University of Birmingham Subwavelength direct laser nanopatterning via microparticle arrays for functionalizing metallic surfaces

. ABSTRACT Functionalized metallic nano-features can be selectively fabricated via ultrashort laser processing, however the cost-effective large-area texturing, intrinsically constrained by the diffraction limit of light, remains a challenging issue. A high-intensity near-field phenomenon that takes place when irradiating micro-sized spheres, referred to as Photonic Nanojet (PN), was investigated in the transitional state between geometrical optics and dipole regime to fabricate functionalized metallic subwavelength-features. Finite element simulations were performed to predict the PN focal length and beam spot size, and nano-feature formation. A systematic approach was employed to functionalize metallic surface by varying the pulse energy, focal offset and number of pulses to fabricate controlled array of nanoholes and to study the generation of triangular and rhombic Laser-Induced Periodic Surface Structures (LIPSS). Finally, large-area texturing was investigated to minimize the dry laser cleaning effect and improve homogeneity of PN-assisted texturing. Tailored dimensions and densities of achievable surface patterns could provide hexagonal light scattering and selective optical reflectance for a specific light wavelength. Surfaces exhibited controlled wetting properties with either hydrophilicity or hydrophobicity. No correlation was found between wetting and microbacterial colonization properties of textured metallic surfaces after 4h incubation of E. coli. However, an unexpected bacterial repellency was observed.


ABSTRACT
Functionalized metallic nano-features can be selectively fabricated via ultrashort laser processing, however the cost-effective large-area texturing, intrinsically constrained by the diffraction limit of light, remains a challenging issue.A high-intensity near-field phenomenon that takes place when irradiating micro-sized spheres, referred to as Photonic Nanojet (PN), was investigated in the transitional state between geometrical optics and dipole regime to fabricate functionalized metallic subwavelengthfeatures.Finite element simulations were performed to predict the PN focal length and beam spot size, and nano-feature formation.A systematic approach was employed to functionalize metallic surface by varying the pulse energy, focal offset and number of pulses to fabricate controlled array of nanoholes and to study the generation of triangular and rhombic Laser-Induced Periodic Surface Structures (LIPSS).
Finally, large-area texturing was investigated to minimize the dry laser cleaning effect and improve homogeneity of PN-assisted texturing.Tailored dimensions and densities of achievable surface patterns could provide hexagonal light scattering and selective optical reflectance for a specific light wavelength.

Surfaces exhibited controlled wetting properties with either hydrophilicity or hydrophobicity. No correlation was found between wetting and microbacterial colonization properties of textured metallic
surfaces after 4h incubation of E. coli.However, an unexpected bacterial repellency was observed.
However, a preliminary step is required to apply the PN technology for surface texturing, i.e. the deposition of a microspheres' monolayer on a substrate.Self-assembly processes such as Langmuir-Blodgett techniques can be applied for microspheres' deposition into hexagonal close-packed (HCP) monolayers, also called contacting particle-lens arrays (CPLA) [20].Then, a laser-irradiation of CPLA results in an array of non-evanescent beams emerging from the shadow-side of microspheres, with diameters as small as the third of the irradiated wavelength and a length of two wavelengths [15].Such a method was applied to laser-fabricate subwavelength holes and convex bumps [21,22], even on materials with low absorption at a given wavelength (λ) [23,24].Various attempts to tailor the PN optical properties were reported, including changing the refractive index (n) and the shape and diameter (2r) of microspheres and also the illumination conditions [15,[25][26][27][28].
Microparticle Arrays for Functionalizing Metallic Surfaces".Submitted on 13/09/2018, Revised on 01/02/2019, Accepted on 07/02/2019.However, large-area texturing for functionalizing surfaces by direct laser ablation remains a challenge.This is mostly due to the detachment of microspheres after any single laser pulse irradiation and this could be explained with the substrate ablation or dry laser cleaning (DLC) effect [29][30][31][32].Contrary to laser cleaning where the laserinduced marking is an undesirable side-effect, to achieve large-area PN texturing the marking must be maximized while minimizing the microspheres' detachment.Several techniques were proposed to address this issue, usually by preserving CPLA by covering it with a transparent material [33][34][35][36].Maintaining CPLA in contact with the substrate allowed multi-pulse processing and thus to texture larger areas homogeneously and at the same time to increase the aspect ratio of the resulting nanoholes [36].
In this research, a novel approach for submicron large-area texturing of metallic surfaces is presented and their multifunctional aspects are described with applications in nano-scale optics, wettability and microbiology.Especially, a near-infrared ultrashort laser irradiation of transparent silica microspheres deposited onto stainless steel plates was investigated.The microspheres were used as CPLA and their sizes were equal to the irradiation wavelength.FE simulations were conducted to predict the PN focal length and the full width at half maximum (FWHM) spot size at the substrate surface while varying laser wavelengths and spheres' radii.Direct laser nanofabrication of nanobumps, nanoholes and LIPSS was investigated and was compared to DLC effect towards large-area texturing.Finally, potential surface functionalities of nanoholes and LIPSS were studied for selective hexagonal scattering, tunable optical reflection, wettability, and microbiology applications.

Sample preparation and laser set-up
Silica microspheres were deposited in a hexagonal close-packed monolayer onto X6Cr17 stainless steel plates by employing a patented method [37] based on the Langmuir-Blodgett technique.The laser irradiation was performed with a near-infrared ultrashort pulsed laser source (Satsuma, Amplitude Systemes) with a pulse length of 310fs and a central wavelength of 1032nm.The laser beam had a circular polarization and was steered over the substrate using a 3D scan head (RhoThor RTA) equipped with a 100mm F-theta telecentric lens.The beam spot size, 2ω0 at 1/e 2 , was adjusted to 30µm diameter at the focal point with a beam expander.Spot sizes at focal offset were calculated following the relation ω(z)=ω0[1+(z/ZR) 2 ] 1/2 with ZR=πω0 2 /λ.Pulse fluences were estimated by dividing the pulse energy by the spot area πω 2 .For post-fabrication of LIPSS, the pulse-to-pulse distance (δ) is the ratio of the scanning speed over the repetition rate and the number of pulse equivalent was calculated according to the formula N=πω0 2 /(σ•δ), with σ the hatching distance.The remaining spheres and any debris resulting from the laser irradiation were removed using an ultrasonic bath in ethanol for 15min.

Optical characterization
A broadband light source (400-1100nm, Ocean Optics 2000+) and a spectrophotometer (resolution 0.1-100 FWHM, Ocean Optics HL 2000) were used to characterize the surfaces' optical properties.Broadband light normally illuminated the metallic nanopatterns and the reflected light was captured though an objective (x20) in reflection mode.Near-field hexagonal scattered/diffracted light was measured through a hemispherical screen (semi-transparent white, 30mm in diameter).Far-field scattered/diffracted rainbow patterns were performed with an image screen (white A4 paper) setup.

Wettability
The wetting properties were analyzed employing an optical tensiometer (Attension Biolin Scientific Theta T2000-Basic+) and purified (Milli-Q, Merck) water was used as probe liquid.All tests were performed under ambient conditions.Static contact angles (CA) of 6μl drops were measured after stabilization employing the sessile drop technique.The CA hysteresis was calculated using the needle-in-drop method and measuring the advancing and receding CA evolution of 6μl at 0.5μl/s.

Microbiology
The textured metallic samples were replicated by a two-step process.First, negative surface structures were obtained from a master with an epoxy resin (Araldit CY212, Plano) via thermal polymerization.Second, the final replicas were produced by employing the epoxy resin master with a perfluoropolyether prepolymer via UV induced polymerization (Fluorolink MD40, Solvay).Bacterial colonization and bacterial adhesion on textured substrates were assessed with E. coli (strain W3110), after 4h incubation in lysogeny broth media (LB, Sigma-Aldrich) at 37°C (overnight culture adjusted to optical density (OD) ~0.4 in fresh LB before).The bacteria were fixed in paraformaldehyde (PFA), then rinsed with phosphate buffer solution and distilled water (milliQ, Merck) and gently dried with nitrogen.Subsequently, the samples were sputter coated with a 10nm gold layer (SCD 050 sputter coater, Balzers) and imaged with a SEM XL30 (ESEM/FEG, FEI).The adherent bacterial cells were counted on SEM images using the cell counter tool in ImageJ (NIH).The experiment was performed two times and a total of 14 images (~150000µm²) per sample were examined.

FE and computational modelling
Computational modelling techniques for particle light scattering were developed, in particular by employing numerical simulation or analytical methods based on the generalized Lorentz-Mie theory [26,30] particles with much larger sizes than the wavelength and the focal spot sizes were calculated to be in the range from 0.8λ down to 0.15λ depending on the refraction indexes of the particles and the propagation medium [21,38,39].Depending on the ratio 2r/λ (x) three different cases could be distinguished.For x>>1, the generic rules of geometrical optics apply and the Snell-Descartes' law should be used to obtain the approximate focal spot size (wg) as follows [30]: wg≈r((4-n 2 ) 3 /(27n 4 )) 1/2  (1) For spheres much smaller than the wavelength, x<<1, the close-packed monolayer behaves as a densely homogenous one and the dipole approximation estimated the spot sizes (wd) in the dipole regime similar to the sphere radii [30], i.e.
wd≈r, which is comparable with some reported simulation results [40].Seldom, the intermediary range between geometrical and dipole regime, i.e. x≈1, was considered and so the focal spot size could be estimated between 0.14λ and 0.40λ [39,41,42].For this third case, Arnold [30] introduced a constant of transition C=((n 2 -1)wg)/((n 2 +1)r) thus unifying the analytical approximation of geometrical and dipole focal enhancement so that the focal spot size (w) could be approximated as follows: where k=2π/λ is the electromagnetic wave number.
To understand the near-field focal enhancement via microspheres, FE based computational modelling was performed in COMSOL Multiphysics (v5.3a).A planar wave (1030nm) propagates through an air medium and focused through silica microspheres.To take into account optical interferences of the spheres, a close-packed 2D array of 5 spheres was considered.The continuity and scattering boundary conditions were considered for the spheres and the domain boundaries.The diameter 2r of the microspheres was varied, and respectively, the domain of simulation (8r x 10r).
Near-field electric field intensity distributions upon normal irradiation of microspheres of different sizes were investigated by varying x from 0.5 to 8. The microspheres are considered as a homogeneous, non-absorbent medium with refractive index of 1.50 at 1030nm wavelength [43].The near-field focused light intensity as a function of microsphere size (x=1.0,1.9, and 2.9) are depicted in Figure 1a.The increase of microspheres sizes of led to an increase of the focused light (D) and PN length (L).For periodic microsphere arrays, the maximum focused light intensity was observed at the center due to the Gaussian distribution nature of the illuminated plane wave.
The resulting PN was analyzed by means of peak intensity (Imax), paraxial focal length (D) and beam spot size (FWHM at Imax).The PN length (L) was estimated as the distance from the substrate surface until the intensity reaches Imax/2 as shown in Figure 1a, by plotting the evolution of intensity along the vertical axis (Figure 1b).The intensity and beam spot size at the point of contact between the microspheres and the substrate was considered, too. Figure 1c showed peak intensity of PN as a function of normalized microsphere size at maximum and surface position.A pre-condition for the near field optical phenomena associated with PN is the electric field propagation along the vertical axis to reach a peak intensity outside the microsphere.Based on the simulation results this pre-condition is met at x≈1.5 and x larger than 2 (see Figure 1d).Therefore, the PN geometrical optic regime is considered established for x strictly larger than 2. For this regime, the relative peak intensity increased with the size of the sphere (see Figures 1cd).However, while the peak intensity is reached at the substrate surface when x=2, a further increase of x led to a reduction of the surface peak intensity compared with the max values, especially up to approximately 50% for x>6.Another well-established PN trend depicted in Figure 1d is that both the focal point and the PN length decrease with the decrease of microspheres' sizes [19,44] until the focal point becomes coincident with the contact point between the microspheres and the substrate for x=2.
Interestingly, the simulation illustrated that a further decrease of x led to an increase of the focal distance, when x=1.5, before a rapid decrease down to negative values where the peak intensity was reached inside the microspheres.This was attributed to a transition state between the geometrical optic and dipole regimes.For the dipole regime (x<<1), electrical fields are predicted to reach a peak intensity not anymore along the vertical axis but sideways [40].The FE simulation exhibited local intensity peaks shifted sidelong the paraxial axis for the transition state of x≈1 and the focal point is close to the microsphere's center.Half maximum intensity was reached outside the microsphere, at a paraxial distance of ~0.2λ, which is considered as a PN tail.
Furthermore, considering characteristics of Gaussian intensity profiles, the PN tail reached an intensity of Imax/e 2 at a paraxial distance of approximately 0.7λ for x≈1.Even if the peak intensity is not reached on the substrate, the intensity at the surface may still exceed the ablation threshold for a given material, opening the way for PN-assisted manufacturing applications.A subwavelength-size damage may appear on substrate surface when the PN energy density is sufficiently high while the PN beam size is smaller than the irradiating wavelength.The analytical approximation for the focal spot size in Eq. ( 2) converted to FWHM by a factor of (2ln2) 1/2 correlated well to the simulation of PN beam size for x lower than 1.5, with FWHM smaller than 0.5λ.Although the analytical model predicts a constant increase of beam size with the increase of x, the FWHM reached a plateau at around 0.5λ in the FE simulation, in particular for x larger than 1.5 (see Figure 1e).A similar stagnation was also reported by McCloskey et al. [26].While the PN FWHM appeared consistently in the subwalength range, with a minimum at ~0.2λ, the FWHM at the surface position remained in the subwavelength range only for x from 0.5 to 4.
Therefore, when PN are used in direct contact with substrates, two fabrication regimes can be discussed.At first regime, an increase of microspheres' sizes inherently leads to larger surface beam sizes and hence a lower intensity at the surface.This effect may be comparable to performing laser processing out of focus, but only to some extent, as any change in the PN beam shape may induce different surface damage morphologies.The simulation results' trend for microspheres much larger than the wavelength used in contact mode may confirm experimental observation of volcano-shaped surface damages [22,45].Several studies investigated how to bring the focal point back to the surface, with a vertical offset of the microspheres [42,46,47] or a non-orthogonal incident irradiation [40,48,49], and this aspect is therefore not considered further in this study.At second regime, the irradiation of particles smaller than the wavelength leads to FWHM at the surface position that are in the subwalength range, but with the lowest peak intensity and PN length.Thus, this constrains the fabrication process to low aspect ratio nanofeatures and therefore should be investigated further.
Employing CPLA for large area texturing, the HCP nature of the deposited monolayer induces a surface damage with hexagonal periodicity equal to the microspheres' size.However, the dimension of the surface damages compared to their periodicity has to be considered.The irradiation density, i.e. the ratio of the area covered by PN spots with intensity larger than half maximum over the whole substrate, can be used as an indicator for the achievable feature sizes and periodicity (Figure 1f).A general trend between the analytical and simulation results was observed, the irradiation density increasing with the decrease of x.The analytical approach predicts an asymptotic minimum at ~2% for very large x and a maximum irradiation density of ~30% can be achieved as x tends to 0. Contrarily, the simulations revealed a local maximum of ~20%, reached when x is equal to 1.There are similar trends in approximation and simulation results but there are discrepancies, too, because the analytical approach does not take into account whether the beam spot sizes are at the substrate surface or not.For PN spots on the surface, the density is also increasing for lower x.The ratio FWHM/2r reached more than 50% when 2r is smaller than λ, indicating that the density achievable with the PN technology can be sufficient for large-area sub-wavelength surface texturing when using CPLA with microspheres smaller than the wavelength.
In this research the case when x is equal to 1 is selected to investigate the PN texturing process experimentally.Especially, this PN processing case is used to validate the technology for producing dense nanoscale features while the irradiation conditions are far from optimum.In particular, these PN texturing conditions lead to the lowest peak intensity and intensity at the surface (Figure 1c), and a PN tail of a few hundred nanometers (Figure 1d), however with a relatively high irradiation density with subwavelength-sized FWHM in a λ-periodic HCP (Figure 1f).

Subwavelength laser ablation
Ultra-short laser pulses in the near-infrared (NI) range (310fs, 1032nm) were used to irradiate a dense monolayer of 1µm-diameter SiO2 spheres deposited onto stainless steel substrates (Figure 2a).The laser beam was steered over the surface using a galvanometer scanner and was focused with a 100mm F-θ-lens.The NI irradiation with a 30μm beam spot creates an array of PN tails that can lead to surface damage when the pulse energy is sufficiently high, resulting in the ejection of parts of the CPLA (Figures 2b-c).Each PN resulting from such CPLA irradiation can be considered as a Gaussian nanobeam, with FWHM and 1/e 2 spot sizes of 0.46λ and 0.74λ, respectively.
Various surface morphologies could be achieved depending on pulse energy and pulse numbers.A selection of PN-processed surfaces is presented in Figures 2d-g.As a result of a repeated irradiation with a pulse energy close to the damage threshold, convex nanobumps were fabricated (Figure 2d).Huang et al. attributed the generation of such surface features to the Marangoni convection induced by PN sidelong the paraxial axis and achieved λ/2-diameter nanobumps on silicon wafers by irradiating a 1-µm-diameter CPLA with UV-nanosecond laser pulses [50].In this research, slightly smaller nanobumps, in particular with diameters and heights of ~λ/3 and ~30nm, respectively, were generated with near-infrared (NI) 310fs pulses.While a single shot was sufficient to create the nanobumps with nanosecond pulses, multiple irradiations, in particular 100 pulses at 10kHz, were required to produce them with femtosecond pulses, and apparently as a result of some cumulative effects.Beyond the damage threshold of the substrate, the surface was locally ablated under the microspheres and an array of nanoholes were generated.First, a surface damage was observed with single pulses of 0.9µJ at the focal point and then up to 1mm focal offset.The nanoholes were circular in shape with a distance between each other corresponding to CPLA periodicity.The minimum diameter was ~λ/6, i.e. ~175nm.The increase of pulse energy resulted in larger holes' diameters, up to 1µm, and thus to form honeycomb structures.For positive or negative surface structures, the aspect ratio remained limited at ~1:10, as depicted by two representative AFM cross-sections in Figures 2k-l.The microspheres' sizes were close to the irradiation wavelength (x≈1) in this research and these processing conditions differ from those used by other researchers for direct NI femtosecond nanopatterning.In particular, wavelengths in the range from 400 to 800nm were utilized to fabricate nanoholes of ~0.3λ diameter and several 10nm in depth on silicon and glass substrates [40,51].
The intensity distribution of the laser beam had a direct influence on the homogeneity of fabricated nanoholes' arrays.A single pulse could produce an array of ~260 nanoholes with diameters from 255 to 940nm (Figure 2h).An increase of pulse energy results in ~730 nanoholes with similar average diameter but with a broader dispersion, from 120nm to 1.23μm (Figure 2i).Such high pulse energies were detrimental to the nanoholes' homogeneity and this was mostly due to the laser beam Gaussian intensity profile.PN induced thermal-effects after irradiations with femtosecond laser pulses were reported on glass [51], silicon wafers [45,52] and metallic substrates [52].Especially, they led to the fabrication of nanoholes larger than the CPLA periodicity and thus clogging contiguous nanoholes.However, by tailoring the focal offset, larger CPLA areas could be irradiated and also the use of laser beams with a top-head Gaussian profiles could enable the fabrication of nanoholes with much smaller spread of diameters.In this work, arrays of nanoholes were produced with a focal offset up to 2mm.Especially, it was possible with the use of 1mm focal offset to produce ~1700 nanoholes with an average diameter 310nm and a spread from 160 to 780nm (Figure 2j).offset and 100 irradiations at 1.3µJ; array of nanoholes fabricated with a single irradiation of (e) 1.3µJ at focus, (f) 6.3µJ at focus and (g) 6.3µJ at 1mm focal offset, respectively.(h-j) Distributions of nanoholes' diameters corresponding to (e-g).The nanoholes are gathered in clusters of 0.1µm and a normal distribution is extrapolated from the distribution.Profiles of nanobumps (k) and honeycomb structures (i) are given with two representative AFM cross sections.

Large area texturing
Parallel nanostructuring refers to the fabrication of nanofeatures over an area significantly bigger than a single-pulse laser spot.The large area covering with PNinduced nanofeatures is intrinsically limited due to the detachment of microspheres after any single laser pulse irradiation.This is similar to the effect referred to as Direct Laser Cleaning (DLC) that was investigated by many researchers both empirically [31] and theoretically [30].Especially, DLC is the result of a mechanical ejection triggered by the rapid thermal expansion of the irradiated microspheres.The evaporation of condensed humidity at the interface sphere-substrate may also increase DLC effect [32].
DLC takes place below the ablation threshold and therefore only cracks and delamination of the CPLA could occur without any surface damage, until a sufficient fluence is reached to eject the microspheres.The reported ejection velocities [53] are in the order of 10m/s and this could explain why there was no increase of nanoholes' aspect ratios with the pulse repetition rates investigated in this research.A pulse energy of 2µJ and processing with 1.5mm focal offset were required to start cracking the microspheres' monolayer.A further increase of pulse energy up to 4.8µJ resulted in a partial removal of microspheres without damaging the substrate surface.At the same time laser processing closer to the focal point and also with higher pulse energies led inevitably to surface damage.The CPLA delamination and ejection of microspheres can be attributed to ablation dynamics, especially the removal of microspheres above nanotextured areas.
A focal offset of the incident laser beam was beneficial for fabricating more homogeneous array of nanoholes, however the tail of the Gaussian beam broadens the area affected by DLC.A multi-pulse approach had the same DLC effect, possibly due to a cumulative effect.Similarly, a higher pulse energy increased not only the thermal effect and nanoholes' sizes but also the DLC area as shown in Figure 3a.The filling ratio, in particular the ratio between the surface area covered with nanofeatures and the total processed area, was optimized by applying various scanning strategies, especially by varying the pulse-to-pulse distance to achieve moon-shaped arrays of nanoholes, as shown in Figure 3b.For example, the maximum filling ratio of 46% was reached with 3.6μJ pulses at focus and a pulse-to-pulse distance slightly smaller than the laser spot diameter (Figure 3c).A further reduction of pulse-to-pulse distance led to non-circular arrays of nanoholes and decreased the filling ratio to ~30% (Figures 3d-e) until the LIPSS outset threshold was reached and PN-induced nanoholes were converted into rippleslike LIPSS (Figures 3f-g).

LIPSS generation with tailored periodicity
Further irradiation of nanoholes fabricated with single-pulses with fluences in the range from 0.18 to 0.29J/cm 2 , as shown in Figure 2g, did not lead to additional material removal but resulted in some improvements of nanoholes' rims, as already reported [24,54].The irradiation intensity was sufficient to displace the microspheres from their initial locations and thus to prevent the fabrication of nanoholes with higher aspect ratio unless preventing such displacements [36].Albeit the fluences used for nanoholes' fabrication were not sufficient to reach the substrate's ablation threshold after the CPLA removal, multiple irradiations led to surface structures referred to as LIPSS [14].Cumulative fluence of ~1.8J/cm 2 was sufficient to trigger the generation of LIPSS over the arrays of nanoholes.Further cumulative effects after reaching ~2.9J/cm 2 were the spreading of LIPSS over the whole surface and thus to join the array of nanoholes.At a fluence of ~11J/cm2 the formation of triangular surface morphologies was initiated (Figure 3h) and, compared with the generation of such triangular LIPSS without CPLA, only half of the necessary cumulative fluence was required [6].Thus, prefabricated surface patterns could be used to lower the cumulative fluence required for LIPSS generation.
Despite LIPSS being wavelength-dependent periodic surface structures, the 1μm periodicity of the CPLA was retained.Various regular LIPSS morphologies were generated by varying pulse numbers and fluence, in particular triangular and rhombic  3h-i), which could be of interest for a single-step LIPSS fabrication.The periodicity remained consistent with the pre-fabricated array of nanoholes until the cumulative fluence was sufficient to break down the established λ-periodic triangular features into ~0.82λ-periodicripple-like LIPSS (Figure 3j).Considering the case in Figure 3h (λ periodic), such triangular LIPSS with ~0.87λ periodicity were achieved with similar irradiation conditions where the triangular surface morphology was attributed to the use of a circular-polarized laser beam [6].A pre-fabricated array of nanoholes can potentially enable the generation of triangular nanofeatures with varying sizes and periodicity.

Optical properties
The sub-wavelength features fabricated on stainless steel were characterized through light illumination in reflection mode.A broadband light source was used to illuminate the textured surfaces at angles from 0 to 180° by controlling the rotation with a step motor.A spectrophotometer was positioned normal to the patterned surface for intensity measurements within the visible spectrum, i.e. a wavelength range from 400 to 700nm.
The optical response of stainless steel surfaces fabricated with different pulse energies, pulse numbers and effective pulse-to-pulse distances were studied.A 0.9μJsingle pulse irradiation resulted in a surface morphology with reflection properties close to 100% with only a slight decrease to around 85% in the violet-blue range (Figure 4a).diffraction patterns.Similar 2D-FFT and diffraction patterns were previously reported for hexagonal arrangements of microstructures [6,56].In agreement with 2D-FFT of nanoholes' arrays, hexagonal scattering patterns were observed in the surrounding of the non-diffracted broadband light (zero-order) (Figure 4g).Since the diffraction angle increases with the wavelength, multiple rainbow patterns were visible covering the range from blue to red.Analogously to diffraction gratings, the diffraction distances from the center spot were larger for the red light compared with the blue one, accordingly to Bragg's law (λ/Λ=2sinφ, where λ is the considered wavelength, Λ the grating's spacing, φ the angle from the surface plane) [57].The angle-resolved far-field images indicated that PN-assisted nano-fabrication enables broad-range (up to θ=70°) diffraction patterns (Figure 4h).However, lower tilt angle (θ=10°) illumination shows bright far-field diffraction compare with higher tilted angle illumination (θ=80°), due to omnidirectional light scattering from the edge of the micro-patterns at the bigger angle (θ=80°).

Wettability and microbiological properties
Wettability properties of nanopatterned surfaces were investigated with 6μl of water (n=3) after prior surface cleaning, with sonication in ethanol solution and air flow drying.While non-textured stainless steel had a static water contact angle (CA) around 90° (between hydrophobic and hydrophilic), nanotexturing had two opposite effects on wetting depending on the pulse-to-pulse distance used during laser processing (Figure 5a).If the process results in arrays of nanoholes, such as for σ=δ larger than 15µm, the static contact angle decreases leading to a hydrophilic surface, with CA down to 59°.If the process results in LIPSS the CA values increased, the surface became apparently hydrophobic, with CA of 101°.In addition to the static CA, dynamic contact angle measurements were conducted for σ=δ equal to 5μm.Although LIPSS seemed to be hydrophobic (increased static and advancing contact angle), the receding contact angle remained close to the reference receding CA (Figure 5b), which leads to sticking droplets on the surface due to the high CA hysteresis (Figure 5c).
In microbial experiments, samples with a structural transition from nanoholes towards LIPSS were compared regarding their bioadhesion properties.The experiments were performed with a lab strain of E. coli inhibition of bioadhesion [8].However, many nanotextured samples did not show an increased amount of initially adhered bacterial cells compared to the non-textured reference (Figures 5d-f).One sample type even exhibited a significantly lower bacterial colonization after 4h compared to unstructured surfaces (Figure 5e).In particular, this sample type processed at σ=δ equal to 10µm represents a transition from the hexagonally ordered nanoholes fabricated with larger pulse-to-pulse distances and selforganized subwavelength structures (LIPSS) fabricated with lower pulse-to-pulse distances (Figures 3c-g).While the bacterial assays did not show a similar monotonous trend as the static wetting measurements, interestingly, a strong increase in bacterial adhesion could be observed for the ripples-like LIPSS.SEM images in Figure 5f show that transition is accompanied by a smoothening of the surface, but periodic surface features are still visible, whereas the roughness on the LIPSS structure seems to be more pronounced.

Figure 1 .
Figure 1.The modelling results of light focusing through silica microspheres.Irradiation wavelength was fixed at 1030nm while the microspheres' radii were varied.(a) Intensity mapping for three cases of x=2r/λ, especially equal to 1.0, 1.9 and 2.9.The color scales are independent for each case, in particular from blue (low intensity) to red (peak intensity).(b) The intensity evolution along the vertical axis, normalized with the sphere's radius for four cases of x.The dashed circle and line indicate the normalized sphere's dimension and surface position.(c) Peak intensity estimations for the considered cases of x from 0.5 to 8. (d) Paraxial focal length and PN length as a function of x.The red colored area suggests combinations of wavelength and radius where the focal point is outside microspheres.(e) FWHM normalized in regard to the irradiation wavelength for the considered x values.(f) Irradiation surface density calculated for a close-packed array of spheres as a function of x.Estimations are taken at the peak intensity positions z=z(Imax) on the substrate surface.Analytical PN waist were plotted according to Arnold [30].

Figure 2 .
Figure 2. A direct laser ablation based nanopattering by irradiating a HCP CPLA of 1µmdiameter spheres with a NI 30µm-diameter Gaussian fs-laser beam.(a) The used experimental setup for processing substrates with a focal offset.(b-c) The schematic steps of the texturing process.The microspheres are partially removed from the textured area and its surroundings.Micrographs of morphologies achieved with 310fs laser pulses at 1032nm wavelength: (d) nanobumps fabricated employing a 2mm focal

Figure 4 .
Figure 4. Optical characterization of nanotextured stainless steel.The evolution of broadband reflection with the increase of pulse numbers with two different pulse energies (a) 0.9μJ and (b) 6.6μJ.(c) The reflection evolution for three selected wavelengths, 440nm (blue), 540nm (green) and 640nm (red), with pulse numbers (constant pulse energy of 4.8μJ).The evolution of reflection properties with the pulseto-pulse distance for (d) fixed σ=25μm and (e) σ=δ.(f) The reflection evolution for selected blue, green, red wavelengths, with the pulse-to-pulse distance σ=δ.(g) Light scattering properties under normal illumination of the surfaces, projected on a 30mmhemispherical semi-transparent surface.(h) Light scattering properties under different illumination angles, from 10 to 80°.
(W 3110) by incubation for 4h in nutrition rich media.Instead of the originally nanostructured stainless steel samples, identical polymeric replicas were prepared for the bacterial assays, which allowed us to keep the original surface clean and available for further studies.It was shown recently that structural periodicities close to the cells size, similar to the 1 µm period of the PNstructures, support bacterial surface colonization and are, therefore, detrimental to the Microparticle Arrays for Functionalizing Metallic Surfaces".Submitted on 13/09/2018, Revised on 01/02/2019, Accepted on 07/02/2019.

Figure 5 .
Figure 5. Wettability and microbiological characterizations of subwalength surface structures.(a) The evolution of static water CA with the pulse-to-pulse distance (σ=δ) on stainless steel.(b) Advancing and receding water CA for polished and σ=δ=5μm textured stainless steel.(c) A tilting test illustrating the sticking effect of water drops onto the σ=δ=5μm textured stainless steel.(d) The evolution of normalized adherent E. coli cells after 4h incubation on PFPE replicas with the fabricated pulse-to-pulse distance, for (d) σ=25μm and (e) σ=δ.Data are shown in box-whisker plots with half of all data points within the box and 100% within the whiskers; black diamonds in the boxes indicate mean values and the black horizontal line the median value (n=14).(f) Representative SEM images of adhered E. coli on replica of (from left to right): non-structured reference, σ=25µm and δ=22.5µm,σ=25µm and δ=15µm, σ=δ=10µm, σ=δ=5µm.The adhered E. coli cells are highlighted in green halos.
nanoholes and LIPSS.Surfaces exhibited efficient broad range light scattering (up to θ=70°).Surface reflection and scattering properties could be tailored and specific visible wavelength range could be absorbed.The wetting of the textured surfaces responded differently for array of holes than for LIPSS, showing a monotonous decrease of the static contact angle from 100° for LIPSS down to 60° for hexagonal-arranged nanoholes, correlating with the gradually change of process parameters.No similar comprehensive trend could be found in microbial adhesion tests, but there was a significant decrease in bacterial surface colonization after 4h on surfaces situated in the transition between nanohole and LIPSS structures.This could open a new pathway for future studies to evaluate the PN-assisted technology for inhibition of bacterial adhesion.