Manufacturing Process and Characteristics of Silica Nanostructures for Anti-Reflection at 355 nm

Abstract

Recent advancements in photonics have intensified the performance requirements for optical systems and present significant challenges for optical coating technologies. Conventional interference coating systems often prove to be insufficient, especially in applications requiring large angles of light incidence or a wide wavelength range. Nanostructures, which consist of an air material mixture, offer promising alternatives. In this work, silica nanostructures are manufactured by the AR-plas2 method, in which first an organic layer is evaporated onto a substrate. This organic layer forms self-organizing nanostructures by a plasma etching step, which are subsequently coated with silica. Finally, the organic residues are removed by additional plasma etching and heat treatment steps, which results in hollow silica structures. The work examines the optical and functional properties of these structures designed for 355 nm to demonstrate their use as anti-reflective coatings for advanced optical systems.

1. Introduction

The rapid development in photonics opens new perspectives for future technologies, particularly in areas such as quantum computing, laser material processing, or semiconductor technology. However, these innovations also require increasingly sophisticated optical systems. That means they must be suitable, e.g., for high laser powers, withstand ultra-high vacuum conditions, address multiple wavelengths simultaneously, or guarantee outstanding optical performance even at high angles of light incidence.

The challenges are particularly evident in systems that utilize ultraviolet (UV) light. Not only is the choice of materials limited, but interactions between UV light and materials can also lead to unwanted absorption, which may result in the destruction of optical components. Anti-reflective (AR) coatings in particular play a key role here, as they not only have to provide excellent optical performance but are often the weakest element in such systems [1,2,3]. The choice of materials and manufacturing processes is particularly important here and should be chosen carefully [4]. For example, optical systems that operate with lasers with a wavelength of 355 nm are of technical importance as they are used in many fields, e.g., semiconductor technology or material processing. That is why several research groups have developed AR coating systems for this wavelength. Hafnia, as a material with a high refractive index in combination with silica (SiO₂), has proven itself as a coating system with a high laser-induced damage threshold and has been investigated by different groups [1,4,5,6,7,8]. Juškevičius et al. [5], for example, present an AR system that consists of silica and hafnia that was applied on fused silica using e-beam coating technology. The coating system achieves a reflection of less than 0.2% at 355 nm and a laser-induced damage threshold (LIDT) of 6.8 J/cm² on non-etched fused silica (measurement wavelength: 355 nm; 1-on-1 test according to [9]; pulse duration: 3 ns; repetition rate: 15 Hz; Nd:YAG laser). Falmbigl et al. [1] investigated a double-sided coated AR coating system on fused silica, also consisting of silica and hafnia, produced by ion beam sputtering. They achieved transmission values above 99.8% at 355 nm. Absorption was measured using photothermal common-path interferometry, and they achieved a value of around 110 ppm. The LIDT value was measured with 15.7 J/cm² (measurement wavelength: 355 nm; 200-on-1 test; pulse duration: 8 ns; repetition rate: 20 Hz). However, LIDT values could only be compared with each other when the same substrate or measurement parameters were used, because they depend on various factors, including the choice of coating method, the material quality and surface condition of the substrate [10,11,12]. Consequently, a more meaningful evaluation of the LIDT values is achieved by comparing them with those of the uncoated substrate, an approach that is employed in this study.

The use of commercial interference coating systems to minimize unwanted reflections has its limitations, as they are often unable to meet the new requirements for optical com-ponents due to the limited choice of materials and therefore predefined refractive indices. An alternative is offered by nanostructures, in which an air material mixture creates an effective refractive index that can be optimally adapted to the design [13,14,15,16,17,18,19]. Several studies show that the combination of interference coatings and nanostructures or the combination of more than one nanostructure provide improved optical performance [20,21]. In addition, the LIDT can be higher than that of interference coatings. This was already demonstrated in a round robin experiment in 2010, in which wet-chemically produced sol gel coatings exhibited the highest LIDT values for a wavelength of 355 nm [4]. Another experiment was performed by Tolenis et al. They compared nanostructures that were produced using glancing angle deposition (GLAD) technology, with a coating based on silica and hafnia made by ion beam sputtering (IBS); both were deposited on fused silica [22]. They achieved a reflection of 0.2% for the GLAD structures and 0.3% for the silica/hafnia coating system at 355 nm. They showed that the LIDT of the GLAD structures was 16.8 J/cm² compared to 5.1 J/cm² for the IBS manufactured coating (measurement wavelength: 355 nm; 1-on-1 test according to [9]; pulse duration: 3.1 ns; repetition rate: 15 Hz; Nd:YAG laser). Other processes to produce nanostructures besides wet-chemical coating processes [23] are described in the literature, e.g., lithographic processes [24,25], reactive ion etching [26,27,28] or production using physical vapor deposition technology [7,22,29]. In this paper, nanostructures made of silica are fabricated using AR-plas2 technology, which uses a commercially available chamber for physical vapor deposition equipped with an advanced plasma source (APS). This technology is cost-effective and suitable for mass pro-duction. A major advantage of this approach is the production of layer systems within one coating process, where nanostructures can be combined with interference layer systems, but also several structures with each other. This results in improved layer systems that meet the increasing requirements, e.g., for optical broadband systems. The method was originally developed for etching polymer surfaces and later adapted to the structuring of organic layers. For this, an organic transfer layer is deposited onto an arbitrary substrate. Subsequently, self-assembling nanostructures are formed by plasma etching, which are then overcoated with silica. The organic components can be removed by additional post-treatment steps, resulting in hollow silica nanostructures.

Due to the technical importance of optical systems with lasers for 355 nm mentioned at the beginning, nanostructures with an AR effect for 355 nm were produced and investigated in this work. Since they have a higher roughness due to their morphology, which in turn leads to higher optical losses, a structural analysis was conducted in addition to the examination of optical properties. The organic material used was Xanthine, which forms nanostructures after plasma etching and has already been introduced in a previous work [30]. Most of the organic material was removed by additional treatment steps, resulting in hollow silica structures. To investigate whether residual organic material was still in the structures, absorption and LIDT measurements were conducted.

2. Material and Methods

2.1. Sample Preparation

The nanostructures were designed for the wavelength of 355 nm and double-sided coated on SK-1300 (Ohara, Japan) glass with a 1 inch diameter and 1 mm thickness. There were no specific requirements for the substrate quality. The substrates were pre-cleaned using ethanol and subsequently cleaned in an ultrasonic bath.

The commercial coating chamber SyrusPro 1100 (Bühler AG, Alzenau, Germany) was used for the AR-plas2 method. The chamber is equipped with an APS to support chemical and physical reactions, a thermal evaporator to deposit organic materials, and an electron beam gun for oxide materials. During the process, a homogeneous Xanthine (Sigma-Aldrich, Taufkirchen, Germany) layer was first deposited from a molybdenum boat onto the substrate at an initial chamber pressure of 2 × 10⁻⁵ mbar. The deposition of the organic material was controlled by optical monitoring to achieve a thickness of 145 nm. Subsequent etching with the APS and an O_2/Ar gas mixture led to self-assembling nanostructure formation, which was coated with 32 nm of silica [30]. It is assumed that both chemical transformation and physical removal led to the formation of the structures. The organic part of nanostructure was removed by subsequent annealing and a standard wet-cleaning procedure, resulting in hollow silica nanostructures.

2.2. Topographical Measurements and Analysis

Roughness analysis was performed using an atomic force microscope (AFM) Dimension Icon (Bruker, CA, USA) and a white light interferometer (WLI) NewView 7300 (ZygoLOT Europe GmbH, Darmstadt, Germany). Using AFM measurements, roughness was analyzed for spatial frequencies from 0.02 µm⁻¹ to 256 µm⁻¹, corresponding to scan areas from (50 × 50) µm² to (1 × 1) µm². WLI measurements were performed for 5× to 50× magnification to calculate the roughness for spatial frequencies down to 0.001 µm⁻¹. By combining the different roughness measurement methods and areas, roughness information can be analyzed for a wide range of surface structures. Therefore, a quantitative evaluation of the data was performed by calculating the two-dimensional power spectral density (PSD) function for each measurement according to the following formula:

$$PSD \left(f_x, f_y\right)=\underset{L\to \mathit{\infty }}{\lim }{\displaystyle \frac{1}{L^2}}{\left|\underset{-{\displaystyle \frac{L}{2}}}{\overset{{\displaystyle \frac{L}{2}}}{\int }}\underset{-{\displaystyle \frac{L}{2}}}{\overset{{\displaystyle \frac{L}{2}}}{\int }}(h\left(x,y\right)e^{-2\pi i(f_{x}x+f_{y}y)}dxdy\right|}^2,$$

(1)

where h(x,y) represents the topographical surface data, L the scan size and fx and fy the spatial frequency components. The PSD function (Equation (1)) describes the relative strength of the individual roughness components in terms of spatial frequencies and thus consider not only the vertical, but also the lateral distribution of the measured topography data [31]. In addition, a scanning electron microscope (SEM) from Zeiss SIGMA SEM was used to magnify the structures.

For the ex situ characterization after laser damage tests of the samples, an OLX-5000 (Olympus, Tokyo, Japan) confocal laser scanning microscope was used, and the images were evaluated using the open-source software package Gwyddion (version 2.66) [32].

2.3. Reflectance, Transmission, Absorption, and Light Scattering Measurements

To investigate the optical performance, transmission (T) and reflection (R) have been measured with a UV/VIS spectrophotometer Lambda 950 (Perkin Elmer, Waltham, MA, USA) at an angle of light incidence of 6°. A special inset was used to measure absolute R und T values of the samples without changing its position [33]. The accuracy of the instrument is given with 0.1% in the visible and 0.2% in the UV spectral range.

Light scattering properties of the nanostructures were quantified by angle-resolved light scattering measurements (ARS; for measurement geometry, see Figure 1) using a 3D scatterometer (system MLS10, developed at Fraunhofer IOF). The ARS is defined as the scattered power (ΔP_s) into the solid angle ΔΩ_s as a function of the scattering angles θ_s and φ_s, normalized to this solid angle and the incident light power (P_i).

Considering the scattering over the entire hemisphere for backward and forward di-rection, the numerical integration of the ARS, excluding specular reflection/transmission, gives the total scattering (TS = Ps/Pi) over the whole hemisphere. TS can be calculated by the integration from the measured ARS, which is scattered into the forward or backward hemispheres. At normal incidence, scattering angles of 2° ≤ θs ≤ 85° must be considered for the measurement of TS (according to the ISO 13696 standard) [34]. A description of the set-up for the light scattering measurements can be found in [35]. The uncertainty is given with 10%.

Absolute absorption (A) measurements were performed using the laser-induced de-flection method. This method utilizes a photothermal technique with a transverse pump sample beam configuration. For the samples, a modified measurement concept was used, the so-called thin disc vertical concept. Further details are provided in [36].

The measurements for light scattering and absorption were performed at 355 nm.

2.4. Laser-Induced Damage Threshold (LIDT) Measurements

The laser-induced damage threshold (LIDT) measurements were conducted using a previously described set-up [37]. This set-up operates in an ISO class 6 clean room under ambient conditions and uses a tripled Q-Smart 450 (Lumibird, Lannion, France) system as the laser source. A schematic of the LIDT set-up is presented in Figure 2.

All tests were conducted with a linearly polarized (p-polarization) pulse with a dura-tion of 8.5 ns in the third harmonic (355 nm) of the neodymium-doped yttrium aluminum garnet (Nd:YAG) laser and a frequency of 10 Hz. The beam diameter, which was fitted with a Gaussian profile at 1/e², varied between 82 and 174 μm. The greatest contribu-tion to the overall error budget was represented by the pulse-to-pulse energy stability and the beam profiler accuracy. The combined error contribution to the tests was calculated to be approximately ± 8%. The LIDT tests were performed in compliance with ISO 21254 [9,38,39,40], using the 1-on-1, S-on-1, and R-on-1 methods.

3. Results and Discussion

3.1. Morphological and Surface Analysis

Figure 3 shows the SEM images of the silica nanostructures. In the cross-sectional image (Figure 3a) it can be seen, that columnar structures have been formed. The top view indicates that the structures are stochastically distributed over the surface (Figure 3b).

In addition to SEM images, topographical data were assessed using AFM and WLI for a quantitative evaluation of the surface by calculating the PSD functions. To capture information over a broad range of spatial frequencies, measurements were conducted in various measurement areas. Exemplarily, AFM images for a (1 × 1) µm² scan area are given in Figure 4.

The power spectral density (PSD) function is a technique for analyzing the distribu-tion of surface roughness over different spatial frequencies. Rather than just average roughness values (e.g., average roughness (Ra) or root mean square roughness (rms)), which only provide a single value containing the vertical information of the surface, the PSD function provides a more comprehensive view by adding lateral information of the surface features in terms of their frequency components. This provides a better under-standing of how the anti-reflective coating interacts with light at different wavelengths (e.g., in terms of light scattering) and can identify corresponding optical behavior. Particu-larly for light scattering the spatial frequencies are directly linked to the scattering direc-tion by the grating equations. The corresponding PSD functions were evaluated (as described in Section 2.2) and combined by geometrically averaging, resulting in a so-called Master PSD function [21]. This approach not only allowed for the analysis of the scattering-relevant spatial frequency range but also provided insights into the structures at both high and low spatial frequencies. This enabled us to evaluate the influence of the coatings on substrate roughness and, consequently, the overall morphology of the samples. The Master PSD functions for both the nanostructure and the uncoated substrate, are shown in Figure 5. Figure 5a) also shows the spatial frequency range relevant for light scattering at a wavelength of 355 nm. The Master PSD function reveals an almost perfect replication of the substrate roughness below f = 0.1 μm⁻¹, which is typical for thin films with low intrinsic roughness. A difference to the substrate occurs in the mid- and high-spatial-frequency range, which can be attributed to the nanostructure, where the roughness becomes relevant. However, it should be noted that only particle-free areas were considered for the roughness analysis. When particles are taken into account, the roughness increases drastically, as can be seen in Figure 5b), which shows the PSD functions for particle-free and contaminated surface areas. The origin of the particles is currently unknown and is already being investigated in more detail.

3.2. Optical Properties and the Calculation of the Effective Refractive Index

Figure 6 shows the corresponding R and T (a) spectra of a double-sided coated substrate and an uncoated substrate, and the optical losses (b), which were calculated by subtracting R and T from 100%. The wavelength range for the measurements was between 300 nm and 400 nm.

The nanostructures show a transmission value of 99.8% and a reflectance value of 0.1% at a design wavelength of 355 nm (see Figure 6a). This is comparable to the value of the system of interference layers given by Falmbigl et al. [1]. The minimum/maximum is shifted to 370 nm. However, the curve can be shifted to shorter wavelengths by further adjusting the coating parameters. The optical losses for the nanostructures are approximately 0.1% at 355 nm (see Figure 6b), which may contain contributions from both light scattering and absorption. However, since the noise of the measuring device prevents the exact determination of the losses, light scattering and absorption were measured separately according to the methods described in Section 2.3. The results are also given in this section.

To calculate the effective refractive index (n_eff), a three-layer model was assumed by reverse data fitting with OptiRe software (version 13.77, OptiLayer GmbH, Ismaning, Germany) using the spectra (Figure 6) and the heights extracted from microscope images taken with the SEM. For this purpose, the structures were divided into three areas and their corresponding n_eff values were estimated (see Figure 7). A height of 40 nm with an n_eff of 1.15 is calculated for H1, as well as a height of 20 nm and an n_eff of 1.3 for H2 and a height of 15 nm with an n_eff of 1.05 for H3. As the structures are not the same height and the material is not equally distributed, this calculation is subject to errors and is therefore only a first approximation.

Light scattering depends on different properties of the coating, e.g., the surface struc-ture, but also volume effects or material properties (e.g., refractive index). However, the surface itself has the greatest influence. Figure 8 shows the ARS curve for the structure in-cluding the uncoated substrate. In addition, the corresponding calculated TS value for the silica nanostructures is (1652 ± 165) ppm, and it is (209 ± 21) ppm for the uncoated sub-strate.

The curves in Figure 8 display distinct peaks at −180° and 0°, corresponding to the directions of specular transmission and reflection, respectively. When evaluating the residual scatter signal in the off-specular direction, an increase in both the back and front scattering directions is observed for the nanostructures compared to the scattering from the substrate. However, based on the examination of the structural properties (see SEM images in Figure 3), a lower TS value would have been expected, as the heights of the single columns are regular and not clustered. Looking again at the ARS curves in Figure 8, oscillations can be recognized for the nanostructures both in the near-and far-angle range. This indicates light scattering induced by particles, which is particularly evident in the forward scattering [41].

The absorption value for the nanostructures was determined to be (166.8 ± 15.0) ppm. This value can be attributed to the residual organic material, which was used during the manufacturing process and can still be contained within the silica structures.

3.3. Laser-Induced Damage Threshold (LIDT) Results

Each sample was tested within 80% of the clear aperture using a closed hexagonal pattern with 127 positions placed 1.55 mm from each other, thus fullfilling the general recommendation for the distribution to prevent cross contamination. For the 1-on-1 and S-on-1 tests, where S is 10, 10 positions with the same fluence were tested for statistical purposes, while 24 positions were tested in the R-on-1 test. For the latter, each position was initially irradiated with a maximum of 100 pulses at a constant fluence. Subsequently, the fluence was increased by 0.5 J/cm² and the sample was irradiated once more with a maximum of 100 pulses until the surface showed visible signs of destruction. The results for all tests are given in

Table 1. Summary of the LIDT values.

Sample	1-on-1 [J/cm2]	10-on-1 [J/cm2]	R-on-1 [J/cm2]
Nanostructure	4.38 ± 0.34	2.87 ± 0.20	1.37 ± 0.42
Substrate	6.49 ± 0.50	4.26 ± 0.30	2.28 ± 0.59

.

The results of the 1-on-1 test indicate that the nanostructures show lower LIDT values to the substrate. Furthermore, the 1-on-1 methodology test provides insights into the fundamental damage mechanisms of the given material. Using high-resolution laser scanning microscopy, different damage morphologies could be distinguished between the nanostructures and the substrate, as shown in Figure 9. Only damaged positions of fluences exceeding the LIDT were considered in the study. For the silica nanostructures, a damage pattern is shown at a fluence of 11.58 J/cm², while for the uncoated substrate, damage is shown at 24.77 J/cm².

Based on the crater depth analysis in Figure 9, it can be seen that the silica nanostructure and the substrate presented different forms of melting and resolidification of the material, either from the coating or in combination with the glass substrate [42]. In addition, the material ejected from the crater is deposited in the nearby zone, visible in the slight color change, in form of a halo around the crater. For the nanostructures, the crater depth exceeded the coating thickness, reaching depths between 150 and 300 µm.

As the results of the 10-on-1 test show, the LIDT values are lower than the values observed in the 1-on-1 test, which was anticipated. Figure 10 presents the microscopic images of the damage morphologies for the samples. The damage morphology of the nanostructures shows a few small circumferential cracks besides melting. There are no cracks, which might be related to the increased absorption. The analyzed surface spot on the substrate was irradiated with an intensity of 12.15 J/cm² and exhibited indications of profound ablation that has overcome the melting phase, which is characteristic for bare substrates or crystals.

Although a 0.5 J/cm² step and 100 pulses for each fluence were selected for the R-on-1 test, the results shown in Table 1 represent the mean LIDT value obtained from 24 individual positions. Looking at the uncoated substrate, a standard deviation of around 25% from the mean value can be seen, whereas the standard deviation for the nanostructures is around 30%. This may be related to small inhomogeneities in the substrate, which can cause local absorption points that are known to lower the LIDT value, especially in the nanosecond regime. In addition, organic residues in the nanostructures can reduce the LIDT, which may correlate with the absorption values in Section 3.2. Figure 11 shows an overview where damage occurred after the nanostructure and the uncoated substrate were irradiated with different fluences. Additionally, damage is found on both the front and back of all samples. The front side presents marks of pulse accumulation and the respective crater formation. The craters tended to be relative large in size and presented marks of molten materials and cracks/flaking from the edges [43]. The damage on the rear side can be explained by the Fresnel diffraction by contamination particles, surface imperfections, and perhaps by the nanostructures themselves [44]. No conditional effects could be identified in the tested samples.

Although the values are generally lower than those reported in the literature, it should be mentioned that the LIDT depends on various factors, such as the surface quality. Therefore, the values should always be compared with the LIDT of the uncoated substrate, which is the highest achievable LIDT.

4. Conclusions

In this study, we successfully produced hollow silica nanostructures specifically designed for a wavelength of 355 nm. The achieved transmittance of 99.8% shows very good optical performance suitable for anti-reflective coatings. The optical properties were described with a model consisting of three layers and calculated by reverse data fitting using transmittance and reflectance spectra. The surface roughness, formed by the thickened heads of the columns, exhibited an rms value of 3.7 nm. A PSD analysis revealed that the roughness of the nanostructures in the mid- and high-frequency range differs from that of the uncoated substrate, as the roughness of the nanostructures only becomes relevant in these ranges. In the low-spatial-frequency range (below 0.1 µm), the Master PSD function shows almost perfect agreement with the substrate roughness, which is typical for thin films with low intrinsic roughness. The total scattering, approximately 1652 ppm, absorption losses of around 170 ppm and the comparatively low LIDT value are presumably due to residual organics in the silica structures. Further reduction of the organics, e.g., by solvents, is currently being investigated. After optimization, the AR-plas2 process seems promising for producing UV laser-resistant AR coatings.