**2. Materials and Methods**

The experimental setup used in this work is the same as that used by the authors in a previous work (Moreno-Diaz et al. [24]). A schematic diagram of the experimental set-up is shown in Figure 1a,b.

A laser pulse of a Q-switched Nd:YAG laser (of 10 ns, 1.06 μm of wavelength, and 2.5 J per pulse) was focused using a 20 cm focal lens on a sample of certified aluminum alloy (Al2024-T351). The composition of the sample included in addition to aluminum around 4% Cu, 1.5% Mg, and 0.6% Mn. As can be observed in Figure 1, a constant water flux was supplied in this experiment. A typical crater, see Figure 2, of ~2 mm diameter was produced in the sample surface after the laser pulse. The image was obtained with a confocal microscope (LEICA DCM 3D´Leica Microsistemas S.L.U., L'Hospitalet de Llobregat, Spain). The irradiance of laser was approximately 10 GW/cm2.

The plasma light was collected by an optical fiber and directed to the spectrograph provided with a diffraction grating of 1800 grooves/mm. The placement of a lens to collect more plasma light may be an alternative in air, but the lens has been shown to be badly damaged by water splashes in the laboratory and probably would be even more so in an industrial environment.

Emission spectra were acquired using a spectrograph (Horiba Jobin Ybon FHR1000, HORIBA UK Limited, Northampton, United Kingdom) equipped with an ICCD camera (Andor, model iStar 334T, Oxford Instruments, Concord, MN, USA).

Measurements were taken 2 mm from the target surface (where the signal/noise ratio was the best). As we mentioned in our previous work, in spite of our attempts, the quality of the signal prevented us from making the Abel inversion, and so our data are spatially integrated and present the temporary integration of the measurement gate time.

A low-pressure Ne lamp (Oriel 6032, Newport, Irvine, CA, USA) was used to calibrate the wavelength scale. The instrumental profiles in the interest range were measured with a He-Ne laser, checking its full width at half maximum. The instrumental bandwidths in this range were found on average to be 0.18 ± 0.01 Å. In this bandwidth, it was found by numerical adjustment that the Lorentzian contribution was practically null.

**Figure 1.** Experimental setup used in the reported Laser Shock Processing (LSP) experiments: (**a**) Scheme; (**b**) Photo.

**Figure 2.** A typical surface change after an LSP experiment. Images obtained via scanning confocal microscopy.

Relevant LSP experiments were performed both under air and water confinement. Under air-confinement conditions, the composition of the plasma reflected the composition of the sample, obtaining spectral lines of Al I and Al II, Mg I and Mg II, Cu I and Cu II, and Mn I and Mn II, in addition to the Hα-line of the hydrogen, which can only be due to very weak traces of hydrogen in the plasma and a possible reaction between the ionized aluminum and the residual humidity of the

air [21]. The presence of the water flow weakened the emissions of spectral lines differently than it did those of the Hα-line, which had a similar profile in air and in water flow in all cases. This effect can be observed in Figure 3.

As the profiles and the displacement of the Hα-line were practically identical under air and water flow confinement (LSP condition), the hypothesis of the similarity of the plasma behavior and properties under both conditions (already recognized in the previous work [24]) is considered sound.

**Figure 3.** Mg II emission lines in air and LSP conditions. Delay time: 2 μs. Gate time: 1000 ns.

In Figure 4a–c, images of the plasma trace at 2.5, 4, and 5 μs after laser pulse (gate time of 300 ns in all cases) and the corresponding spectra can be observed. As previously indicated, the presence of neutral hydrogen, although important for the obtained results, is residual, and confirms the practical non-absorption of the Hα line. The hostile conditions in which these measures must be carried out, with water in flow and splashing, makes it very difficult to experimentally contrast this claim. Nevertheless, the Hα lines did not show any self-absorption signs. In other words, there is full symmetry and no anomalous dip in the center. These facts have already been mentioned by El Sherbini et al. [19,21] in the electron density range of these kinds of experiments (described in Moreno-Diaz et al. [24]). Another relevant feature was the appearance of a red shift of the peak of the Hα-line which, in turn, was observed to decrease with the delay time.

In order to perform the analysis of the spectral lines, the obtained experimental profiles were adjusted to Voigt profiles generated numerically. To obtain the Lorentz experimental broadening of the line, the contribution of the instrumental profile was discounted.

Unlike the Lorentzian contribution, which was practically null, the Gaussian contribution of the instrumental profile of the used experimental device was important in relation to the Gaussian contribution found in the Voigt profiles of the Hα-line analyzed in this work. This fact advises against the use of this contribution to estimate the temperature of thermal agitation of hydrogen, which only in conditions of full thermodynamic equilibrium would coincide with the sought electronic temperature.

Figure 5 shows four examples of the fitting from a Voigt profile to experimental line emissions of the plasma. The Hα-lines corresponding to 4 μs delay time and gate times (emission integration times) of 100 ns, 300 ns, 500 ns, and 1000 ns are presented.

(**c**)

**Figure 4.** Spectrally resolved images of the plasma (with a gate time of 300 ns) at (**a**) 2.5 μs, (**b**) 4 μs, and (**c**) 5 μs after laser pulse with a wavelength range from 6530 to 6590 Å and spectra obtained from these images of the Hα-line in LSP conditions.

**Figure 5.** Fitting from a Voigt profile to experimental line emissions of the plasma with gate times of (**a**) 100 ns, (**b**) 300 ns, (**c**) 500 ns, and (**d**) 1000 ns at 4 μs delay time with a wavelength range from 6530 to 6590 Å.

The corresponding Stark broadening and red shift appear in the left side of every spectrum. As can be observed, the best line profiles are at 300 and 500 ns. At 100 ns and 1000 ns there are slight asymmetries that are attributed to the fact that signal integration times are not adequate. At 100 ns, the signal integration time is too short and at 1000 ns, it is too long. It is also observed that despite these disadvantages for the 100 ns and 1000 ns gate times, the Stark widths and shifts obtained are the same in all gates within the experimental error margins.
