Sub-Threshold Fabrication of Laser-Induced Periodic Surface Structures on Diamond-like Nanocomposite Films with IR Femtosecond Pulses

In the paper, we study the formation of laser-induced periodic surface structures (LIPSS) on diamond-like nanocomposite (DLN) a-C:H:Si:O films during nanoscale ablation processing at low fluences—below the single-pulse graphitization and spallation thresholds—using an IR fs-laser (wavelength 1030 nm, pulse duration 320 fs, pulse repetition rate 100 kHz, scanning beam velocity 0.04–0.08 m/s). The studies are focused on microscopic analysis of the nanostructured DLN film surface at different stages of LIPSS formation and numerical modeling of surface plasmon polaritons in a thin graphitized surface layer. Important findings are concerned with (i) sub-threshold fabrication of high spatial frequency LIPSS (HSFL) and low spatial frequency LIPSS (LSFL) under negligible surface graphitization of hard DLN films, (ii) transition from the HSFL (periods of 140 ± 30 and 230 ± 40 nm) to LSFL (period of 830–900 nm) within a narrow fluence range of 0.21–0.32 J/cm2, (iii) visualization of equi-field lines by ablated nanoparticles at an initial stage of the LIPSS formation, providing proof of larger electric fields in the valleys and weaker fields at the ridges of a growing surface grating, (iv) influence of the thickness of a laser-excited glassy carbon (GC) layer on the period of surface plasmon polaritons excited in a three-layer system “air/GC layer/DLN film”.


Introduction
Femtosecond laser-induced periodic surface structures (LIPSS), brightly called a scientific evergreen [1], remain to be of undiminishing interest with regard to both the LIPSS formation mechanisms and various applications of laser-structured surfaces [2][3][4][5][6][7][8][9][10][11][12]. For diamond-like carbon (DLC) films, femtosecond (fs) laser ablation and LIPSS formation are preceded by a structural transformation (graphitization) in a thin surface layer [13][14][15][16][17]. The surface graphitization changes surface properties and affects the ablation process and periodic surface nanostructuring during multipulse irradiation [3,10,[15][16][17]. Generally, the fs-LIPSS formation on DLC films is attributed to the excitation of surface plasmon polaritons (SPPs) in the graphitized layer, responsible for the periodic enhancement of local laser fields and spatial modulation of ablation rates on the film surface [4,7]. The structure of the graphitized DLC surface, where LIPSS are produced, is that of glassy carbon (GC), according to Raman spectroscopy studies [3,10,15]. The surface relief analysis of the laser-irradiated swelled regions gives an estimation of a few hundred nanometers for the graphitized layer thickness [10,13,16]. The presence of such a 'glassy carbon' layer makes a dielectric DLC film a two-layer GC/DLC coating with a top conductive layer having identical optical properties and controlling the interference between surface plasmons and incident light. This can qualitatively explain similar 'LIPSS period-on-fluence'

Materials and Methods
The DLN films on Si substrates, produced by plasma-assisted chemical vapor deposition from polymethylphenylsiloxane vapors, were used in fs-laser processing experiments. The DLN films of 2.5-µm thickness were characterized by the hardness 27 GPa, elastic modulus 143 GPa, and compressive stress 268 MPa. All details about the deposition technique and mechanical properties of the films are given in our recent paper [22]. The optical properties of the DLN films, including the refractive index n = 2, extinction coefficient k a = 0.021, and absorption coefficient α = 2.6 × 10 3 cm −1 at the laser wavelength λ = 1030 nm, were reported in our previous article [17].
Laser ablation processing of the DLN films was carried out using a SATSUMA HP2 femtosecond laser system [30,31] (from Amplitude Systèmes, Pessac, France), generating pulses of τ = 320 fs duration at the wavelength λ = 1030 nm. In the experiments, the average power (P) was varied from 100 to 154 mW at the pulse repetition rate (f ) 100 kHz; the pulse energy (ε p ) was changed from 1 to 1.54 µJ. The laser beam was focused with a 160-mm f-theta objective, and the beam radius (w o ) of 17.5 µm at the 1/e 2 level was determined with a scanning slit beam profiler from Thorlabs (Newton, NJ, USA). For the used pulse energies, the peak fluence F = 2ε p /πw o 2 was ranged from F = 0.21 to 0.32 J/cm 2 , with these values being smaller than the single-pulse graphitization and spallation thresholds-F gr = 0.3 J/cm 2 and F spal = 0.35 J/cm 2 [17]. A high-precision galvanometer scanner in-telliSCAN se from Scanlab (Puchheim, Germany) was applied to control the scanning beam velocities (v s ) and to produce linear surface microstructures (microgrooves) of 10-20 µm width and 15 mm length at different irradiation conditions. The microgrooves were produced at the fluences increasing from F = 0.21 to 0.32 J/cm 2 , two different scanning velocities of 0.08 m/s and 0.04 m/s, and the pulse repetition rate of 100 kHz. For the two scanning velocities, the pitch distance (v s /f ) was 0.8 µm and 0.4 µm, respectively. The ablation rate was determined as the groove depth divided by the effective pulse number N = w gr /(v s /f ) (w gr is the groove width). The laser beam was linearly polarized and the direction of the beam polarization was perpendicular to the scanning direction. All laser processing experiments were carried out in ambient air at the normal beam incidence onto the film surface.
The dimensions and surface morphology of the microgrooves were examined using scanning electron microscopy (SEM), white-light interferometry (WLI), and atomic force microscopy (AFM). The data of the DLN structure transformation in a thin surface layer was obtained from Raman spectra measured at the excitation wavelength 532 nm. The two-dimensional Fast Fourier Transform (FFT) analysis of SEM and AFM images was applied to determine the periods of LIPSS.

Results and Discussion
SEM images of microgrooves produced at low fluences (below the F gr and F spal thresholds) and two scanning velocities are shown in Figure 1, along with the ablation characteristics (Figure 1d,e) and Raman spectra (Figure 1f) of the grooves. used pulse energies, the peak fluence F = 2εp/πwo 2 was ranged from F = 0.21 to 0.32 J/cm 2 , with these values being smaller than the single-pulse graphitization and spallation thresholds-Fgr = 0.3 J/cm 2 and Fspal = 0.35 J/cm 2 [17]. A high-precision galvanometer scanner intelliSCANse from Scanlab (Puchheim, Germany) was applied to control the scanning beam velocities (vs) and to produce linear surface microstructures (microgrooves) of 10-20 μm width and 15 mm length at different irradiation conditions. The microgrooves were produced at the fluences increasing from F = 0.21 to 0.32 J/cm 2 , two different scanning velocities of 0.08 m/s and 0.04 m/s, and the pulse repetition rate of 100 kHz. For the two scanning velocities, the pitch distance (vs/f) was 0.8 µ m and 0.4 µ m, respectively. The ablation rate was determined as the groove depth divided by the effective pulse number N = wgr/(vs/f) (wgr is the groove width). The laser beam was linearly polarized and the direction of the beam polarization was perpendicular to the scanning direction. All laser processing experiments were carried out in ambient air at the normal beam incidence onto the film surface. The dimensions and surface morphology of the microgrooves were examined using scanning electron microscopy (SEM), white-light interferometry (WLI), and atomic force microscopy (AFM). The data of the DLN structure transformation in a thin surface layer was obtained from Raman spectra measured at the excitation wavelength 532 nm. The two-dimensional Fast Fourier Transform (FFT) analysis of SEM and AFM images was applied to determine the periods of LIPSS.

Results and Discussion
SEM images of microgrooves produced at low fluences (below the Fgr and Fspal thresholds) and two scanning velocities are shown in Figure 1, along with the ablation characteristics (Figure 1d,e) and Raman spectra (Figure 1f) of the grooves.  . SEM images of microgrooves produced on the DLN film with a scanning beam of the IR fs-laser at (a) F = 0.21 J/cm 2 and V s = 0.08 m/s, (b) F = 0.32 J/cm 2 and V s = 0.08 m/s, (c) F = 0.32 J/cm 2 and V s = 0.04 m/s, the polarization direction is perpendicular to the beam scanning direction; (d) the ablation rate vs. fluence for the two scanning velocities, arrows indicate the single-pulse graphitization and spallation thresholds; (e) AFM surface profiles of the microgrooves shown in (a,b); (f) Raman spectra (normalized to the maximum intensity) from central regions of the microgrooves, produced at F = 0.21 J/cm 2 and F = 0.32 J/cm 2 and two scanning velocities of 0.08 and 0.04 m/s. Changes in the groove morphology with increased fluence at the v s = 0.08 m/s, shown in Figure 1a,b, evidence an appearance of the relief periodicity, which is transformed to LIPSS (Figure 1c) at the lower velocity v s = 0.04 m/s and correspondingly larger pulse number. Details of the surface morphology in the central regions of the microgrooves fabricated at different fluences (0.21, 0.25, and 0.32 J/cm 2 ) and two scanning velocities are presented in SEM images of higher magnification in Figure 2. These images confirm the formation of LIPSS at all the fluences and low scanning velocity v s = 0.04 m/s, attributing the surface relief changes at the higher velocity v s = 0.08 m/s (lower effective pulse number) to an initial stage of LIPSS formation. and Vs = 0.04 m/s, the polarization direction is perpendicular to the beam scanning direction; (d) the ablation rate vs. fluence for the two scanning velocities, arrows indicate the single-pulse graphitization and spallation thresholds; (e) AFM surface profiles of the microgrooves shown in (a,b); (f) Raman spectra (normalized to the maximum intensity) from central regions of the microgrooves, produced at F = 0.21 J/cm 2 and F = 0.32 J/cm 2 and two scanning velocities of 0.08 and 0.04 m/s. Changes in the groove morphology with increased fluence at the vs = 0.08 m/s, shown in Figure 1a,b, evidence an appearance of the relief periodicity, which is transformed to LIPSS (Figure 1c) at the lower velocity vs = 0.04 m/s and correspondingly larger pulse number. Details of the surface morphology in the central regions of the microgrooves fabricated at different fluences (0.21, 0.25, and 0.32 J/cm 2 ) and two scanning velocities are presented in SEM images of higher magnification in Figure 2. These images confirm the formation of LIPSS at all the fluences and low scanning velocity vs = 0.04 m/s, attributing the surface relief changes at the higher velocity vs = 0.08 m/s (lower effective pulse number) to an initial stage of LIPSS formation.  The results of SEM, AFM, and Raman spectroscopy examination of the microgrooves show a number of interesting features of sub-threshold ablation processing with an IR femtosecond-laser. At all fluences used, the fs-laser processing and periodic surface nanostructuring of the DLN film occur without noticeable surface swelling arising from  The results of SEM, AFM, and Raman spectroscopy examination of the microgrooves show a number of interesting features of sub-threshold ablation processing with an IR femtosecond-laser. At all fluences used, the fs-laser processing and periodic surface nanostructuring of the DLN film occur without noticeable surface swelling arising from the surface graphitization [10,16]; sharp swellings of~50 nm height seen in the surface profiles in Figure 1e are due to protruding thin layers caused by their spallation and fracture at the groove edges. It is different from the previous data of fs-LIPSS in DLN [16,17] and ta-C [10] films with the LIPSS formed on swelled graphitized regions of the film surface. These findings are consistent with the Raman spectra shown in Figure 1f, which indicate minor changes in the structure of laser-graphitized surface layers with increased fluence/pulse number and very weak surface graphitization during sub-threshold processing and LIPSS formation. The spectrum of the original film is characterized by an asymmetric broad G-peak near 1490 cm −1 . In the Raman spectra of the microgrooves, one can see only an appearance of a 'shoulder' of the slightly increased intensity in a region of the D peak around~1300-1350 cm −1 and hardly observable changes near the G-peak. The observed minor changes in the spectra evidence that the thickness of a laser-graphitized surface layer is very small. Using the procedure of the original film spectrum subtraction described in the recent papers [26,32], we estimate the graphitized layer (GL) thickness of <10 nm on the groove surface under the processing conditions studied. The fs-laser processing without noticeable surface graphitization is indicative of a thermo-mechanical mechanism of ablation to be dominant at sub-threshold fluences. This is also supported by the clear evidence of laser-induced spallation on the tops of hill-like regions of the surface gratings, shown in Figure 2d,e, and by the presence of numerous nanoparticles (of a few tens nanometers size) on the surface of microgrooves after processing at the higher scanning velocity, as is seen in the images in Figure 2a-c. The formation of very specific nanostructures from nanoparticles is found to occur at the beginning of the LIPSS growth, which is considered to result from the interference of incident linearly-polarized light with surface plasmons (SP) and variation of local ablation thresholds in valleys and hills of a growing surface grating.
The typical surface morphology inside laser-ablated microgrooves is shown in Figure 1a, also observed in the previous laser structuring experiments of DLN films [23,26]. It is characterized by arc-like curves/regions consisting of nanoparticles and corresponding to the back edges of the laser spot moving in the direction from left to right with the pitch length of v s /f = 0.8 µm. Most interestingly, these arc-like regions are found to transform to sinusoid-like 'nanoparticle' curves in the groove central areas at the higher fluence F = 0.32 J/cm 2 , shown in Figures 1b and 2c. So they also have been the back edges of the laser spot moving from left to right with a pitch length of 0.8 µm. Similar sinusoid-like structures from nanoparticles are noticeable at lower fluences as well, in Figure 2b and even in Figure 2a, but they are weakly pronounced due to a smaller period of nascent LIPSS and relatively large size (a few tens of nanometers) of nanoparticles.
We assume that the edges of the moving spot 'show' the equi-field (equi-fluence) lines coinciding with the ablation (spallation) threshold of the groove surface or the ablation threshold of nanoparticles redeposited in the groove during processing. Under this assumption, we estimate the SP fields using the data of Figures 1a and 2a,c, as described below.
The radial distribution E L (r) of the electric field of the incident laser beam is given by where E 0 is the electric field in the spot center, w o is the beam radius at the 1/e 2 level, x and y are the coordinates parallel and perpendicular to the beam scanning direction. The beam polarization is directed normally to the scanning direction, i.e., along the y coordinate.
The electric field oscillations of the standing wave of SP (along the y axis) are described by where λ SP is the wavelength of the SP, E SP,MAX is the amplitude of the tangential component of the electric field (the amplitude of the SP field is assumed constant as the 'spot edge' width ∆r is much less than its radius r th , see below). The positive values of E SP (y) mean in-phase oscillations of the SP and laser fields and the negative values mean anti-phase oscillations of the fields.
Taking the radius r th to define the ablation threshold of the groove surface, the threshold laser field in the absence of the SP is determined from Equation (1) In the presence of the SP, the radius r th is the average value between two boundaries at the distances of r th ± ∆r, as marked in Figure 4a by a green curve and two red curves, respectively. This is due to the interference of the laser field with the tangential (parallel to the surface) component of the electric field of the surface plasmon generated in the graphitized layer. Decomposing E L (r) into a Taylor series, we obtain the equation.
from which the amplitude of the tangential component of the SP field is determined by   Figure 4a), we estimate the tangential component of the electric field of the SP generated in the graphitized layer at the lowest fluence F = 0.21 J/cm 2 . For the optical properties of the graphitized surface layer taken identical to the GC properties εGC = 3.6 + 4.5i at 1000 nm [18] and laser pulse duration τ = 320 fs, we obtain the following values of (i) laser intensity in the spot center (in the GL) I0 = (1 − R) F/τ = 5.9 × 10 11 W/cm 2 , (ii) laser field in the spot center E0 = (2 I0/(c n ε0)) 1/2 = 1.52 × 10 9 V/m, (iii) Eth = 5.6 × 10 8 V/m, and (iv) ESP,MAX = 1.3 × 10 7 V/m (R = 0.1 is the reflection coefficient of the GC surface).
At the higher fluences, especially at F = 0.32 J/cm 2 , the values of Δr increase up to 600 nm, see Figure 2c. According to Equation (4), the SP field is estimated to be more than three times the above value of ESP,MAX = 1.3 × 10 7 V/m, as the Eth remains the same and the rth increases with fluence. It is seen in Figures 1b and 2c that the amplitude of the equi-field sinusoid-like curves changes along the groove which could be dealt with a variable depth of a growing surface nanograting and enhancement of evanescent electric fields with the grating depth at the corrugated surface [33]. Most importantly, the location of the equi-field 'nanoparticle' curves on the emerging surface grating of 10-15 nm depth, shown in an AFM image in Figure 5, proves that the larger local electric fields are in the valleys and the weaker fields are at the ridges of the growing LIPSS. In addition, the observed fluence-dependent transformation of the convex 'averaged Eth' curves (related to the back edges of the laser spot, in Figure 1a) to the slightly concave ones (in the central area of the groove, in Figure 1b) is very unusual and not understood. It may be The electric field oscillations of the SP are shown schematically in Figure 4b. The interference of the SP and incident linearly-polarized light defines the boundary of the ablation threshold inside the groove by the equation.
The solution of Equation (5) is x = (w o 2 ln(E 0 /(E th − E SP (y)) − y 2 ) 1/2 ; it is shown in Figure 4c by a sinusoid curve in the coordinates x' = r th + ∆r-x, y, with the positive E SP,MAX values corresponding to the ablation threshold field E th at the (r th + ∆r) radius and the negative E SP,MAX values-to the E th at the (r th − ∆r) radius.
Taking the radius of the spot edges equal approximately to r th = 17.5 µm (from Figure 1a) and ∆r = 200 nm (from Figure 2a or Figure 4a), we estimate the tangential component of the electric field of the SP generated in the graphitized layer at the lowest fluence F = 0.21 J/cm 2 . For the optical properties of the graphitized surface layer taken identical to the GC properties ε GC = 3.6 + 4.5i at 1000 nm [18] and laser pulse duration τ = 320 fs, we obtain the following values of (i) laser intensity in the spot center (in the GL) I 0 = (1 − R) F/τ = 5.9 × 10 11 W/cm 2 , (ii) laser field in the spot center E 0 = (2 I 0 /(c n ε 0 )) 1/2 Materials 2022, 15, 4506 7 of 13 = 1.52 × 10 9 V/m, (iii) E th = 5.6 × 10 8 V/m, and (iv) E SP,MAX = 1.3 × 10 7 V/m (R = 0.1 is the reflection coefficient of the GC surface).
At the higher fluences, especially at F = 0.32 J/cm 2 , the values of ∆r increase up to 600 nm, see Figure 2c. According to Equation (4), the SP field is estimated to be more than three times the above value of E SP,MAX = 1.3 × 10 7 V/m, as the E th remains the same and the r th increases with fluence. It is seen in Figures 1b and 2c that the amplitude of the equifield sinusoid-like curves changes along the groove which could be dealt with a variable depth of a growing surface nanograting and enhancement of evanescent electric fields with the grating depth at the corrugated surface [33]. Most importantly, the location of the equi-field 'nanoparticle' curves on the emerging surface grating of 10-15 nm depth, shown in an AFM image in Figure 5, proves that the larger local electric fields are in the valleys and the weaker fields are at the ridges of the growing LIPSS. In addition, the observed fluence-dependent transformation of the convex 'averaged E th ' curves (related to the back edges of the laser spot, in Figure 1a) to the slightly concave ones (in the central area of the groove, in Figure 1b) is very unusual and not understood. It may be due to the surface relief of the microgroove changing with fluence ( Figure 1e) when the effects of the groove slope on the local electric fields should be considered in further analysis and checked in experiments. More particular experimental data on the ablation mechanism and formation of nanoparticles during the sub-threshold processing with IR fs-laser are also required for a better understanding of the LIPSS growth on the DLN surface in dependence on the irradiation conditions. The SEM images (in Figure 2) of the LIPSS formed at the vs = 0.04 m/s evidence that the LIPSS period (Λ) increases with fluence within the range from 0.21 to 0.32 J/cm 2 , very similar to the period-on-fluence dependences for the fs-LIPSS produced on other carbon-based coatings [10]. The data of the LIPSS period was obtained from the FFT analysis of SEM images of 6 μm × 4 μm size and AFM images of 5 μm × 5 μm size. The one-dimensional FFT spectra of the SEM images shown in Figure 2d-f, characterizing the periodical nanorelief on the bottom of the microgrooves, are presented in Figure 6. It is seen that the LIPSS period increases from Λ = 140 nm to Λ = 840 nm, being significantly smaller than the laser wavelength (Λ < λ/7) at F = 0.21 J/cm 2 and becoming close to the laser wavelength at F = 0.32 J/cm 2 . Following the proposed terminology of LIPSS [1,12], this means that the change from the high spatial frequency LIPSS (HSFL) to the low spatial frequency LIPSS (LSFL) occurs during sub-threshold processing of DLN films with IR fs-laser pulses in the narrow fluence range under consideration. The SEM images (in Figure 2) of the LIPSS formed at the v s = 0.04 m/s evidence that the LIPSS period (Λ) increases with fluence within the range from 0.21 to 0.32 J/cm 2 , very similar to the period-on-fluence dependences for the fs-LIPSS produced on other carbon-based coatings [10]. The data of the LIPSS period was obtained from the FFT analysis of SEM images of 6 µm × 4 µm size and AFM images of 5 µm × 5 µm size. The one-dimensional FFT spectra of the SEM images shown in Figure 2d-f, characterizing the periodical nanorelief on the bottom of the microgrooves, are presented in Figure 6. It is seen that the LIPSS period increases from Λ = 140 nm to Λ = 840 nm, being significantly smaller than the laser wavelength (Λ < λ/7) at F = 0.21 J/cm 2 and becoming close to the laser wavelength at F = 0.32 J/cm 2 . Following the proposed terminology of LIPSS [1,12], this means that the change from the high spatial frequency LIPSS (HSFL) to the low spatial frequency LIPSS (LSFL) occurs during sub-threshold processing of DLN films with IR fs-laser pulses in the narrow fluence range under consideration.
periodical nanorelief on the bottom of the microgrooves, are presented in Figure 6. It is seen that the LIPSS period increases from Λ = 140 nm to Λ = 840 nm, being significantly smaller than the laser wavelength (Λ < λ/7) at F = 0.21 J/cm 2 and becoming close to the laser wavelength at F = 0.32 J/cm 2 . Following the proposed terminology of LIPSS [1,12], this means that the change from the high spatial frequency LIPSS (HSFL) to the low spatial frequency LIPSS (LSFL) occurs during sub-threshold processing of DLN films with IR fs-laser pulses in the narrow fluence range under consideration.  An interesting feature of the obtained 1D-FFT spectra is related to the appearance of a second peak of higher frequency (smaller period) in the FFT spectra of the SEM images of the groove surface processed at two low fluences, especially in the spectrum shown in Figure 6b. The higher-frequency peak, marked as Λ 2s in Figure 6b (and Λ 1s in Figure 6a), is due to high-brightness edges of thin layers originating from laser-induced spallation on the tops of hill-like regions of the forming surface grating. This makes the Λ 2s (and Λ 1s ) peak not relevant to the true period of the formed LIPSS, as it is defined principally by the yield of secondary electrons but not the surface grating parameters. The data of the FFT analysis of the periodical surface nanostructures formed on the DLN film in sub-threshold fs-laser processing are summarized as the LIPSS period-on-fluence dependence in Figure 7. An interesting feature of the obtained 1D-FFT spectra is related to the appearance of a second peak of higher frequency (smaller period) in the FFT spectra of the SEM images of the groove surface processed at two low fluences, especially in the spectrum shown in Figure 6b. The higher-frequency peak, marked as Λ2s in Figure 6b (and Λ1s in Figure 6a), is due to high-brightness edges of thin layers originating from laser-induced spallation on the tops of hill-like regions of the forming surface grating. This makes the Λ2s (and Λ1s) peak not relevant to the true period of the formed LIPSS, as it is defined principally by the yield of secondary electrons but not the surface grating parameters. The data of the FFT analysis of the periodical surface nanostructures formed on the DLN film in sub-threshold fs-laser processing are summarized as the LIPSS period-on-fluence dependence in Figure 7. It should be noted that the low spatial frequency peak (Λ3) is present in all the FFT spectra of Figure 6, although weakly pronounced in the spectra in Figure 6a,b for two low fluences. In the frame of the SPP model in multilayer systems [34] (also referred to as a thin-film SPP model [35]), simultaneous observation of the low spatial frequency and It should be noted that the low spatial frequency peak (Λ 3 ) is present in all the FFT spectra of Figure 6, although weakly pronounced in the spectra in Figure 6a,b for two low fluences. In the frame of the SPP model in multilayer systems [34] (also referred to as a thin-film SPP model [35]), simultaneous observation of the low spatial frequency and high spatial frequency peaks seems to be a consequence of the localization of coupled SPPs at the two interfaces (air/GL and GL/DLN) in a thin graphitized layer on the film surface and fluence-dependent effect of the excited SPP modes on the LIPSS formation. In addition, a superposition of the HSFL and LSFL is found to occur in local regions of the microgroove produced at F = 0.25 J/cm 2 and v s = 0.04 m/s, as shown in SEM and AFM images and FFT spectrum in Figure 8; the values of the HSFL and LSFL periods are Λ 2 ≈ 280 nm and Λ 3 ≈ 900 nm. Whether it is a further proof of the coupled SPPs manifestation or a short-time instability of the pulse energy during beam scanning (e.g., pulse energy increases by 20-30% in several successive laser shots)-both will need more experimental data to clarify the emergence of LSFL under the above lower fluence processing. Generally, the SEM image in Figure 8a is a unique picture of the LIPSS-structured DLN surface. It illustrates the effect of fluence on the HSFL period increasing from the edge to center regions due to the Gaussian intensity distribution and occurrence of the LSFL in local areas of the HSFL-structured surface. The discussed-above 'artificial' nature of the Λ 2s peak is evidenced as well, as the peak amplitude is strongly reduced in the FFT spectrum of an AFM image which shows the true relief of the surface grating formed on the groove bottom (see Figure 8b,c). For the analysis of fs-LIPSS periods obtained under the low-fluence processing and weak surface graphitization of the DLN film, we applied the SPP model for multilayer systems (consisting of conducting and dielectric layers) [34] to calculate the periods of SPPs excited within a thin graphitized layer of different thickness ranging from 5 to 100 nm. The geometry of a three-layer system, including the air environment (medium 2), graphitized layer of the thickness h (medium 1), and DLN film (medium 3), is given in Figure 9. The concentration of the laser-induced plasma (Ne) was assumed to be uniform For the analysis of fs-LIPSS periods obtained under the low-fluence processing and weak surface graphitization of the DLN film, we applied the SPP model for multilayer systems (consisting of conducting and dielectric layers) [34] to calculate the periods of SPPs excited within a thin graphitized layer of different thickness ranging from 5 to 100 nm. The geometry of a three-layer system, including the air environment (medium 2), graphitized layer of the thickness h (medium 1), and DLN film (medium 3), is given in Figure 9. The concentration of the laser-induced plasma (N e ) was assumed to be uniform throughout the entire volume of the graphitized layer, and the DLN electronic subsystem was considered unexcited. The properties of the graphitized layer were taken identical to the optical properties of glassy carbon at λ = 1000 nm [18,36]. For the analysis of fs-LIPSS periods obtained under the low-fluence processing and weak surface graphitization of the DLN film, we applied the SPP model for multilayer systems (consisting of conducting and dielectric layers) [34] to calculate the periods of SPPs excited within a thin graphitized layer of different thickness ranging from 5 to 100 nm. The geometry of a three-layer system, including the air environment (medium 2), graphitized layer of the thickness h (medium 1), and DLN film (medium 3), is given in Figure 9. The concentration of the laser-induced plasma (Ne) was assumed to be uniform throughout the entire volume of the graphitized layer, and the DLN electronic subsystem was considered unexcited. The properties of the graphitized layer were taken identical to the optical properties of glassy carbon at λ = 1000 nm [18,36]. Figure 9. The geometry of a three-layer system used in numerical calculations of the SPP excitation in a thin glassy carbon layer on the DLN film.
The dielectric permittivity of the graphitized layer is given by ε1(ω) = εGC − ωp 2 /ω 2 (for simplicity, the damping of free carriers is ignored), where εGC = 3.6 + 4.5i is the dielectric constant of the unexcited GC material, ω is the frequency of the laser wavelength, ωp = [Ne The dielectric permittivity of the graphitized layer is given by ε 1 (ω) = ε GC − ω p 2 /ω 2 (for simplicity, the damping of free carriers is ignored), where ε GC = 3.6 + 4.5i is the dielectric constant of the unexcited GC material, ω is the frequency of the laser wavelength, ω p = [N e e 2 /(ε 0 m)] 1/2 is the plasma frequency, e is the electron charge, ε 0 is the dielectric permittivity of vacuum, m = 0.5 m e is the optical effective mass of electrons. The dielectric constants of air and DLN film are ε 2 = 1 and ε 3 = 4 + 0.084i. The excitation of SPPs at the metal-dielectric interfaces is allowed for Re(ε 1 ) < 0, Re(ε 2,3 ) > 0. The condition of Re(ε 1 ) = 0 is fulfilled at N e =1.9 × 10 21 cm −3 for the laser wavelength λ = 1030 nm, and calculations were carried out for the N e values from 1.9 × 10 21 to 1 × 10 23 cm −3 .
The SPP wave numbers of a dielectric-metal-dielectric system are described by an expression [34].
where k 2 i = β 2 − k 2 0 ε i is the component of the wave vector perpendicular to the interfaces in the medium i (i = 1, 2, 3), a = h/2 is a half the thickness of a metal layer (i.e., GC layer in the calculations), β is the SPP wave number, and k 0 is the wave vector of light in vacuum.
For a → ∞, the Equation (6) transforms to the equation of SPPs at the respective interfaces (air/GC and GC/DLN) between the two half-spaces [34] Equation (6) was numerically solved for β for different thicknesses of the conducting GC layer and optical constants ε i . Then, the SPP period Λ SPP was calculated as Λ SPP = 2π/Re(β) [35]. The results of the numerical calculations for the GC layer thickness of h = 5, 10, 20, and 100 nm are shown in Figure 10. The data calculated for the single interfaces (air/GC and GC/DLN) using Equation (7) are also presented for comparison. It follows from Figure 10 that the periods of the HSFL Λ 1 = 140 ± 30 nm and Λ 2 = 230 ± 40 nm are in good agreement with the SPP periods for the carrier densities in the range of N e~5 × 10 21 − 1.4 × 10 22 cm −3 . It is worth noting that the solutions of Equation (6), for which Im(β) > Re(β), are not considered due to the high absorption of such SPP modes, making their appearance unlikely. For the very thin laser-excited GC layers (of 5-20 nm thick), our calculations show that Im(β) > Re(β) is observed at N e < 4.7 × 10 21 cm −3 . The LSFL period Λ 3 = 830-900 nm is found to be close to periods of SPPs localized at the air/GC interface both in the thin-film and single-interface SPP models.
good agreement with the SPP periods for the carrier densities in the range of Ne ~ 5 × 10 21 − 1.4 × 10 22 cm −3 . It is worth noting that the solutions of Equation (6), for which Im(β) > Re(β), are not considered due to the high absorption of such SPP modes, making their appearance unlikely. For the very thin laser-excited GC layers (of 5-20 nm thick), our calculations show that Im(β) > Re(β) is observed at Ne < 4.7 × 10 21 cm −3 . The LSFL period Λ3 = 830-900 nm is found to be close to periods of SPPs localized at the air/GC interface both in the thin-film and single-interface SPP models. Figure 10. Period of SPP modes vs. free-carrier density, calculated using Equation (6) for different thicknesses of the laser-excited GC layer (5, 10, 20, and 100 nm); for comparison, the ΛSPP data calculated for the interfaces air/GC and GC/DLN using Equation (7)  Short remarks about the GL thickness and fs-laser processing at sub-threshold fluences are to be made. The very low GL thickness (<10 nm) together with low ablation Figure 10. Period of SPP modes vs. free-carrier density, calculated using Equation (6) for different thicknesses of the laser-excited GC layer (5, 10, 20, and 100 nm); for comparison, the Λ SPP data calculated for the interfaces air/GC and GC/DLN using Equation (7)  Short remarks about the GL thickness and fs-laser processing at sub-threshold fluences are to be made. The very low GL thickness (<10 nm) together with low ablation rates of 5-16 nm/pulse (which are much smaller than the optical absorption length 1/α~100 nm in the GL at λ = 1030 nm) are indicative of nonlinear absorption of IR fs-laser pulses in the surface layer followed by specific thermo-mechanical ablation process when the heated nm-thick layers are removed from the surface by spallation. (α ≈ 10 5 cm −1 is the absorption coefficient of GC [36]; data on the two-photon absorption coefficient of DLC are lacking in the literature [37]). Indeed, our previous data of numerical simulations of the temperature and surface graphitization in a 2.5-µm-thick DLN film at linear absorption and similar fluence evidenced the heat-diffusion-assisted growth of the graphitized layer under the actions of successive IR fs-laser pulses [17], very different from the results presented in this work. The total thickness of the ablated and graphitized surface layers h t = h abl + h gr is estimated to be 10-25 nm per laser shot at the minimal and maximal fluences (0.21 and 0.32 J/cm 2 ). The h t is equal to the maximal thermal length, which defines the time of removal (ablation, spallation) of the laser-excited layer as t abl = h t 2 /χ~0.5-3 ns, consistent with the typical timescales of the material response to fs-laser excitation [38]. So, the ablation rate defines the thickness of the laser-excited graphitized layer which is used in the numerical calculations and responsible for the observed LIPSS periods, while the thickness of the graphitized surface layer (determined after processing) is related to the material properties responsible for characteristics of the ablation process. In our experiments, these values almost coincide, characterizing the sub-threshold ablation processing and LIPSS formation without noticeable surface graphitization.

Conclusions
The important findings of this work result from distinctive features of sub-threshold processing of DLN films with the IR fs-laser pulses, characterized by nanoscale ablation, spallation, nanoparticle formation, and very weak surface graphitization. The formation/dimensions of LIPSS on the DLN film surface are shown to depend on the laser fluence and scanning beam velocity. Very specific 'sinusoid-like' nanostructures from nanoparticles are revealed to form on the ablated surface at an initial stage of the LIPSS formation, considered as a visualization of equi-field lines by nanoparticles due to the interference of incident linearly-polarized light with surface plasmons and variation of local ablation thresholds in valleys and hills of a growing surface grating. The observed location of the equi-field 'nanoparticle' curves on the emerging surface grating of 10-15 nm depth can be concluded as a direct experimental proof of the larger electric fields in the valleys and the weaker fields at the ridges of the growing LIPSS. Based on the FFT analysis of SEM/AFM images, it is found that the change from the HSFL (with the periods of Λ 1 = 140 ± 30 nm and Λ 2 = 230 ± 40 nm) to the LSFL (Λ 3 = 830-900 nm) occurs during sub-threshold processing of DLN films in the narrow fluence range of F = 0.21-0.32 J/cm 2 . The simultaneous observation of the low spatial frequency and high spatial frequency peaks in FFT spectra indicates the localization of coupled SPPs at the two interfaces (air/GL and GL/DLN) in a thin graphitized layer and the fluence-dependent effect of the excited SPP modes on either HSFL or LSFL formation. The HSFL periods are shown to be in good agreement with numerical calculations of the periods of SPPs excited within a GC layer of 5-20 nm thickness, performed with the SPP model for a three-layer system "air/GC layer/DLN film". The LSFL period is shown to be close to the periods of SPPs localized at the air/GC interface. The obtained results demonstrate that fabrication of both the HSFL and LSFL with negligible surface graphitization of hard DLN coatings can be realized during IR fs-laser processing at sub-threshold fluences.