm^-3. This difference also affects the alpha particle yield, which depends not only on the density of the boron ions but also on the effective proton range through the target.