1. INTRODUCTION

Powerful high-energy laser systems increasingly impact renewable energy [1], directed energy [2], exploration of basic light–matter interactions [3], and compact and unique particle and irradiation sources such as x-ray sources for use with microscopy and medicine [4,5], neutron and proton sources for non-destructive testing and medical use [6], and electron emission for lithography [7]. Antireflective (AR) coatings serve the crucial roles of increasing the overall efficiency of the high-energy laser system while simultaneously reducing stray light to safely maximize the output intensity. However, traditional AR coatings consisting of multi-layer dielectrics (MLD) and sol-gel coatings have inherent limitations that restrict the applications that can be handled. MLD coatings, while typically environmentally stable, suffer low laser-induced damage thresholds by virtue of the multiple materials and material interfaces within the fabrication design. At 355 nm (7.5 ns pulse and 0.58 mm diameter), MLD coatings have a laser-damage threshold of ${10.6} \pm {9.3}\;{{\rm J/cm}^2}$ [8], while at 1064 nm (20 ns pulse, 1 mm diameter) MLD coatings have a damage threshold of ${28} \pm {3.5}\;{{\rm J/cm}^2}$ [9]. Sol-gel coatings, on the other hand, are generally very robust under high laser fluence but are limited by environmental stability [10], uniformity, and the range of optical materials they are compatible with—i.e., the need for index matching. At 355 nm (7.5 ns, 0.58 mm diameter), sol-gels have a damage threshold of ${37.7} \pm {9.3}\;{{\rm J/cm}^2}$ [8], while at 1064 nm (10 ns, 0.17 mm diameter as estimated by scaling to other known data), sol-gels begin to damage at roughly ${60}\;{{\rm J/cm}^2}$ [11].

An alternative approach proposed to avoid these traditional AR coating shortcomings is the incorporation of metasurfaces and nanostructured surfaces (NS) as AR coatings, aided by recent advances in surface fabrication technology [12–15], where these surfaces are defined as artificially structured surfaces with optical properties obtained by unit structure rather than the constituent material composition [16]. The unit structure may be used to modify different material characteristics such as the refractive index [17,18], thermomechanical properties [19], or electrical properties [20]. The similarity between these surfaces is that the properties are controlled by the structure rather than the material composition, with those NS features occurring either periodically [21] or randomly [12,22,23] on the surface. Optical nanostructured surfaces consisting of subwavelength elements decorating the surface are the focus of this work exploring the feasibility of using a NS for high laser power AR applications using fused silica as a test material, although the findings here could be extended to other transparent optical materials. Furthermore, engraving NS can have implications for flat optics, focusing lenses, etc., if the features can be spatially tuned [12].

Substrate engraving, as a method to generate subwavelength features, represents a significant advance to NS AR coatings as it yields features that are monolithic to the substrate, avoiding the introduction of additional materials or material interfaces. Consequently, a NS fabricated in this way may generate AR coatings with both an increased laser-induced damage threshold and environmental stability relative to traditional AR coatings. To realize the possibility of using substrate-engraved nanostructured surfaces for AR large-aperture high-energy or high-peak-power laser applications, there needs to be a fabrication process capable of generating subwavelength features on the large scales required by high-power laser systems, such as optics up to the ${\sim}{0.5 - 1.0}\;{\rm m}$ scale utilized on the National Ignition Facility (NIF) [24]. However, the ability to generate subwavelength features via standard milling processes, i.e., e-beam, focused ion beam milling, and deep ultraviolet processes, becomes impractical with increasing optic aperture size, and can be considered economically unfeasible for meter-size optics. To overcome this limitation, structures randomly distributed across surfaces have been harnessed for AR applications (rAR). Fabrication of rAR may be obtained through dry-etching processes that either take advantage of the creation of in situ nucleation centers [25] or utilize a fabricated mask to guide the etching process [26]. Recently, solid state diffusional dewetting has shown to be a tunable process capable of generating uniform arrays of nanoparticles to function as etching masks [27]. Etching masks open the possibility to deterministically etch features with controlled design, such as more control over the depth profile of the index of refraction. The ability to control sidewall slope, i.e., cylindrical features versus right circular cones, presents an opportunity to tune the mechanical properties of the generated surface, as the fragile nature of needle-like structures when loads applied to the pillar tops induce torque is minimized with cone structures.

For surfaces with subwavelength embedded structures, the question is open on whether they are inherently more susceptible to damage compared to conventional AR coatings due to potential field intensifications at asperities or sharp features. In addition, the mechanical stability could also become an issue when manually handling optics with NS in general. In this work, we present, to the best of our knowledge, a first demonstration of a dewetting-assisted dry-etching process to illustrate the ability to design a NS to act as AR and show the durability of these structures relative to conventional AR coatings. We show the reflectance of dewetting-assisted NS at the fundamental and third harmonic wavelengths at the NIF (${1}\omega$ and ${3}\omega$, 1053 nm and 351 nm, respectively) and compare it to the predictions based on transmission matrix calculations as a function of nanostructured surface fabrication parameters. Short pulse (small beam) laser-damage performance at ${1}\omega$ and ${3}\omega$ is reported for these nanostructured surfaces, and a new mode of damage is discussed that, to our knowledge, has not been reported on before. These results answer fundamental questions about the damage mechanisms of surfaces composed of nanoscale features and verify that unintended gold diffusion into the substrate bulk during dewetting is not an issue. Results from mechanical testing via static indents are discussed to address the mechanical robustness and AR response under mechanical loading of these surfaces, demonstrating the ability of this method to tailor the mechanical and optical properties for durable optical nanostructured surfaces. This has advantageous implications for the broader metaoptics community.

2. RESULTS AND DISCUSSION

Nanostructured surfaces generated via dry etching through a self-assembled dewetting mask may be described by a few key features, such as the pit depth, center-to-center feature spacing (${\Lambda}$), sidewall sloping function (SF), and fill factor (FF), as shown pictorially in Fig. 1. Details of the fabrication process are given in Supplement 1. During the etching process, the mask nanoparticles may be purposefully etched away, resulting in sloped sidewalls of the NS features as seen in Fig. 1(b). This sloping is defined by a sloping function (SF) that varies from 0, for pillars with completely vertical sidewalls, to 1 for cones with completely sloped walls in which the cross section consists of triangular features adjoined at the base. The term fill factor is defined as the area fraction of NS features relative to the underlying substrate; see Fig. 1(c) for a visual representation.

 

Fig. 1. Fabrication process and design parameters for nanostructured surfaces. (a) Production of substrate-engraved nanostructured surfaces, in which a thin gold film is dewet to produce a dry-etching mask. (b), (c) Surface cross section and top view, respectively, with the sloping function (SF) and period (${\Lambda}$) shown in (b) and described in the text. The fill factor is given by the area fraction of NS features relative to the substrate surface area, where the ‘A’ in (c) represents the NS feature surface area projected to the underlying substrate.

Download Full Size | PDF

 

Fig. 2. Structural parameters (pit depth and masking fill factor) key to minimizing the NS reflection for different depth index grading cases. (a) Vertical sidewalls (${\rm SF} = {0}$); (b) ${\rm SF} = {1}$. Cross-section illustrations of the simulated nanostructured surfaces are shown below the contour plots, where the contours (color scales) give percent reflectance.

Download Full Size | PDF

The into-the-layer effective refractive index profile for an engineered fused silica NS of this design can be approximated by the FF according to Bruggeman’s mixing formula applied to a stack of thin planar layers [28]; see Figs. 1(b) and 1(c) for idealized cross section and top view. The index will vary with the mixing ratio of the substrate and void volumes; i.e., for cones with a gradual transition from 100% air at the apex to 100% substrate material at the base of the cone, the index will transition smoothly from ${{n}_{\rm air}}$ to ${{n}_{\rm substrate}}$ with a rate of change dependent upon the feature height—stemming from the concept of into-the-layer graded index. To contrast this scenario, pillars with vertical sidewalls offer the opposite situation in which the volumetric fraction of the substrate medium does not change vertically through the layer, so the effective index of this substrate/voids layer remains constant for a given pillar height. In this situation, the minimal reflection condition is satisfied for an appropriate NS layer thickness. To function as an AR without appreciable scattering, the period ${\Lambda}$ needs to be smaller than $\lambda /{2}$ [28]. Consequently, the features need to be smaller than the wavelength of interest, but by fabricating features that are not too small there are two obvious advantages: (1) the mechanical stability of the features may be retained and (2) a more dynamic range of the into-the-layer index profile can be capitalized on. The design of an optimal AR will depend on the specific application (e.g., the required reflectivity over a bandwidth, acceptance angle, the mechanical stability) [28], and therefore the ability to predictively tailor, as we demonstrate here, is of high significance.

Transmission matrix calculations [28] based on this NS geometry are presented in Figs. 2(a) and 2(b) for an incident 351 nm source in the cases of 0% SF and 100% SF, respectively, highlighting that with control over the NS feature height and sloping, it is possible to design a tailored AR. Illustrations of the modeled NS cross section are shown below the respective contour plots. For vertical pillars (${\rm SF} = {0}\%$), the reflectance minimum occurs at 45% FF and a depth ${\sim}{75}\;{\rm nm}$ (with additional higher order solutions at taller NS features). For ${\rm SF} = {100}\%$ [see Fig. 2(b)], the fill-factor dependence falls away at the expense of needing taller structures. To reach a reflectance equivalent to the ${\rm SF} = {0}\%$ reflectance minima, the pit depth must be considerably greater—on the order of ${\sim}{300}\;{\rm nm}$—emphasizing the flexibility in the optical design gained by the degrees of freedom. The fabrication tolerances of NS topography in this process are very accommodating to the shape and feature etching depth.

For this work, rAR NS were fabricated for 351 nm (i.e., ${3}\omega$) and 1053 nm (i.e., ${1}\omega$) and are shown in Fig. 3. These two wavelengths were chosen due to their paramount importance for high-power laser systems, such as the NIF [24], as these bound the UV and IR cases for most high-power lasers of interest. The top views of the ${3}\omega$ and ${1}\omega$ NS are given in Figs. 3(a) and 3(c), respectively, while the cross sections are shown in Figs. 3(b) and 4(d). Refer to Supplement 1 for specific values of the FF, SF, and feature height for the fabricated surfaces.

 

Fig. 3. Nanostructured surface features fabricated for a wavelength of 351 nm in (a) and (b) and 1053 nm in (c) and (d). (a), (c) Top views. (b), (d) Cross sections. Note: magnifications vary across the SEM micrographs.

Download Full Size | PDF

The predicted and measured reflectance for the ${3}\omega$ NS are given in Figs. 4(a) and 4(b), respectively, and agree well with each other. The back-surface reflection has been removed to leave only the reflection from the front nanostructured surfaces. The reflectance ($p$ polarization) of reference unetched fused silica is shown in Fig. 4(b) to exemplify reflectance reduction versus a planar fused silica optic. The red indicator on the reflectance contour map in Fig. 4(a) marks the pit depth and fill factor of the ${3}\omega$ AR that was created; this predicted reflectance value is then portrayed by a red indicator on the measured reflectance plot Fig. 4(b). Measurements at 351 nm showed 0.37% reflectance per etched interface, while the predicted value [see Fig. 4(a)] is 0.32%, a deviation by 0.05% reflectance. Analysis predicts that by reducing the NS feature height by 30 nm, the reflectance would drop to 0.04%.

 

Fig. 4. Predicted and measured reflectance for the generated NS AR. Reflection predictions for ${3}\omega$ and ${1}\omega$ are depicted in (a) and (c), respectively, while the measured reflectance for ${3}\omega$ and ${1}\omega$ are given in (b) and (d), respectively. The red indicators in (a) and (c) denote where the generated NS lies on the contour maps, while the red indicators in (b) and (d) illustrate the predicted reflectance for the fabricated topographies. Panels (b) and (d) plot the NS reflectance against unetched fused silica.

Download Full Size | PDF

Similarly, the predicted and measured reflectance for the ${1}\omega$ NS are shown in Figs. 4(c) and 4(d), respectively, and are in good agreement. For the manufactured NS at ${1}\omega$, the measured reflectance was 0.42% and the predicted reflectance was 0.45%, resulting in a difference of 0.03% reflectance. Having validated the model for these test conditions, one point should be emphasized: further etching of the ${1}\omega$ NS to generate features with 100 nm greater heights would reduce the reflectance to 0.05%. Variations between the predicted and measured reflectance may be attributed to uncertainty of the measured reflectance.

Following demonstration of the desired optical performance as an AR coating, laser damaging revealed a new mode of damage near the damage threshold associated with the removal and flattening of nanoscale features of the NS. Both the ${1}\omega$ and ${3}\omega$ NS were subjected to high peak fluence pulses, and the laser-induced damage was measured in accordance with ISO 21254. Upon exposure to sufficiently high laser fluences, damage to the input-surface nanostructured surfaces may occur in the form of both traditional fractures as well as a new mode of damage related to the NS features that has not been reported before—henceforth referred to as “NS damage.” This NS damage was observed for both ${1}\omega$ and ${3}\omega$ exposures, and is shown as a top-view SEM image in Fig. 5(a) and as a cross section in Fig. 5(b). NS damage causes removal of the etched features, with the cross-section images in Fig. 5(b) revealing that the features completely vanish at the center of the damaged region, with an intermediate region displaying partial feature smoothing. NS damage typically results in micrometer-scale regions of the NS that have experienced feature deterioration, and consequently the local reflectance increases relative to the initial AR value. However, due to the limited size of these featureless regions, full aperture transmission is minimally impacted. It is important to note that in all cases of observed NS damage, fracturing was not detected, indicating that the features are not mechanically removed by an explosive mechanism. Consequently, NS damage can be considered benign, as successive laser shots to a previously NS damaged region did not cause damage growth. Characterization of the damage threshold following NS damage is ongoing, with early results suggesting that because NS damage does not grow, after this damage has occurred the optic has a higher local damage threshold. In other words, following the creation of benign (non-growing) NS damage, the optic can withstand increasing fluences until the next laser-damaging mechanism is reached. This emphasizes the great benefit of nanostructured surfaces.

 

Fig. 5. Nanostructured surface damage morphology top view SEM in (a) for ${3}\omega$ exposure, and cross section SEM in (b) following ${1}\omega$ exposure, with an accompanying reflection micrograph to depict the location relative to the damaged region. (c) Proposed damage mechanism.

Download Full Size | PDF

The origin of NS damage, observed for both ${1}\omega$ and ${3}\omega$ exposure, is likely a thermally driven material reflow and/or evaporation mechanism. All observed instances of NS damage were accompanied by plasma emissions, detected visually and audibly. Resultant NS damaged sites were not centered with the incident beam (i.e., not necessarily in peak fluence regions, data not shown), were generally orders of magnitude smaller than the ${{1/e}^2}$ diameter of the incoming beam, and typically preserved circular symmetry, all of which are indicative of plasma-initiated damage. The proposed plasma-initiated NS damage mechanism is depicted in Fig. 5(c). X-ray photoelectron spectroscopy (see Fig. S1 in Supplement 1) revealed that the only contaminants on the etched and cleaned surfaces were carbon, sulfur, and nitrogen, with atomic concentrations of ${4.2} \pm {0.4}\% \;{\rm C}$, ${0.4} \pm {0.06}\% \;{\rm S}$, and ${0.5} \pm {0.06}\%$ quaternary ${N}$ at the air/nanostructured surface interface. The error bars represent standard deviation from analysis of multiple locations on the sample. The etching mask material gold was not detected, indicating that gold diffusion into the substrate bulk is negligible and does not present a laser-damage precursor. While the concentrations of the detected surface contaminants are considered small, they are the expected plasma sources as any potential contaminant aggregation could serve as a potent light absorber [29,30]. Still, other defects from the NS structures are possible, but none could be correlated to laser damage.

Laser-induced damage of the fabricated NS is described by an onset value of around ${30}\;{{\rm J/cm}^2}$ under ${3}\omega$ irradiation with ${{1/e}^2}$ diameter of 1 mm and pulse length of 8 ns. NS damage and traditional fracture damage as a function of fluence incurred during ${3}\omega$ one-on-one testing, i.e., one shot per site, is depicted in Fig. 6(a). A micrograph in reflectance mode was taken of the site prior to the incident laser pulse, and another image was taken following the laser exposure. The difference between these two reflection maps reveals the laser-induced damage; see Fig. 6(a) inset. A laser fluence contour map was superimposed on this image, resulting in the percent damaged area per fluence bin. The damaged area plot in Fig. 6(a), representing a single irradiated area, reveals an onset of damage around ${30 - 35}\;{{\rm J/cm}^2}$, where Fig. 6(a) depicts a combination of both traditional (fracture) laser damage and NS damage. It can furthermore be inferred that 100% of the area irradiated by ${3}\omega$ fluences greater than ${50}\;{{\rm J/cm}^2}$ is damaged, signifying a 100% damage probability. By employing a relatively large beam and capturing an in situ spatial fluence distribution of the input laser beam, damage statistics can be collected from a single pulse laser exposure. The damage area probability plot, which is obtained as described above, can be considered the one-on-one damage probability curve (ISO 21254). All fluence values reported are peak fluence.

 

Fig. 6. Laser-induced damage testing results of (a), (b) ${3}\omega$ NS AR and (c), (d) ${1}\omega$ NS AR. (a) Percent damaged area per fluence bin, with the inset showing a typical microscope image with fluence contour map to calculate fractional damaged area. (b) Percent damaged area (black color) on the left axis, and damage density on the right axis (red color) under ${3}\omega$ exposure. Damage density data was generated from ${10}\;{{\rm mm}^2}$ total irradiated area. (c) Peak fluences at which nanostructured surface damage was induced over an array of 100 damage sites under 1ω exposure. (d) Damage probability (at ${1}\omega$) of an etched and subsequently cleaned nanostructured surface (red), laser-conditioned nanostructured surface (black), and reference unetched fused silica substrate (blue). All dashed lines are linear fits at the onset of damage to guide the eye.

Download Full Size | PDF

In an effort to identify the onset of laser damage, i.e., the threshold at which the optic could be operated without single-shot damage, only the lower fluences of Fig. 6(a) are of interest. Pursuant of this onset threshold, the percent damaged area data centered around when the optic began to damage is plotted against the damage density (red data) in Fig. 6(b). To acquire damage density data, 13 damage sites (induced by 1 mm beam diameter, 8 ns pulse duration) were investigated in regions of lower incident fluence to enable individual damage site counting. The damage density was then averaged across the 13 laser shots to generate the shown density plot. In support of Fig. 6(a), linear fitting of the damage density also indicates that damage onset for 3ω exposure occurs at roughly ${30}\;{{\rm J/cm}^2}$. To place this damage threshold into context, the upper echelon of the probability density function for ${3}\omega$ fluence on the NIF is ${\sim}{15}\;{{\rm J/cm}^2}$ [31].

Nanostructured surface damage of the designed AR experiences an onset value of ${62}\;{{\rm J/cm}^2}$ during R-on-1 testing with a 4 ns ${1}\omega$ pulse, 165 µm beam diameter. For the R-on-1 testing an initial peak fluence of ${20}\;{{\rm J/cm}^2}$ was used, which was increased by 10% for each subsequent pulse at a frequency of 10 Hz. As the surfaces being tested were engraved on 1 mm thick substrates, exit-surface damage to the thin optics was a concern with R-on-1 testing, and exposure to a previously damaged region was avoided to prevent irreparable damage to the optic via fracture growth. Consequently, the sample was moved to a different location for testing when damage occurred. In this way, 100 sites were irradiated, but due to termination for exit-surface damage, only a subset of the 100 sites generated NS damage. The resultant NS damage is given in Fig. 6(c), showing damage onset at ${62}\;{{\rm J/cm}^2}$. Of the 100 tested sites, 11% of the sites damaged at the exit surface first at fluences $\gt 145 \; {\rm j/cm}^2$, so NS damage statistics do exist even beyond ${150}\;{{\rm J/cm}^2}$, as indicated in Fig. 6(c).

As the likely mechanism of NS damage is material reflow induced by local heating when a plasma event occurs, if the plasma precursors are removed, the NS damage onset should shift to higher fluences. To explore this, two regions of a fabricated NS were probed—one of which was in the as-fabricated and cleaned state, while the other region had been laser conditioned (initially exposed to lower laser fluences to remove plasma precursors). Laser-induced damage threshold testing comparing these two regions to a cleaned fused silica reference (without NS) is shown in Fig. 6(d) using a $1 \omega$ one-on-one testing procedure following the laser conditioning process. All data points in Fig. 6(d) represent the average across tested sites; raw data can be found in Fig. S2. For the NS without laser conditioning, the first fluence yielding a non-zero probability occurs at ${49}\;{{\rm J/cm}^2}$, corresponding to a 0% probability intercept of ${39}\;{{\rm J/cm}^2}$ from the linear fit. Meanwhile, for the laser conditioned NS the first non-zero probability occurs at ${88}\;{{\rm J/cm}^2}$ with a linear fit 0% probability intercept of ${74}\;{{\rm J/cm}^2}$, a 90% increase relative to the NS without laser conditioning. The reference fused silica begins to damage at a comparable fluence of ${81}\;{{\rm J/cm}^2}$. This illustrates the possibility of optical cleaning, either laser conditioning or rapid thermal annealing, to remove plasma-inducing precursors, substantially broadening the range of acceptable fluences. It is important to note that R-on-1 testing, shown in Fig. 6(c), laser conditioned the NS to an extent, hence the NS damage onset at ${62}\;{{\rm J/cm}^2}$. Figure 6(d) more accurately depicts the capabilities of these NS at ${1}\omega$ with and without rigorous laser conditioning.

Static indents using a 200 µm radius canonical tip confirm that the nanoscale features of the fabricated NS are mechanically durable, and the AR properties of the NS can withstand high pressures. Preliminary testing consisting of applying fingerprint pressure, drag wiping, and sonication did not damage the AR as detected by reflectance optical microscopy. To quantify the onset of AR failure for these structures, much higher pressures are required. Loads, ranging from 50 mN to 50 N, were applied to the ${1}\omega$ and ${3}\omega$ nanostructured surfaces using static indentation normal to the substrate surface. Reflectance variations for these applied loads were monitored through confocal microscopy; refer to Figs. S4 and S5 for raw data. For the NS designed for 351 nm, a UV confocal microscope (370 nm) was used, while for the 1053 nm NS an off-design wavelength (630 nm) was used; under 630 nm irradiation the ${1}\omega$ NS has a reflectance of 0.65%. For the ${3}\omega$ NS [see Fig. 7(a)], the reflectance remained fairly constant—although exhibiting some damage—for the 50, 100, and 500 mN loads and started increasing, i.e., increasing AR damage, at loads greater than 1 N. Confocal images at various loads, located above Fig. 7(a), depict typical 370 nm reflectance measured by confocal microscopy. Likewise, the ${1}\omega$ NS exhibited a similar trend of a subtle reflectance increase at 1 N applied load, with reflectance increasing for larger loads; see Fig. 7(b). Utilization of a 200 µm radius canonical indenter ensures that at an applied load of 1.0 N, the smallest possible applied pressure—corresponding to the 1 N load being distributed over the entire 200 µm radius hemisphere—is on the order of 4 MPa, which is two orders of magnitude higher than the average fingerprint pressure of ${\sim}{20}\;{\rm kPa}$ [32,33]. In actuality, the applied pressure during the 1 N load was likely much higher than 4 MPa, as evidenced by the damaged regions having radii much smaller than 200 µm. If the applied 1 N load is distributed over the AR damaged surface area (projection of the indent to the substrate surface) with a radius of 13 µm, as indicated by Fig. 7(a), then the applied pressure responsible for AR damage is on the order of 2 GPa. This signifies that substrate-engraved NS can maintain AR capability even when subjected to pressures that are orders of magnitude larger than those caused by accidental contact during optic handling.

 

Fig. 7. Static indents reveal NS robustness. A 200 µm radius canonical tip was used to load (a) the ${3}\omega$ NS and (b) the ${1}\omega$ NS, with loads ranging from 50 mN to 50 N. For both plots, the color legend from (a) is used. Confocal microscopy displays changes to the NS reflectance under these applied loads, with the images above (a) displaying typical 370 nm reflectance at the indicated applied loads. Diffraction patterns induced by subsurface damage are evident in the 50 N loads for both ${3}\omega$ and ${1}\omega$ NS.

Download Full Size | PDF

Analysis of damaged regions by SEM reveals that the mechanical mode of damage to the NS is not fracture-induced, but plastic deformation of the NS features. Mechanically loaded regions of the surfaces were investigated to explore how the NS features were being damaged. Images of these regions are given in Fig. 8 for the ${3}\omega$ NS, with the loading condition specified at the bottom of each image. Both ${1}\omega$ and ${3}\omega$ NS damaged the same way under loading; refer to Figs. S6 and S7 for electron micrographs of pressure-induced damage to both surfaces. Under loading the features appear to ‘flatten’, as evident in Fig. 8(a), which displays the boundary between damaged and undamaged NS features, with the entire damaged region visible in Fig. 8(b). The feature fill factor increases by 15% across the damage interface seen in Fig. 8(a); this is a result of the features growing radially and densely packed features coalescing into single plateaus. Features far from the damaged region (uncompressed) are shown in Fig. 8(c). Viewing the center of the damaged region [Fig. 8(d)] shows that the NS features are not sheared off, as there are no indications of fracturing parallel to the substrate surface. The features now have a smooth plateau, in contrast to the more cone-like initial topography in Fig. 8(c). The plateauing effect, coupled with the increased fill factor within the damaged region, indicates that the features compress vertically and expand radially. Additional evidence for this mechanism is given in Fig. 8(d), where fractures perpendicular to the surface are visible in many of the features. This is expected, as radial growth will induce tension and ultimately lead to fracture.

 

Fig. 8. SEM of NS damage induced by static indenting reveals plastic deformation of NS features. Panel (a) is a higher magnification of (b), showing the interface between damaged and undamaged NS. (c), (d) Unloaded NS features and features at the center of the region loaded with 0.5 N, respectively. These images display ${3}\omega$ NS damage, but ${1}\omega$ and ${3}\omega$ were observed to damage in similar ways.

Download Full Size | PDF

The radial growth mechanism under compression is consistent with the ${1}\omega$ NS AR resilience (no change in reflectance until 1 N loading), while for ${3}\omega$ an increase in reflectance is visible for all applied loads; see Fig. 7. For an applied load of 0.5 N [refer to Fig. S8 for SEM], the ${1}\omega$ NS was found to have a mean plateau major axis length of 200 nm (elliptical fitting), while the same loading produced a mean plateau major axis of 320 nm for the ${3}\omega$ NS due to the more densely packed NS feature coalescence. This increased plateau size for the ${3}\omega$ NS will increase scattering, while a plateau size of 200 nm for the ${1}\omega$ NS ($\lt \lambda /{3}$ for the 630 nm confocal) will have a minimal impact on reflectance. Thus, it is seen that lower loads are necessary to reach a critical plateau size for the ${3}\omega$ NS. The extent to which the NS features may have densified has not yet been investigated, but it is likely that densification is the precursor to radial growth. Both nanostructured surfaces have shown to be mechanically durable under loading normal to the substrate, demonstrating the robust antireflective property of these surfaces.

3. CONCLUSION

We have demonstrated and tested, for the first time, to the best of our knowledge, the ability to generate durable substrate-engraved nanostructured surfaces for scalable and designable AR that are potentially suitable for high-power lasers. The scalability is a product of the process simplicity—dry etching through dewetting-induced self-assembled masks. Designability follows as a result of the process simplicity and separable controls: dewetting allows for control over mask nanoparticle size and spacing, while the etching process permits regulation over the feature side slopes, giving rise to tunable nanostructured surface-driven AR coatings. Using this approach, the scalability of nanostructured surfaces—even for the short wavelength range of 351 nm AR applications—has been obtained.

Upon exposure to sufficiently high fluences, a new mode of damage, nanostructured surface damage, has been observed. This damage, likely the result of thermally driven material reflow induced by plasma initiation on the nanostructured surface, locally removes NS features and results in featureless regions of the NS on the micrometer scale. Under 1053 nm exposure, the onset of NS damage occurs at ${\sim}\;{39}\;{{\rm J/cm}^2}$ for a 4 ns pulse without laser conditioning of the nanostructured surface. Through a laser conditioning process the 1053 nm laser-damage performance was further enhanced to ${74}\;{{\rm J/cm}^2}$, an improvement of ${\sim}{90}\%$, and comparable to the observed substrate reference damage performance. For an 8 ns pulse at 351 nm without laser conditioning, the onset occurs at ${\sim}{30}\;{{\rm J/cm}^2}$. Damage values at both 1ω and 3ω are demonstrably higher than conventional AR damage thresholds [8,10,34]. Significantly, NS damage is inherently stable, and does not grow with successive laser exposure.

Initial cleaning tests consisting of sonication and drag wiping did not damage the AR as detected by optical microscopy. Under normal incidence mechanical loading with a 200 µm radius canonical tip, the mechanical durability and AR resilience of the fabricated NS were quantified. For the loads investigated, the reflectance of both the 351 nm AR and the 1053 nm AR was minimally impacted below 1 N. At a load of 1 N, the minimum applied pressure is 4 MPa—two orders of magnitude higher than an average fingerprint pressure. This indicates that accidental handling contact is not detrimental to the NS AR properties. For loads as high as 10 N with the 200 µm radius indenter tip, the mechanical damage mode is plastic deformation of the NS features themselves, which compress vertically and grow radially.

These nanostructured surfaces show great promise to impactfully expand the applications of high-power lasers, including compact new particle and x-ray sources, renewable energy, directed energy, and fundamental light–matter interaction exploration.