Established high-power Ti:sapphire laser systems^[1^–9^] are able to deliver laser pulses up to several PW at a high repetition rate of 0.05–1 Hz. Focusing them to relativistic intensities allows one to create laser-driven secondary sources in a wide range from ionizing radiation^[10^–13^] to extreme ultraviolet (XUV) and THz pulses^[14^–17^]. Solid-density metal targets are being used to create ion sources^[18^,19^], flashes of highly energetic X-rays^[20^,21^] and XUV light sources^[22^]. A high-repetition-rate operation is important for many applications in medicine and fusion science^[23^,24^], but poses a challenge for system integrity.

Debris management is an important aspect of ultrahigh-intensity laser–solid interaction at a high repetition rate. The amount of ejected mass is of the order of hundreds of μg per laser shot^[25^,26^] and the deposition of the ablated material is observed to deteriorate beamline components^[22^,25^,27^]. Available detailed characterizations of debris have been limited to non-relativistic laser intensities just above the ionization threshold^[22^,28^,29^] and high-energy long-pulse lasers^[30^]. First characterization attempts for relativistic high-power laser interactions show a timeline of small debris particles ejected earlier, with a fast ejection speed, and successively larger projectiles with lower velocity^[27^]. These studies further indicate an asymmetry of ejection early on, with more debris being ejected away from the side on which the laser interaction takes place, but lack a characterization of the spatially resolved debris deposition.

This paper presents a characterization of ejected debris with spatial resolution, for the first time, that will allow an evaluation of mitigation strategies to avoid damage and deterioration of beamline components, diagnostics and metrology devices.

The paper is structured as follows: (i) after a brief introduction of the novel methodology that is used to derive spatially resolved measurements from flatbed scans in Section 2, (ii) we present results from an experimental campaign at a high-power laser in Section 3 that shows two distinct types of debris, and (iii) close with the discussion and conclusion in Sections 4 and 5, respectively, evaluating the amount of ejected debris, relating results to an available model.

Experiments for this work are conducted at the Extreme Light Infrastructure Nuclear Physics (ELI-NP) facility^[31^] with a high-power 1 PW Ti:sapphire laser delivering on target ${E}_{\mathrm{L}}\approx 22\;\mathrm{J}$ within a pulse duration of ${\tau}_{\mathrm{L}}\approx 30\;\mathrm{fs}$ (giving a total power on target of $\approx 0.7\;\mathrm{PW}$ ). The energy is extrapolated from calibrations recorded at low energy and the pulse duration is measured on-shot with a frequency-resolved optical gating (FROG) system that diagnoses a picked-up reflection from a small elliptical mirror positioned before the focusing parabola. The laser pulse is focused with an incidence angle of 45° onto $\left(50\pm 5\right)\;\mu \mathrm{m}$ thick nickel disk targets, with a focal spot diameter of ${d}_{\mathrm{L}}\approx 4\;\mu \mathrm{m}$ full width at half-maximum (FWHM). The focal spot at high energy is estimated to be the same as for low-energy measurements^[32^], even if the Strehl ratio might be different^[33^,34^]. The setup is shown in Figure 1, with the focusing parabola (off-axis parabola (OAP)) in the back and targets mounted on a wheel.

Two 1 mm thick and $50\;\mathrm{mm}\times 50\;\mathrm{mm}$ squared sputter plates produced from fused silica (FS) are used to catch debris that is emitted away from the respective target front and rear sides. The plates were originally meant for another scope, solely to protect the optics of a probe beam setup (not further discussed hereinafter). The front-sided sputter plate is placed in front of a polarizer facing the laser-interaction side, while the rear-sided sputter plate is placed in front of an imaging lens. Note the auxiliary character of this arrangement of catchers, as the OAP is by default protected with a thin pellicle. The plates’ centres are not perfectly collinear with the laser-interaction point, but shifted by 3 mm down with respect to the target, and their surfaces are parallel to the target surface. The distance of the rear plate to the interaction point is $\left(125\pm 5\right)\;\mathrm{mm}$ , while the front plate is positioned at $\left(95\pm 10\right)\;\mathrm{mm}$ .

After the experiment, the sputter plates are scanned with an EPSON V-750-PRO flatbed scanner to obtain the spatially resolved deposited debris thickness ${z}_{\mathrm{Ni}}$ as a function of the transmittance $T={I}_{\mathrm{t}}/{I}_0$ . Here, ${I}_0$ is the intensity of the incident wave and ${I}_{\mathrm{t}}$ the intensity at the exit of the double-layer system. The details of the scanning procedure are outlined in Appendix A and the calculation of the transmission of a flat double-layer system is revisited in Appendix B. The theoretically predicted transmittance of evaporated nickel deposit is shown in Figure 2 as a function of the layer thickness for three channels of a colour scan. One notes the good agreement between the different colour channels, which points to a flat spectral response.

Raw scans of debris collected on the sputter plates are shown in Figure 3, with both plates having recessed areas that were protected from debris by mounting structures. Different interference patterns can be observed in Figures 3(a) and 3(b), which stem from the slightly different thicknesses of the silica plate. However, their influence on the scanned intensity is smaller than the error bars of the measurements performed. Debris originates from three laser shots, two shots on targets with diameter ${d}_{\mathrm{t}}=0.5\;\mathrm{mm}$ and one shot on a disk of 2 mm diameter. The sputter plates are coated by a surfacic deposition of debris, a weak but distinct areal deposition that uniformizes towards the edges of the plates. The uniformity of the deposition can be observed in contrast to the protected area by the sputter plate’s holders (white area in Figure 3).

In addition, three distinct marks are observed towards the target normal (highlighted with elliptical dashed lines). Slight target misalignment of less than 5° might be responsible for the spatial separation of the marks. This hypothesis is supported by the diametrical opposition of structurally similar marks with respect to the target position. Further, one notes that two small marks (highlighted with red dashed lines) contrast one large mark (highlighted with blue dashed lines) and it is reasonable to assume that small marks correspond to shots on small disk targets. A detailed characterization of the marks is done using white-light interferometry, profilometry and a scanning electron microscope (SEM), as shown in Figure 4 and available as dataset^[35^]. The surface characteristics change abruptly from uniform deposition outside the marks to a complex ablation–redeposition pattern, likely to be mechanical damage. Ablation craters reach depths of several tens of μm and diameters of hundreds of μm. The average depth is $10\;\mu \mathrm{m}$ (as measured with the profilometer) and the ablated volume within the marks amounts to $\approx 0.0231\;{\mathrm{mm}}^3$ atop the target front side and $\approx 0.0095\;{\mathrm{mm}}^3$ atop the target rear side (as deduced from white-light interferometry). With a glass density of $5\;\mathrm{g}\;{\mathrm{cm}}^{-3}$ , the mass of ablated material from the FS plates is $\approx 115\;\mu \mathrm{g}$ atop the target front side and $\approx 47\;\mu \mathrm{g}$ atop the target rear side. An additional element analysis via energy-dispersive X-ray spectroscopy (EDS) in an SEM reveals the atomic composition of the surfacing layer of the sample: $\left(36.1\pm 0.4\right)\%$ Ni deposition and $\left(26.41\pm 0.5\right)\%$ of Si and O that originate from SiO₂ glass. Further fractions are from contaminations and impurities amounting to $\left(6.6\pm 0.6\right)\%$ of O, $\left(2.2\pm 0.2\right)\%$ of Cl and $\left(2.3\pm 0.2\right)\%$ of Al. The surfacic deposition of glass indicates the redeposition of ablated material.

The transition from regions that show no ablation to regions that are heavily ablated is detailed with colour microscopy images in Appendix C.

Concerning the surfacic deposition, we measured the transmittance through the centre of the rear-side sputter plate (marked with an X point in Figure 3(a)) with spectral resolution using a compact Czerny–Turner spectrometer, as shown in Figure 5. The integrated surface element is $1\;{\mathrm{mm}}^2$ . The measurement shows a flat spectral response in accordance with the theoretical prediction presented in Section 2 for the transition metal nickel deposited on an FS plate.

The transmittance of the debris in Figure 3(a) is shown in Figure 6 in a squared region of interest (ROI) selecting a region of uniform deposition (green squares in Figure 3) far from the central marks. For conversion from scan intensity to transmittance we follow Equation (A3) from Appendix A, which reads as follows:

$$\begin{array}{r}{T}_{mn}={10}^{(\ln \left[I_{mn}\right]/B)-C}.\end{array}$$  (1)

The raw data are analysed separately for the distinct RGB channels, revealing no opaque zones, which allows for quantitative analysis of the full surface. For all three channels the transmittance shows a similar behaviour, as can be seen in Table 1. The surfacic deposition on the rear side has a uniform mean transmittance of $\left(84.4\pm 2.9\right)\%$ in the ROI. Towards the front side, the mean transmittance amounts to $\left(92.1\pm 1.8\right)\%$ in the ROI.

Considering that the debris on the sputter plates consists only of deposited nickel, we calculate the thickness of the deposited debris using Equation (B1) from Appendix B, which reads as follows:

$$\begin{array}{r}T(z_{\mathrm{M}},z_{\mathrm{S}})={\int }_{{\omega }_{-}}^{{\omega }_{+}}\parallel \mathfrak{T}(z_{\mathrm{M}},z_{\mathrm{S}},\omega ){\parallel }^2\cdot s\left(\omega \right)\mathrm{d}\omega .\end{array}$$  (2)

The results are shown in Figure 7, and the characteristic values are given in Table 2. The surfacic deposition on the rear-side ROI has a uniform mean thickness of $\left(0.6\pm 0.1\right)\;\mathrm{nm}$ . Towards the front side, the mean thickness in the ROI amounts to $\left(0.3\pm 0.1\right)\;\mathrm{nm}$ .

The mass can be calculated as ${z}_{\mathrm{Ni}}\cdot {p}^2\cdot \rho$ with the pixel size $p=10.6\;\mu \mathrm{m}$ and assuming solid density $\rho =8.9\;\mathrm{g}\;{\mathrm{cm}}^{-3}$ . The total mass of nickel deposited on both plates, within the ROIs from Figure 6, amounts to $3.3\;\mu \mathrm{g}$ , $\left(2.6\pm 0.8\right)\;\mu \mathrm{g}$ towards the rear side and $\left(1.3\pm 0.4\right)\;\mu \mathrm{g}$ towards the front side. In terms of corresponding debris emission, the average production is $\left(24\pm 5\right)\;\mu \mathrm{g}\;{\mathrm{sr}}^{-1}$ towards the front side and $\left(83\pm 15\right)\;\mu \mathrm{g}\;{\mathrm{sr}}^{-1}$ towards the rear side. Note the non-linear relationship between measured transmittance and derived debris thickness. The emission detected within the surfacic deposition is slightly asymmetric, with larger ejection towards the rear side of the target.

Available modelling^[26^] suggests that shots on small disk targets emit more debris than shots on large disk targets. The prediction for small disks is the total emission of $\left(257\pm 50\right)\;\mu \mathrm{g}$ , while it is $\left(99\pm 20\right)\;\mu \mathrm{g}$ for large disks. This difference between large disks and small disks can be explained by a larger fraction of the laser-heated electrons being held back by stronger fields in the case of smaller targets. With a larger refluxing cloud of near-relativistic electrons there are more electrons available to transfer heat to the bulk material. The model applies to cases where the evaporating mass is not limited by the available target mass (here $>429\;\mu \mathrm{m}$ diameter disks), and where the target sizes are smaller than the maximum expansion of the laser-generated target potential during electron discharge (here $<12\;\mathrm{mm}$ disks).

Experimentally, the total mass of the debris can be extrapolated from the measured mean surfacic deposition in the ROIs (towards the rear and front sides) of $54\;\mu \mathrm{g}\;{\mathrm{sr}}^{-1}$ assuming a spherically uniform emission. The extrapolated result is $\left(672\pm 127\right)\;\mu \mathrm{g}$ and compares well with the modelled total value of $\left(613\pm 83\right)\;\mu \mathrm{g}$ (obtained from the sum of the contribution of two small disks and one large disk) within the margins of uncertainty. The observed small asymmetry of the spherical emission (with more deposition towards the target rear side) might be owed to the asymmetry of the charge distribution in the environment of the laser interaction, with more electrons deposited in the laser forward direction. The asymmetry might be also related to the target thickness, which motivates future parametric studies.

The damaged areas on the sputter plates might be produced by the impact of high-velocity debris particles or dense flares of debris. The constraint area of damage reveals this population to be rather directional towards both target normal directions. There is visibly slightly more damage on the sputter plate facing the target front side than that facing towards the target rear side. The amount of ablated glass on the front side is $245\%$ larger than that on the rear side. This might point to the observation of a larger quantity of directional debris towards the target normal on the side of the laser interaction, similar to an earlier observation of this behaviour by Booth et al.^[27^]. This directional emission of destructive debris is favourable in situations of tight laser focusing. The latter is required to reach ultrahigh intensities, but brings the precious final focusing optic into close vicinity to the debris source. Directional debris can be mitigated by choosing the laser-incidence angle large enough to avoid an intersection of target normal and focusing optics. The population of debris that is emitted spherically uniformly poses a much lower risk as it can be addressed by available mitigation schemes, that is, spinning protection disks^[36^].

The experimental results show two small directional marks next to one large mark, which is counter-intuitive when compared with the modelling that predicts a larger total emission of debris for smaller targets^[26^]. If the presumption is correct that both small marks correspond to both shots on small targets, then the directional fraction of debris is smaller for small disk targets than for large ones. However, the larger recirculating electron population for smaller targets may yield evaporation to higher temperature, and heating for longer times. Therefore, the amount of spherically emitted debris can be higher for smaller targets than for larger targets. A larger fraction of spherically emitted debris will constrain the directional population.

The characteristic hourglass shape of the directional debris marks might encode valuable information about the laser–target interaction. Studies on laser-induced forward and backward transfer in the long-pulse regime show the ejection of debris dependent on the laser pulse width, laser pulse energy density and target–catcher distance^[37^–39^]. Further investigation is required to evaluate if debris can be an auxiliary metrology on the laser focal spot profile and temporal laser contrast.

This work took advantage of the benefit of the uniform absorption curve of nickel across the visible spectrum to introduce a fast spatially resolved method of debris characterization. When using spectrometers instead of a flatbed scanner, surface plasmons might be a way to characterize not only the thickness of a layer but also the size of nano-structures when using materials that exhibit a large surface plasmon strength^[40^].

We present a novel method for the characterization of thin layers of debris deposit based on RGB transmission scans that can be performed with commercial flatbed scanners. Initially transparent debris shields from FS are successfully used as debris catchers during experiments with high-power ultra-relativistic laser pulses irradiating solid-density targets. Scans reveal two distinct types of debris: (i) narrow emission cones away from target front- and rear-side normal direction and (ii) spherical emission. While more debris of type (i) is emitted away from the target front, type (ii) shows a slight asymmetry favouring the target rear side. The former agrees with previous works^[27^], while the latter might be due to the overall asymmetric space charge distribution induced by the laser–plasma interaction.

However, the method developed here is applicable only to materials with a smooth transmittance in the optical regime, and cannot be applied to target materials such as semimetals (aluminium) and noble metals (gold), which exhibit plasmons. Nevertheless, with enough sensor sensitivity, this method can be applied to plastic targets. It is further important for the direct applicability of this method that the surface of the sample is flat, such that the amount of transmitted light is not further reduced by diffuse reflection, which is not taken into account.

The quantitative characterization of the amount of debris and the direction of ejection can be used to promote the implementation of novel schemes that mitigate its deleterious effect on optical components and diagnostics.