Irregular LIPSS produced on metals by single linearly polarized femtosecond laser

Currently, supra-wavelength periodic surface structures (SWPSS) are only achievable on silica dielectrics and silicon by femtosecond (fs) laser ablation, while triangular and rhombic laser induced periodic surface structures (LIPSS) are achievable by circularly polarized or linear cross-polarized femtosecond laser. This is the first work to demonstrate the possibility of generating SWPSS on Sn and triangular and rhombic LIPSS on W, Mo, Ta, and Nb using a single linearly polarized femtosecond laser. We discovered, for the first time, SWPSS patches with each possessing its own orientation, which are completely independent of the light polarization direction, thus, breaking the traditional rules. Increasing the laser power enlarges SWPSS periods from 4–6 μm to 15–25 μm. We report a maximal period of 25 μm, which is the largest period ever reported for SWPSS, ∼10 and ∼4 times the maximal periods (2.4 μm/6.5 μm) of SWPSS ever achieved by fs and ns laser ablation, respectively. The formation of triangular and rhombic LIPSS does not depend on the laser (power) or processing (scan interval and scan methodology) parameters but strongly depends on the material composition and is unachievable on other metals, such as Sn, Al, Ti, Zn, and Zr. This paper proposes and discusses possible mechanisms for molten droplet generation/spread/ solidification, Marangoni convection flow for SWPSS formation, and linear-to-circular polarization transition for triangular and rhombic LIPSS formation. Reflectance and iridescence of as-prepared SWPSS and LIPSS are characterized. It was found that besides insufficient ablation on W, the iridescence density of Ta-, Mo-, Nb-LIPSS follows the sequence of melting temperatures: Ta > Mo > Nb, which indicates that the melting temperature of metals may affect the regularity of LIPSS. This work may inspire significant interest in further enriching the diversity of LIPSS and SWPSS.


Introduction
Laser induced periodic surface structuring (LIPSS) was discovered in 1965 [1]. The following decades witnessed rapid advancements in producing LIPSS on different metals, polymers, semiconductors, and glasses and their applications in various fields [2][3][4][5][6][7]. LIPSS structures possessing either a colorful iridescence [7] or a black color [8] are excellent biomimetic interfaces [6]. The convenience of producing LIPSS on different materials has made the direct laser writing technique more attractive than traditional expensive lithography techniques for nanostructuring. According to the decreased ratio of the LIPSS period to the laser wavelength, LIPSS can be divided into supra-wavelength periodic surface structures (SWPSS) and low/high/ultrahigh spatial frequency LIPSS (LSFL/HSFL/UHSFL). The SWPSS, whose period is normally larger than the laser wavelength, is generally prepared by nanosecond (ns) lasers because of their distinct thermal effects, but seldom by femtosecond (fs) lasers. Table 1 displays the reported scanning electron microscope (SEM) morphologies of SWPSS and summarizes the preparation conditions and the SWPSS periods. SWPSS is merely achievable on silica dielectrics [9][10][11][12] and silicon [13,14] by fs laser ablation in air (FLAA). Although SWPSS with periods in the range of 3-6.5 µm has been produced on chromium (Cr) and gold (Au) films using nanosecond lasers [15][16][17], no report has developed SWPSS on a metal target via fs laser ablation. All SWPSS were produced by static multi-pulse fs ablation. So far, the maximal period of SWPSS obtained by fs laser ablation is 2.4 µm [14]. One may wonder how SWPSS will evolve if a large-scale area of SWPSS is scanned and whether the periods of SWPSS can be further enlarged, and to which extent the period of SWPSS can be reached. The formation of SWPSS is associated with Marangoni convection hydrodynamics such as convection roll-driven hydrodynamic phenomena [10] and Marangoni convection properties such as capillarity convection force [15]. It is still questionable whether some other factor that can govern the SWPSS states has not yet been discovered and reported. This work seeks answers to these questions on the basis of fs laser of SWPSS on tin (Sn) metal targets.
Many efforts have also been devoted to manipulating the properties (orientation, period and morphology) of LIPSS and enriching LIPSS's diversity by changing the atmospheres [8,18], modulating light properties [19][20][21], and introducing external stimuli such as bubbles [22] and shockwaves [23]. Among all novel LIPSS structures, triangular and rhombic LIPSS are one of the most interesting series because of their high homogeneity and the possibility for large-scale production. Triangular and rhombic LIPSS have been produced on tungsten (W) [24,25], chromium (Cr) [26], cobalt (Co) [27], nickel (Ni) [7], stainless steel [28][29][30][31], and metal alloy [32] by a single beam with a circular/azimuthal/radial polarization or two time-delayed fs laser beams with cross linear polarizations, as summarized in table 2. Triangular and rhombic LIPSS may be formed via the non-linear convection flow mechanism [31] or self-organization driven by Coulomb repulsion [27] or noncollinear excitation of two surface plasmons [26]. No report has proven the feasibility to produce triangular and rhombic LIPSS using a single linearly polarized laser.
This work breaks the empirical laws for the preparation of both SWPSS and triangular/rhombic LIPSS and, for the first time, demonstrates the possibility of developing SWPSS on Sn and triangular and rhombic LIPSS on Ta, Mo, and Nb via single linearly polarized FLAA. The surface structures obtained by FLAA were characterized by SEM, which can be used for a direct comparison with the counterparts presented in tables 1 and 2. The possible mechanisms for the yield of SWPSS and triangular and rhombic LIPSS on different metals are discussed.

Experimental
A femtosecond laser (pulse duration of 400 fs, wavelength of 1030 nm, and repetition rate of 400 kHz) was used for the FLAA of Sn, Ta, Mo, Nb, W, Ti, Al, Zn, and Zr, as schematically illustrated in figure 1(a). The samples were scanned at a scan speed of 200 mm s −1 by a high-speed galvanometer scanner along different scan directions, as illustrated in figure 1(b). Special structures (figures 1(c) and (d)) were only discovered on Sn, W, Ta, Mo, and Nb, while normal LIPSS was produced on the Ti, Al, Zn, and Zr surfaces. The scan line intervals were set at 5, 10, 15 and 20 µm. Single scan and double scan were both performed for FLAA of Mo to demonstrate the repeatability of triangular and rhombic LIPSS. Different laser powers of 15, 8, 5, and 2 W were used for FLAA. The surface morphologies were characterized by SEM (NOVA NanoSEM 230, FEI). The reflectance of the samples in UV-to-NIR and NIR-to-MIR ranges was analyzed by UV-Vis (Lamda 950, PerkinElmer, Inc.) and FTIR spectroscopy (Nicolet 6700, Thermo Fisher). To simplify the description of the samples obtained under different conditions, the samples are named by material-power-interval. For example, Sn-2W-15 µm refers to a sample that was prepared by FLAA of Sn at a laser power of 2 W, and scan interval of 15 µm. The scan speed was constant at 200 mm s −1 for all samples. For iridescent observation, an LED white light at the power of 1 W was used for illumination well above the sample. A cell phone camera (Hua wei mate 20) was used to capture the optical images of ablated surfaces. Figure 2 displays the SEM and confocal microscopy morphologies of the structures obtained by FLAA of Sn at laser powers of 2, 5, and 8 W and a scan interval of 15 µm. The periods of SWPSS obtained at 2 W ranged from 4 to 6 µm (figure 2(a)), about 4-6 times the laser wavelength (1.03 µm), much larger than the periods of already published SWPSS prepared by fs lasers (table 1). The depth of such SWPSS is ∼10 µm (figure 2(d)). Increasing the laser power to 5 W for FLAA Table 1. Summary of SWPSS prepared by fs laser ablation. All figures are adapted with permission of publishers. Reprinted figure with permission from [10], Copyright (2016) by the American Physical Society. Reprinted with permission from [9] © The Optical Society. Reproduced from [11]. CC BY 4.0. Reprinted from [13], Copyright (2020), with permission from Elsevier. Reprinted with permission from [14]. Copyright (2018), AIP Publishing LLC. Reproduced from [12]. CC  led to the enlargement of SWPSS's periods to the range of 15-25 µm (figure 2(b)) and increased the SWPSS's depth to ∼20 µm (figure 2(d)). Currently, the largest SWPPS period is 25 µm, ∼10 and ∼4 times the maximal period (2.4 µm [14]/6.5 µm [17]) of SWPSS ever achieved by fs and ns laser ablation, respectively. It was reported that a higher substrate temperature facilitates the formation of SWPSS and can trigger the evolution of LSFLs into SWPSS on silica when the substrate temperature exceeds 850 • C [11], which may explain the increase in the SWPSS period. However, when the laser power increased to 8 W, many regions were covered by smooth wavy patches (figures 2(e) and (f)), which are deemed to originate from the solidification of strong molten layers. Figure 2 indicates that increasing the laser power significantly enhances the thermal effect because of the low melting temperature (232 • C) of Sn, which creates a favorable environment for the formation of larger and wider SWPSS. But the thermal effect should be controlled in a certain range; otherwise, the surface structures will be covered by the molten layers (figure 2(e)).

SWPSS on Sn
Of great interest, SWPSS with different orientations were produced (figures 2(a)-(d) and 3(a)-(f)) completely independent of the linear light polarization direction, which is considered the most crucial factor to govern the LIPSS's direction [2]. This discovery further enriches the diversity of SWPSS ever reported, including quasi-periodical SWPSS [15] and transverse and inclined SWPSS [17]. The formation of unusual SWPSS indicates the emergence of a new mechanism that differs from the conventional mechanisms, including interference, surface plasmon polaritons (SPPs), and second  harmonic generation [2,18]. Since these irregular SWPSS are very similar to the convection patterns triggered by temperature gradients [33], it is proposed that Marangoni convection flows propagating at speeds of 60-80 m s −1 [15,34,35] are  Figure 4 displays the possible formation mechanism of SWPSS during the FLAA of Sn at laser powers of 2, 5, and 8 W. The melting states and sizes of molten droplets depend on the laser powers. Higher laser powers generate larger molten droplets and induce higher melting rates. Previous reports have shown the sequential, layer-by-layer assembly of molten Sn droplets for three-dimensional manufacturing [36]. So, it is reasonable to deduce that at a laser power of 2 W, the interaction of fs laser and the Sn target will generate many separated molten droplets. High temperatures of molten droplets verify that remelting the substrate produces a robust bond between the spreading molten droplets and the ablated substrate [36]. The spread of molten droplets leads to the formation of SWPSS patches. Each molten droplet has a unique temperature gradient which drives the movement of the Marangoni convection flow from the hot region to the cold region [33]. Meanwhile, the convection forces, including the thermo-capillarity (Marangoni) shear stress and the stress originating from variations in the radius of the curvature of molten droplets, govern the orientations of SWPSS [15]. Due to the gentle thermal effect (2 W-FLAA), Marangoni convection flows, which are the precursors of SWPSS, tend to solidify very fast, so SWPSS patches with different orientations formed. The formation of microscale holes among the SWPSS arrays (figure 3) may be attributed to the gas capsulation [37] because the existence of ripple structures on molten Sn droplets can increase the possibility of gas capsulation [38] or complex fluidity of Marangoni flow such as Marangoni bursting [39] which has been proven to take place during laser ablation of metals [8].
When a stronger thermal effect was triggered at 5 W-FLAA, no patched SWPSS were generated due to the formation of larger droplets, which can promote the connection/coalescence of molten droplets with each other during their ejection/ spread/cooling processes. In addition, higher power must have increased the ablation productivity, so the density of molten droplets should have increased. The molten droplets may be concurrently and sequentially coalesced, analogous to the gradual increase in the volume of persistent bubbles generated during line-by-line fs laser ablation in water [22]. Further increasing the laser power to 8 W enhanced the thermal effect, which generated larger molten droplets with a longer lifetime. They will flow/eject on existing SWPSS, leading to  the formation of smooth islands among SWPSS (figures 2(c), (d) and 4). Previously, only microporous structures featured by a density of nanodroplets and cavities have been produced on Sn via fs laser ablation in ethanol [40]. Besides enriching the SWPSS diversity, this work also diversifies the variety of Sn surface structures, which have the potential to be applied for wettability control [41,42] and underwater bubble manipulation [43][44][45].

Triangular and rhombic LIPSS on W, Mo, Ta and Nb
Eight other kinds of metals, including Al, Zn, Ti, Zr, W, Mo, Ta, and Nb were ablated by FLAA at the laser power of 2 W to check the structure difference. Normal LIPSS  5(h)), which formed among traditionally longitudinal LIPSS. The period of normal LIPSS was ∼650 nm, while the period of triangular LIPSS was ∼760 nm. High spatial frequency LIPSSs were generated on a Ti surface with a period of ∼380 nm (figure 5(c)). UHSFLs [8] with periods of tens of nm are located in the trenches of LIPSS on the ablated Nb sample (figure 5(i)). Because surface melting and resultant Marangoni bursting has been proven to be necessary for the formation of LIPSS [46] and UHSFLs [8]. It was deduced that gentle or cold melting occurs to ablated metals during FLAA even though the melting temperature of Mo, Ta, and Nb are as high as 2620 • C, 3017 • C, and 2477 • C, respectively.
We wondered whether the formation of triangular and rhombic LIPSS strongly depends on the laser and processing parameters. Choosing the Mo substrate as a representative, FLAA was performed with a laser power of 5 W, scan intervals of 5, 10, 15, and 20 µm, and two scan methodologies (single-scan and double-scan). Figures 6(a)-(c) show that triangular LIPSS formed among normal LIPSS (figure 6(b)) for the Mo-5 W-5 µm sample. Increasing the scan intervals to 10, 15, and 20 µm also enabled the formation of rhombic LIPSS (figures 6(d) and (f)) and triangular LIPSS (figure 6(e)). Samedirection double-scan (figures 6(g)-(i)) and cross-direction double-scan (figures 6(j)-(l)) are also capable of creating triangular and rhombic LIPSS among normal LIPSS.
Unlike uniform triangular and rhombic LIPSS prepared by circularly polarized laser and linear cross-polarized beams [27,31], the layouts of triangular and rhombic LIPSS are heterogeneous (figures 6(j)-(l)), indicating that the influential factor is confined to local regions and may vary dramatically in different regions. Compared to the cross-polarized beams, which only allow a processing parameter window to yield triangular and rhombic LIPSS [27], this work shows that their formation is independent of laser power, scan line interval, and scan methodology while adopting a single linearly polarized fs laser. But the homogeneity and the area of triangular and rhombic regions are not comparable.
Many mechanisms have been proposed for the formation of triangular and rhombic LIPSS. One is the non-linear convection flow [31] triggered by strong temperature gradients. Another example is the hexagonal convection flow matrix [33]. If this mechanism occurs, the active convection flows should be subjected to the irradiation of tens of pulses. As a consequence, the flow structure should be very complex like Sn-SWPSS (figures 2(a)-(d)) rather than leading to the formation of homogeneous LIPSS with sharp edges. Hence, the non-linear convection flow mechanism can be ruled out for the formation of triangular and rhombic LIPSS.  Figure 6(h) displays the possibility to generate split LIPSS, which has also been discovered on stainless steels treated by fs laser ablation [47]. It is deemed that the localization of the incident laser light in the protuberances of LIPSS leads to the formation of split LIPSS [47]. Hence, the triggering of extra electric fields or SPPs upon the interaction of the incident light and the existing LIPSS should take place. Hence, besides the SPPs responsible for the formation of LIPSS, two other local SPPs in the other two directions should also be generated [26], which promote the formation of triangular and rhombic LIPSS. Table 3 lists the melting temperatures of the transition metals used for our experiments. It is clear that triangular and rhombic LIPSS mainly formed on the Ta/Mo/Nb transition metals with melting temperatures between 2000 • C and 3000 • C. Triangular and rhombic LIPSS were seldom found on the W sample. In our previous report, we performed FLAA on W with different scan intervals at a laser power of 5 W [48]. Only split LIPSS was produced (similar to figures 6(g) and (h)) because of a strong coupling of the incident laser and the surface plasmons in high-density plasma states [47]. Laser ablation of metals normally induces a significant increase in the surface temperature [49]. Metals with high meltingtemperatures are resistant to high temperature increases during FLAA, so it is possible to trigger the 'metasurfaces' functions of LIPSS, such as linear-to-circular polarization transition. In contrast, metals with lower temperatures may become melted and restructured by the following incident pulses or may be subjected to a high oxidation rate and nucleation/growth of metal oxide crystals. That is why porous LIPSS rather than smooth LIPSS is attained on Zn surface. Ishikawa et al reported that ZnO particles produced by laser ablation in water start growing at the temperature of 60 • C [50], which is easy to obtain during FLAA [51].
LIPSS obtained by FLAA of metals are normally composed of both pure metals and metal oxides [48] with the latter being the dielectric component. Although metal and dielectric components are not layer-by-layer constructed like metasurfaces [52], it is speculated that they should function similarly to metasurfaces (figure 7). One function of metasurfaces is linear-to-circular polarization transition, which is realizable using dielectric gratings with periodic metallic strips [53] or metallic zigzag arrays [54], analogous to LIPSS structures presented in this work. Hence, it is deduced that existing LIPSS may act as a linear-to-circular polarizer to trigger the formation of triangular and rhombic LIPSS upon successive laser irradiation. Triangular LIPSS can only be produced by adopting optimized laser fluences and scanning speeds [29]. Unoptimized conditions produces a mixture of normal LIPSS and triangular LIPSS [29]. In our experiments, laser power and scanning speed were unoptimized. The scanning speed and repetition rate were set at 200 mm s −1 and 400 kHz, respectively, which means there were two pulses in every micron distance and a very high overlap ratio of pulses. Since each laser pulse can induce the formation of LIPSS [55], it is highly likely that the existing LIPSS structures interact with the incoming light to change the optical field properties and result in the alteration of LIPSS's layout. Anisotropic LIPSS structures have already been used to tune light polarization by altering the scattering of orthogonal and parallel polarized light [56]. Hence, the modulation of the linearly polarized light into an azimuthally/circularly polarized light by LIPSS takes place. No matter whether single scan or double scan methodologies are adopted, the mixture of parallel and triangular LIPSS are achievable, as shown in figure 6. If the linear-to-circular transition really takes places, the chance of this happening should be rare and confined to a local region, which can be deduced from very limited areas of triangular and rhombic LIPSS and the randomness of their layouts.
Another formation mechanism that may induce the formation of triangular and rhombic LIPSS is the interference of surface plasmon waves with incident light, which is capable of producing crosshatch structures upon laser irradiation of square or triangle Au metal structures [57]. This work and another work published by Jia et al [47] explore the formation of split structures via SPP enhancement. The ejected or deposited particles irradiated by laser pulses is another factor that can induce the formation of bent LIPSS [58] by enhancing the electric field [59]. However, this factor can only induce random bending of LIPSS, but it cannot induce the triangular LIPSS arrays shown in figure 6(g). Other than split and bent LIPSS, nanocavities and nanotrenches on the ablated Ti surface indicate the complexity during FLAA of Ti, which may be associated with the third harmonic generation [60].  Figure 8(a) displays the reflectance of an unablated Sn sample and Sn-SWPSS samples produced by FLAA at laser powers of 2, 5, and 8 W. It is clear that laser ablation can significantly reduce reflectance, so the ablated surfaces are good candidates for antireflective applications [8,24,61]. The antireflective performances of ablated surfaces depend on the structure's properties. Larger and wider Sn-5 W-SWPSS (blue line in figure 8(a)) are more antireflective than smaller and shallower Sn-2 W-SWPSS (red line in figure 8(a)), which is due to the higher trapping ability for UV-to-NIR broadband light by deeper SWPSS trenches. This deduction is confirmed by the superior antireflectance performance of Sn-8 W-SWPSS (green line in figure 8(a)) than that of Sn-2 W-SWPSS even though Sn-8 W-SWPSS is covered by molten layers (figures 2(e) and (f)). Figure 8(b) shows the reflectance of all nine samples prepared by FLAA at a laser power of 2 W. Even though the emergence of triangular and rhombic LIPSS can increase the surface areas of Mo-, Ta-, and Nb-LIPSS, their antireflectance performances are not the best. The best UV-to-NIR ultrabroad antireflective surface is Ti-, while the best visible-range antireflective surface is Zr-LIPSS, which may be due to the formation of ZrO x in light of reflectance similarity [62]. Except the Zn sample, which has a distinct reflectance valley in the NIR range, the reflectance of most LIPSS/SWPSS surfaces increases as the wavelength increases. Hence, it can be concluded that the antireflective performance is a synergistic effect combing both surface composition and surface structure. The analysis of the effect of surface composition on antireflectance is beyond the scope of this manuscript, which will be done in another work.

Antireflectance and iridescence of SWPSS and LIPSS
LIPSS is a kind of refraction grating that enables the observation of various iridescence under different observation angles [7,41,63], so they are good candidates to be directly used as biomimetic colorful surfaces [6] or in anticounterfeiting applications [64][65][66]. In this regard, we characterize the iridescence of as-prepared structured surfaces under the illumination of white light, as shown in figure 8(c). The optical images of Sn, Zr, Zn, Ti, W, Al, Nb, Mo, and Ta samples taken at the capture angle of 45 • and illumination angles of 0 • , 15 • , and 30 • are shown in figures 8(d)-(l). Sn-SWPSS does not have iridescence because of multi-orientations and the existence of enormous pores, which can efficiently trap light rather than diffract light. Furthermore, Zr, Zn, and Ti samples do not possess iridescence (figures 8(e)-(h)) due to their high roughness and structural irregularity (figures 5(b)-(d)). Reducing the roughness of Al-LIPSS ( figure 5(a)) revealed a weak iridescence ( figure 8(i)). The weak iridescence of W-LIPSS is due to the irregularity of LIPSS and the existence of smooth regions ( figure 9(a)). The formation of smooth regions originates from insufficient ablation. That is, very high melting temperatures of W only enables a partial melting during FLAA at a laser power of 2 W. Our previous report demonstrated the possibility of inducing complete ablation via FLAA at 5 W to achieved a good iridescence on W [48]. Interestingly, the iridescence of Nb, Mo, and Ta follows the sequence of melting temperatures shown in table 3. Since iridescence is greatly dependent on the regularity and uniformity of LIPSS [67], it is reasonable to deduce that the regularity of LIPSS follows the sequence of Ta > Mo > Nb, which is confirmed by their structure morphologies (figures 9(b)-(d)). LIPSS has been verified to be strongly dependent on the surface melting [18,46], so the melting temperature of the ablated materials should be a critical factor that determines the states of LIPSS.

Conclusion
In summary, this work demonstrated the creation of unique SWPSS on Sn and produced triangular and rhombic LIPSS on W, Mo, Ta, and Nb by FLAA. Increasing the laser power from 2 W to 5 W significantly increased the periods of SWPSS from 4-6 µm to 15-25 µm, thus, allowing us to achieve the largest SWPSS ever reported. The generation of different sizes of molten Sn droplets and their coalescence/spread/ solidification together with Marangoni convection flows are considered to be the driving force behind the formation of SWPSS patches with each patch having its own SWPSS orientation. Further increasing the laser power produced an ultrastrong thermal effect with a longer lifetime of molten layers, which will solidify on the SWPSS. The formation of triangular and rhombic LIPSS is dependent on the material property, independent of the processing parameters. Our analysis of multiple LIPSS's iridescences, surface morphologies, and metal melting temperatures, suggests that transition metals with melting temperatures higher than 2000 • C are excellent substrates to be endowed with triangular and rhombic LIPSS, which form via linear-to-circular polarization transition mechanism upon the impact of a single linearly polarized fs laser. The SWPSS and LIPSS containing certain percentages of triangular and rhombic LIPSS are good candidates for antireflective applications. Mo and Ta LIPSS surfaces possess good iridescence with the potential to be used for anti-counterfeiting applications.