1. Introduction

The fabrication of optical components in fused silica from an initial liquid state, such as sol gel processes [1] or modern glassomer techniques [2] have gained considerable attention in the last years and decades. These techniques combine an exceptional flexibility in producible element types with the outstanding physical properties of fused silica such as its high chemical and thermal stability, as well as their excellent optical characteristics. The broad diversity of optical elements in fused silica manufactured from an initial liquid phase comprise diffractive elements [2], micro-cone textures for solar cell application [3], three-dimensional suspended hollow microstructures [4], micro-structured optical fibers [5], and microlens arrays [6] to name only a few. In essential, all these listed elements profit from the transparency of fused silica.

In this contribution, we demonstrate that a liquid based sol-gel process for fused silica is highly advantageous for non-transparent optical applications as it allows the fabrication of extremely absorbing and diffuse reflecting optical elements.

Diffuse reflective surfaces are used e.g. in integrating spheres or as test targets to calibrate spectrophotometers or image sensors. In particular, such diffusers are key components for the calibration of earth observation satellite sensors necessary for climate research, agriculture or environmental monitoring [7–9]. Especially for space applications, the diffusers are exposed to extreme and harsh conditions.

On the other hand, a high absorbance is of enormous importance for applications such as solar cell technology, improving the performance of infrared cameras or for preventing stray light in optical instruments such as telescopes, microscopes or projection units. Here, vertically aligned carbon nanotube (VACNT) showed their superior absorbance properties with an extremely low reflectance from the visible to the far-infrared [10–12] but suffer from touch-sensitivity and their vulnerability to external mechanical influences.

For both the highly absorbing and the diffuse reflecting fused silica, we modified a sol-gel process and applied a subsequent surface modification step creating sub-wavelength sized structures in the materials surface. In essential, the modified sol-gel process involves the adding of specific filler particles to the initial sol state and an appropriate treatment of the resulting xerogel in a step before the final high temperature sintering process is applied.

The highly absorbing and diffuse reflecting properties of the material are mainly based on volume effects caused by the micrometer and sub-micrometer sized filler particles or remaining cavities in the bulk material. A further improvement of diffusing properties and absorbance is achievable by suppressing disturbing specular reflection effects caused by the refractive index step at the interface between the two media of the fused silica substrate and the surrounding ambient air. In particular, subwavelength structured optical surfaces have proven their ability to minimize reflection effects, specifically in terms of broadband capabilities and usability over an extremely wide range of incidence angles [13,14]. These subwavelength antireflective (AR)-structures are also found in nature, for example on transparent butterfly wings [15] or on the corneal surfaces of night active insects [16,17], and with relation to their natural model, they are also called moth-eye-structures. In essential, the AR properties of moth-eye structures are based on a gradual decrease of the effective refractive index offering a smooth transition from the environmental medium to the bulk material, and therefore allow to minimize Fresnel reflection losses. For the application on volume modified sol-gel fused silica, we applied a specific manufacturing process for these moth-eye-structures, based on a combination of block copolymer micelle nanolithography (BCML) followed by a reactive-ion-etching (RIE) process [18]. The combined BCML-RIE process offers a high flexibility and the useable wavelength range can be extended to both the ultraviolet (UV) [19] and to the near-infrared (NIR) [20].

2. Manufacturing process for diffuse reflecting and highly absorbing fused silica

The manufacturing process of the substrate material for diffuse reflecting and highly absorbing fused silica is mainly based on well established procedures for clear or highly transparent sol-gel quartz glass [6]. The essential modification concerns the inclusion of concentrated cavities with dimensions in the micrometer range or respectively a low amount of nanometer sized, absorbing carbon particles. Since the basic process is the same for both the diffuse reflecting and the absorbing material, we simultaneously describe in the following the manufacturing steps in detail and indicate the essential differences. For visualization, the essential process steps are schematically depicted in Fig. 1.

 

Fig. 1. Sol-gel process to prepare diffuse reflecting (above) and highly absorbing (bottom) fused-silica substrates.

Download Full Size | PDF

In a first step the chemical precursor solution sol is mixed from demineralized water, tetra-ethyl-orthosilicate (TEOS: Si(OC₂H₅)₄; Dynasylan A, Evonik) and colloidal pyrogenic silica (OX50; Evonik) in a molar ratio of 25/1/4 until hydrolysis between TEOS and water ends. A pH value of 4.9 is adjusted by adding a 0.1 Mol/l ammonia solution drop by drop. According to this ratio, the hydrolyzed sol has a mass density of approximately 1.1 g/cm³.

In this stage, micrometer sized spherical particles, respectively the carbon black nanoparticles are added to the sol and the mixture is stirred for several minutes. In the case of the added carbon black nanoparticles, the mixture is additionally treated by ultra-sonic. In order to suppress agglomeration of the carbon black particles in the sol, it is advantageous to pretreat the carbon black particles with surfactant- or dispersing-agents. In analogy, to prevent sedimentation or floating up of the added spherical micro particles, their mass density should be close to those of the sol.

In particular, we used monodisperse polymethyl methacrylate (PMMA) spheres with a diameter of 5.5 µm (PM006UM; microParticles GmbH). To finally create an opaque silica glass body containing appr. 2 × 10⁹ cavities/cm³ we added PMMA spheres with a mass-percentage of 3% to the sol. On the other hand, for the highly absorbing fused silica, we used Acetylen carbon black particles with an average diameter of 42 nm (MFCD00133992; abcr GmbH) pretreated with a surfactant agent (Triton - X114; Fisher Scientific UK). Here, 0.075% (weight percent) Acetylen carbon black was mixed to the sol.

After stirring and ultra-sonic treatment, the mixture is poured into a mold where a gelation process starts, forming the randomly linked three-dimensional silica network with liquid filled nanometer sized pores (wet-gel) in approximately 30 minutes. The wet-gel carries the integrated PMMA microspheres or carbon black particles, respectively, in a homogenous distribution.

Thereafter, the wet-gel is removed from the mold and dried in air for 7 days at constant room temperature and at humidity levels between 80–30% following a defined drying regime. During the drying process the initial wet-gel is transferred into a xerogel by removal of the liquid phase in the nanometer pores. In this phase, the gel experiences already a partial volume shrinkage, which will finally end in an overall volume reduction of 50%.

The subsequent process steps for the manufacturing of both diffuse reflecting and highly absorbing fused silica differ in detail from each other so that they are separately described in the following.

In case of the diffuse reflecting fused silica, the PMMA spheres have to be removed completely from the silica network in the next step. This is necessary to guarantee high quality quartz characteristics dominated by scattering properties and minimizing potential absorption effects. The PMMA spheres are removed from the xerogel in an oven at approximately 500 °C under oxygen atmosphere. To prevent the formation of cracks the temperature is increased slowly over a period of 24 hours. As a cleaning procedure the xerogel is then exposed for several hours to a hydrogen chloride atmosphere with a simultaneous increase of the temperature to 800 °C. After completion of the cleaning process, several flushing steps with oxygen are performed. In the xerogel, the remaining cavities have the same size as the initial polymer spheres. Finally, the xerogel is densified in a sintering process under a helium atmosphere at 1350 °C for about 10 minutes. Here, the nanometer sized pores in the silica network of the xerogel collapse, while the cavities of the initial spheres are shrinking but still remain in the medium. The final cavities in the silica glass have diameter of appr. 3.5–4 µm, are helium gas filled and represent a homogenously distributed non-open porosity.

For the highly absorbing fused silica the prepared xerogel is cleaned by an exposition to an inert gas atmosphere for several hours. In analogy to the process for diffuse fused silica, also here the temperature is increased to 800 °C and simultaneously accompanied with several flushing steps. This procedure helps to remove the residual water as well as the free gaseous oxygen from the silica network and protects the carbon black particles from burning. In a final step, the xerogel is densified in a sintering process under the same inert gas atmosphere at 1300 °C in about 60 minutes. Here, the nanometer pores in the silica network of the xerogel collapse and trap the carbon black particles into the glassy SiO₂ network. The final silica glass appears visually deep black and contains appr. 2.500–3.000 ppm (weight) carbon black particles homogenously distributed.

During the sintering phase the fused silica was cleaned by hydro-chlorine gas to remove trace impurities. In this stage OH-groups will be removed automatically, leading to an average OH-content of 0.1 ppm (weight) in the final dense bulk material.

The resulting fused silica substrates are small cylindrical plates with a diameter of 20 mm and a thickness of 1 mm. These substrates are mostly smooth only locally interrupted by some distributed intrusions where the filler particles partially appear in the surface. Incoming light incident on the untreated substrate surface experiences a refractive index step which is associated with reflections following the Fresnel-laws. In case of the cavity-filled silica glass, the specular reflections disturb the spatial homogeneity of the diffuse stray-light, or for the carbon black filled silica glass the reflections significantly limit the maximum absorbance. In order to avoid these effects, it is necessary to equip the fused silica surfaces with anti-reflection properties. Commonly, anti-reflection coatings for optical elements are based on single or alternating dielectric layer systems, which make use of interference effects for reflection suppression. These layer-based systems are only applicable to a limited wavelength range and are also characterized by restricted tolerances against the incidence angle.

An alternative are surface relief structures with spatial periodicities in the sub-wavelength range, which are also called moth-eye-structures in accordance with their natural model. These moth-eye-structures offer a graded refractive index transition between the ambient air and the substrate material and therefore drastically reduce Fresnel-reflections. In addition to their broad wavelength applicability and the large incidence angle tolerances moth-eye-structures offer also the advantage that they can be directly incorporated into the substrate material so that no additional layer material is necessary. This possibility increases the stability against temperature changes and enhances laser radiation stability [20].

For our purpose, to equip the volume modified sol-gel fused silica with moth-eye-AR structures we applied a process combination of block copolymer micelle nanolithography (BCML) followed by reactive-ion-etching (RIE) [18]. The combined BCML-RIE process is highly flexible and allows to cover an extended wavelength range from the ultraviolet (UV) [19] to the near-infrared (NIR) [20,21]. The process, schematically depicted in Fig. 2, starts with the formation of spherical micelles by dissolving polystyrene-blockpoly(2)-vinylpyridine (PS-b-P2VP) in o-Xylene and the loading of the cores of the micelles with a gold salt (HAuCl₄). Subsequently, the loaded micelles were transferred from the solution to substrates by spin coating. The resulting mono-micellar film shows a hexagonal pattern with a center distance of ∼105 nm ± 10 nm. For our applications we aspired a periodic mask structure, because a randomized disturbance of the lateral periodicity is assumed to reduce the transmission of the incoming light into the bulk material especially in the short wavelength range [22]. In particular, for the carbon black filled samples this effect will disturb the light absorption property.

 

Fig. 2. Preparation of anti-reflective moth-eye-structures on the surfaces of the volume modified fused silica. The process starts with BCML in which micelles are arranged quasi-periodically on top of the substrate. This is followed by a reactive ion etching step. The indicated hollow (half sphere) in the fused silica represents a residual cavity occurring randomly in the sol-gel process for the diffuse reflecting sample.

Download Full Size | PDF

In particular, in case of the diffuse reflecting fused silica, the mono-micellar film covers both the flat areas of the substrate as well as, at least partially, the hollow induced by the removing of the spherical micro-meter sized PMMA particles. After a plasma treatment in which the polymer matrix is removed and the gold precursor is reduced, a mask of elemental gold nanoparticles remains on the substrate, which serves as an etching mask for the following RIE process. The height of the moth-eye-structures is adjusted by different etching recipes, involving the composition of the etching gas and the sequence of the particular etching steps. In particular, we manufactured samples with different nanopillar heights representing specular reflectance minima at wavelengths of about 500 nm, 750 nm, 1000 nm and 1500 nm, respectively. For spectral position adjustment, the specular reflectance minimum was measured successively accompanying the etching process. The final corresponding nanopillar heights are found by scanning electron microscopy (SEM) to be approximately 200 nm, 305 nm, 420 nm and 680 nm. In [20] it was demonstrated that the reflectance minimum can be shifted by increasing the pillar heights to wavelengths larger than 2400 nm.

A contamination of the surface during the moth-eye-structuring process is not observable, in particular, the etching-mask is completely disappearing during the RI-etching. Potential contaminations arising later during the use of the moth-eye-structures are removable by appropriate cleaning steps [23]. The manufactured moth-eye-structures offer a high stability, especially they show a significant higher laser-induced damage threshold (LIDT) than classical multilayer based anti-reflection coatings [20].

In the following, the different samples are indicated by a letter which distinguishes between diffuse reflecting (‘d’) or highly absorbing (‘z’) samples followed by a number denoting the position of the specular reflectance minimum. For example, d1000 represents the diffuse reflecting sample with a specular reflectance minimum at 1000 nm, and z500 stands for the highly absorbing sample with a specular reflectance minimum at 500 nm.

To visualize the effect of moth-eye-structured surfaces on both highly absorbing and diffuse reflecting samples, Fig. 3 shows photographs in which the AR-coated samples are compared with the respective non-surface treated fused silica plates. In the upper image of Fig. 3, both non-structured (left) and AR-equipped (right) plates of the highly absorbing samples are located on a white underlay. Behind the fused silica samples, a black background paper is placed on which letters and symbols are printed in white color. On the non-structured sample, a mirror image of the letter ‘A’ and the ‘smiley’ symbol becomes clearly visible, indicating a reduced but still recognizable reflection. In opposite, the moth-eye-structured sample on the right appears completely dark showing no reflection effects. The lower image of Fig. 3 shows the situation for the diffuse reflecting samples. Hereby, the colors of the underlay and the background with the letter and symbols are inverted. Also here, the non-structured sample (left) shows a recognizable mirror image, whereas on the moth-eye-equipped sample reflections are hardly visible.

 

Fig. 3. Photographs of the highly absorbing (top) and diffuse reflecting (bottom) fused silica samples. The black and white samples are placed on a contrast inverted underlay. In the background letters and symbols are printed on black or white paper. On the non-structured samples (left), mirror images of the letter and symbols become clearly visible, indicating Fresnel reflections. In contrast, these reflections nearly vanish for moth-eye-equipped samples (right).

Download Full Size | PDF

Figure 4 displays SEM images of the resulting structures for exemplary diffuse reflecting and highly absorbing samples. For the SEM images, the sample surfaces were covered with a thin chromium layer. Figure 4(a) shows the surface of the diffuse reflecting sample d750 in an overview with low magnification and in a perpendicular view. A flat surface interspersed with statistically distributed spherical hollows becomes visible. The hollows, all with similar diameters smaller than ∼ 4 µm, are the residuals of the initial PMMA spheres, which are removed in the final stage of the substrate manufacturing process. An image analysis of Fig. 4(a) results in a coverage of ca. 8% by the hollows with respect to the entire surface. At this magnification level, the surface of all other diffuse reflecting samples (d500 … d1500), which are not shown, have the identical appearance compared to d750.

 

Fig. 4. SEM images of a diffuse reflecting sample ((a), (b)) and of a highly absorbing sample ((c), (d)). The left column ((a), (c)) displays an overview for both samples, in the right column ((b), (d)) a FIB-cut allows a detailed view also from the volume of the samples.

Download Full Size | PDF

For a more detailed view Fig. 4(b) shows a section of the d750 sample in higher magnification, captured at an oblique angle of 54°. From the observed area a part with a trapezoidal base was extracted by the application of a focused ion beam (FIB). The volume extraction by the FIB allows to get an inside view into the 3D volume structure of the sample substrate. The figure shows in the center three prominent spherical shaped hollows in the surface, from which the one on the left is cut in the center by the FIB. The flat surface of the sample is covered by homogeneously distributed moth-eye-structures. It can be seen that the moth-eye-structures are also covering the bottom of the hollows. The moth-eye-structures have a height of approximately 305 nm. When looking at the outer parts of the hollows it follows that the heights of the moth-eye-structures are decreasing with increasing sidewall steepness. With the FIB sectioning, also an empty sphere with a diameter of ∼ 4 µm becomes visible inside the substrate volume. For a better visualization the respective part of the image is contrast enhanced and indicated by a green dotted rectangular.

Figures 4(c) and 4(d) show SEM images of the highly absorbing sample z1500 recorded at an oblique angle of 54°. In particular, Fig. 4(c) displays an overview of the sample surface. The surface is homogeneously covered with moth-eye-structures and especially surface defects, comparable to the hollows occurring in the diffuse reflecting samples, are not recognizable. The rectangular shaped flat area at the rear on the right is an artificially prepared platinum layer, which serves for stabilizing and to achieve smooth cutting-edges during the subsequent FIB-cutting process. Figure 4(d) depicts the structures at a FIB cutting edge. The moth-eye-structures below the stabilizing platinum layer have a height of ∼ 680 nm. The transition between moth-eye-structured surface and underlying substrate is clearly indicated by a contrast change. The moth-eye-structures appear not to be freestanding individual pillars but seems to be randomly connected at their upper ends. The edge of the FIB-cut shows small intrusions inside the substrate volume, which are attributed to the particles or small aggregates of the carbon black.

3. Wavelength dependent specular and diffuse reflectance

Essential characteristics of the diffuse reflecting and highly absorbing fused silica samples concern their specular and diffuse reflectance properties, as well as the total hemispherical reflectance. For the respective wavelength dependent measurements, we used a spectrophotometer (Lambda 1050; Perkin Elmer) covering a wavelength range from 250 nm up to 2500 nm with a measurement uncertainty of < 0.5%. In order to analyze the scattering properties in more detail, the angle resolved scattering [24,25] was measured using the 3D scatterometer Albatross developed at Fraunhofer IOF [26] in combination with a supercontinuum light source that can be spectrally filtered between 470 nm and 1700nm using acousto-optical tunable filters. It should be noted that the highly absorbing samples with a thickness of 2 mm show no detectable transmitted light signal. For the diffuse reflecting samples, the ratio between reflected and transmitted portion can be tailored by adjusting the quantity of the PMMA spheres in the initial sol and by the thickness of the samples. For the presented diffuse reflecting samples the specular and diffuse scattering was only measured in reflection. In contrast, for the angle resolved scattering measurements the signals were recorded for both forward (transmission) and backward (reflection) direction.

3.1 Specular and diffuse reflectance for highly absorbing and diffuse reflecting fused silica

In a first step, we measured the reflectance for perpendicular incidence. Therefore, the samples were mounted at the exit aperture of the spectrophotometers integrating sphere. In this setup, the wavelength adjusted probe ray enters the integrating sphere at the entrance aperture and directs towards the sample. In order to enable that also the back reflected specular contribution is kept inside the integrating sphere, we adjusted the incidence angle not exactly to 0° but selected a slightly larger angle (∼ 6°). In direction of the specular reflection, the integrating sphere is equipped with a detachable segment, which allows to differentiate between measurements of pure diffuse reflectance (removed segment - specular contribution leaves the integrating sphere) and of total hemispherical reflectance (integrated segment - both diffuse and specular contribution are detected).

In the following, we present the measurement results of diffuse and specular reflectance for the different moth-eye-structured’ highly absorbing and diffuse reflecting fused silica samples, as well as for the respective unstructured reference samples. For the specular reflectance, we subtracted the measured diffuse reflectance from the total hemispherical signal.

Figure 5(a) shows the measured specular reflectance for the highly absorbing fused silica. The surface-unstructured reference sample shows a reflectance which is very similar to pure fused silica. For comparison we also added the reflectance properties of pure fused silica calculated from literature values of the refractive index [27] (dark yellow line). Across a broad spectral range, the reflectance of the surface-unstructured highly absorbing reference sample is nearly identical to the curve representing the calculated literature values. Small deviations of both curves appear in the deep UV and in the long IR wavelength range, which may be attributed to surface defects of the highly absorbing fused silica sample.

 

Fig. 5. Specular reflectance (above) and diffuse reflectance (bottom) for both highly absorbing (left) and diffuse reflecting fused silica.

Download Full Size | PDF

The samples equipped with moth-eye-structures show a completely different specular reflectance behavior. All four curves representing structured surfaces with different profile heights show qualitatively a similar behavior. In the short wavelength range, all curves show a local reflectance maximum which is smaller than 0.5%. With increasing wavelength, the curves are reaching their minimum at a level of the measurement accuracy, which measures a reflectance of less than 0.1%. With a further increase of the wavelength, the specular reflectance is again rising. The wavelength position of the absolute minimum shifts to longer wavelengths with increasing profile height of the moth-eye-structures. In particular, for the four samples we measured reflectance minima at 500 nm, 750 nm, 1000 nm and 1500 nm, respectively. These values of the reflectance minimum positions were also used to indicate the curves of the different samples in the diagram (z500…z1500). All four curves show a specular reflectance which is smaller across the entire spectral range than for the unstructured reference sample. In particular, it is remarkable that the specular reflectance curve for the largest profile height (z1500), exhibiting a reflectance minimum at 1500 nm, shows a reflectance smaller than 0.6% across the entire addressed wavelength range.

The measurements of the specular reflectance for the diffuse reflecting samples show a very similar behavior compared to the respective highly absorbing fused silica (see Fig. 5(b)). Again, the surface structured samples exhibit a reflectance minimum in the range of the measurement accuracy of approximately less than 0.1% at their individual target wavelengths. Also here, for the largest profile height and a reflectance minimum at about 1500 nm the curve shows a small specular reflectance with values below 0.5% over the total spectral range.

A comparison of the corresponding curves shows, that the measured specular reflectance for the diffuse reflecting samples is always a little smaller than for the highly absorbing fused silica. In particular, a maximum difference of 0.5% is found for the samples which are not equipped with moth-eye-structures. The samples with moth-eye-structures show a smaller difference with a minimum of smaller than 0.25% for the largest profile heights. This finding may be attributed to the randomly distributed hollows occurring in the surface of the diffuse reflecting samples and which are caused by the removal of the PMMA spheres during the sol-gel process (compare Fig. 4(a)). The hollows in the surface induce scattering effects and the appearing diffuse stray light reduces the contribution of the specular reflectance. In the highly absorbing fused silica samples surface defects are widely missing.

In Figs. 5(c) and (d) the measured diffuse reflectance is displayed for both highly absorbing and diffuse reflecting fused silica. For these measurements the detachable segment is removed from the integrating sphere so that only the stray-light is detected with the exception of the specular reflected contribution.

The highly absorbing fused silica shows a very small diffuse reflectance over the entire spectral bandwidth (see Fig. 5(c)). The curves for the unstructured reference sample and for the moth-eye-structured samples with different profile heights show only very small differences. In detail, the reference sample exhibits a slightly higher (∼0.025%) diffuse reflectance in the wavelength range from 1000 nm up to 2200 nm. Overall, all samples show a very small diffuse reflectance. In the short wavelength range the diffuse reflectance measures less than 0.5%, which uniformly decays to the infrared wavelengths with a minimum smaller than 0.17% at 2250 nm. In the deep UV and also in the IR at around 2500 nm, the reflectance increases discernible.

The diffuse reflecting samples show a diffuse reflectance of about 70% over the investigated spectral range (see Fig. 5(d)). Because of the limited thickness of the samples (∼1 mm) not all of the incident light is reflected (diffuse or specular) and a residual amount is transmitted as diffuse background without a specular component. Within a collection cone of ±0.3° centered around the specular beam direction, the transmitted light power amounts to 0.001% for all four structured samples as determined during the angle resolved light scattering measurements. For the highly absorbing samples, the specular and scattered transmitted light was below 10 ppm during the light scattering measurements. In the spectral range from approximately 500 nm to 2500 nm the unstructured reference and all moth-eye-structure equipped samples show nearly uniform curves for the diffuse reflectance, which slightly decay to the larger wavelengths. In detail, for the reference sample the diffuse reflectance is noticeable smaller compared to the structured samples. This effect can be attributed to the differences in the specular reflection of the different sample types.

The differences in the wavelength dependent behavior of the four samples equipped with moth-eye-structures are mainly based on their different specular reflection properties. The diffuse hemispherical reflectance is highest for the sample with the largest profile height for wavelengths larger than ∼1200 nm, which matches to the minimum of specular reflectance for this sample in the respective spectral range. In the adjacent shorter wavelength range (< 1200 nm) the sample with the next smaller profile height (indicated as d1000) shows the minimum specular reflectance. This correlates with the maximum diffuse reflectance for this sample in the corresponding wavelength range. With a further decrease of the wavelength (< 800 nm) the sample with the subsequent smaller profile height should show maximum diffuse reflectance. This is not observed but the respective curve shows the smallest diffuse reflectance for all moth-eye-structure equipped samples. This finding is not fully clear but may origin on a slightly different sample thickness compared to the other samples.

In the very short wavelength range (< 450 nm) the unstructured reference sample on the one hand and the moth-eye-structured samples on the other hand show a significant different behavior. While the diffuse reflectance for all moth-eye-structured samples is considerably increasing, the unstructured sample shows a remarkable decay. The increase of diffuse reflectance for the moth-eye-structured diffuse reflective samples in the sub-450 nm-range corresponds to the similar behavior for the highly absorbing samples. In opposite, the deviation of the unstructured diffuse reflective reference sample may be explained with the reduction of scattering effects of the hollows with decreasing wavelength.

3.2 Angle dependent specular reflectance for diffuse reflecting and highly absorbing fused silica

In the next step we investigated the specular reflectance for the different samples for varying incidence angles. For these measurements we changed the experimental setup and used the TAMS setup of the spectrophotometer (TAMS: Total Absolute Measurement System). The samples are fixed in an angle-adjustable mounting and the specular reflected light is captured by the input aperture of an integrating sphere. The angle position of the detector aperture is also adjusted by the TAMS. The diameter of the detector aperture is a little larger than the spot size of the specular reflected light. This means, that in addition to the specular reflected light also potentially a small amount of diffuse reflected light is detected with this configuration.

For both highly absorbing and diffuse reflecting samples we measured the angle dependent specular reflectance for the moth-eye-structured samples with the maximum profile heights and for the respective unstructured reference samples. The incidence angles were varied between 10° and 60° in steps of 10°.

Figures 6(a) and (b) show the angle dependency for both unstructured reference samples. All curves show a nearly uniform behavior with a slight decay for increasing wavelength. For incidence angles up to 30° the specular reflectance curves for each unstructured sample show only marginal differences in their absolute values.

 

Fig. 6. Angle dependency for both diffuse reflecting ((a), (c)) and highly absorbing fused silica ((b), (d)). The upper row displays the behavior for the unstructured reference samples, the row at the bottom shows the angle dependency for substrates equipped with moth-eye-structures exhibiting a specular reflectance minimum at about 1500 nm at perpendicular incidence. With varying incidence angles the spectral position of the minimum is shifted.

Download Full Size | PDF

A further increase of the incidence angles results in a drastic increase of specular reflectance. In particular, we measured approximately 8% for the unstructured highly absorbing sample and slightly above 7% for the respective diffuse reflecting sample. Again, the small difference between highly absorbing and diffuse reflecting samples can be attributed to additional scattering effects caused by the surface hollows of the diffuse reflecting sample.

The moth-eye-structured samples show a different behavior compared to the unstructured reference samples but present a similar characteristic among highly absorbing and diffuse reflecting samples (see Figs. 6(c) and (d)).

For small incidence angles (≤ 30°) the specular reflectance curves are similar to the previously discussed perpendicular incidence (see Figs. 5(a) and (b)). Across the entire wavelength range, the specular reflectance measures less than 1%. With increasing incidence angle, the specular reflectance significantly increases in the long wavelength range but remains low at short wavelengths. Additionally, the curves indicate that the minimum in the specular reflection shifts to smaller wavelength with increasing incidence angle.

Again, the slightly lower values of specular reflectance for the diffuse reflecting samples compared to the highly absorbing samples over the entire spectral range are due to the additional scattering caused by the hollows of the diffuse reflecting sample.

Essentially, a comparison of all curves shows that also for large incidence angles the specular reflectance is well below the comparable value of the respective unstructured sample. Particularly wide differences occur for wavelengths shorter than approximately 1500 nm. In this range the moth-eye-structures reduce the specular reflectance from approximately 8% to less than 1.5%.

3.3 Angle resolved scattering (ARS) measurements

In order to study the scattering properties in more detail, the angle resolved light scattering was measured within the plane of incidence from 470 nm to 1700 nm with the scatterometer Albatross. The incidence angle was fixed to 3° with respect to the surface normal. Exemplary, results for the diffuse reflecting fused silica are shown in Fig. 7. In the following, the mathematically positive polar scattering angle is counted with the respect to the macroscopic sample normal in the plane of incidence. Thus, the angles from -90° to +90° correspond to the backward scattering direction and the scattering angles ranging from -90° to -270° are indicating the forward scattering direction, respectively. Close to 90°, -90°, and 270° the sample holder blocks the scattered light, leading to a drop of the scattering signal.

 

Fig. 7. Angle resolved light scattering measurements of diffuse reflecting fused silica at an angle of incidence of 3°: (a) Backward scattering distribution in the plane of incidence at a wavelength of 1000 nm for different samples (step size polar scattering angle = 1°); (b) Wavelength dependency for sample d750. At a polar scattering angle of 3°, the incident light is blocked by the detector causing the obscuration in the measurement signal. The specular reflected beam can be observed at -3°.

Download Full Size | PDF

Figure 7(a) shows a logarithmic plot of the ARS as a function of the observation angle at a wavelength of 1000 nm for the four different surface structured diffuse reflecting samples. For comparison, the scattering distribution of a Spectralon sample (Spectralon SRS-99; Labsphere), which can be regarded as an almost perfect Lambertian scatterer [28], is also displayed. The Spectralon sample exhibits a total hemispherical reflectance (specular and diffuse part) of 99%.

All four structured diffuse reflecting fused silica samples are very similar in their scattering behavior with an almost identical ARS that is comparable to the scattering distribution of the reference Spectralon, however with a slightly lower amplitude. The moth-eye structured samples exhibit a small specular reflectance peak, which is missing for the Spectralon. This peak is however continuously decreasing in amplitude for increasing structure heights. The strong similarity of the scattering distribution for the surface structured diffuse reflecting fused silica compared to a Lambertian scatterer is also preserved for the visible and NIR spectral range, which is illustrated in Fig. 7(b). Explicitly, Fig. 7(b) shows the ARS of sample d750 for both backward (reflection) and forward (transmission) direction for a broad wavelength range in a color-coded diagram. The color display shows no significant changes in the angle dependency for the whole observed wavelength spectrum. In the forward scattering direction (transmission), the angle resolved scattering is also Lambertian-like and roughly a factor of 2 below the scattering level of the backward direction. A distinctive specular peak cannot be identified in forward scattering direction.

For the highly absorbing fused silica, the results from the angle resolved scattering measurements are plotted in Fig. 8. Again, the angle of incidence was set to be 3°. Figure 8(a) displays a logarithmic diagram of the ARS as a function of the observation angle at a wavelength of 1000 nm for the four different surface structured highly absorbing samples. Instead of the white Lambertian scatterer, a black Spectralon sample (Spectralon SRS-02; Labsphere) including absorbing carbon particles, is introduced as a reference which exhibits a total hemispherical reflectance of 1% at 1000 nm.

 

Fig. 8. Angle resolved light scattering measurements of highly absorbing fused silica at an angle of incidence of 3°: (a) Backward scattering distribution in the plane of incidence for a wavelength of 1000 nm (step size polar scattering angle = 1°); (b) Wavelength dependency for sample z750. At a polar scattering angle of 3°, the incident light is blocked by the detector causing the obscuration in the measurement signal. The specular reflected beam can be observed at -3°.

Download Full Size | PDF

The highly absorbing fused silica samples exhibit a diffuse scattering distribution as well, which is similar among the four structured samples. Their shape is similar to the black Spectralon, which suggests that the light scattering is mainly caused in the bulk material. The surface structuring only affects the specular reflected peak, which is comparable in amplitude to those of the diffuse fused silica. Except for the specular beam direction, the scattering signal for all structured highly absorbing samples is below the respective values for the black Spectralon, which indicates stronger absorption of the fused silica samples compared to the reference. Figure 8(b) displays the ARS of sample z750 in the backward direction ranging from 470 nm to 1700 nm. This wavelength plot also reveals a very homogenous scattering distribution in the entire covered spectral range.

For a further comparison to a Lambertian scatterer, the encircled energy around the specular beam in reflection direction was determined for the different surface structured fused silica samples. Therefore, the ARS-functions displayed in Figs. 7(a) and 8(a) were integrated with the assumption of an isotropic scattering distribution, meaning that the scattering is equal for all azimuthal scattering angles. The resulting encircled energy plots are presented in Fig. 9.

 

Fig. 9. Encircled energy around specular beam in reflection direction: (a) diffuse reflecting fused silica in comparison to a Lambertian scatterer at 1000 nm and (b) highly absorbing fused silica in comparison to a black Spectralon at 1000 nm.

Download Full Size | PDF

In case of the diffuse reflecting fused silica, all samples show a very similar behavior for the encircled energy as a function of the polar scattering angle (see Fig. 9(a)). In comparison to the reference of a Lambertian scatterer (Spectralon SRS-99) the curves have the same shape but differ in the achieved maximum value, which is only 60% for the fused silica samples and 100% for the reference, respectively. This gap is mainly attributed to the transmission effects associated with the fused silica samples. In more detail, when combining both the backward and the respective forward scattered light, the resulting curve for the encircled energy is nearly identical to the reference curve for the Lambertian scatterer. This is shown as the dotted line in Fig. 9(a), which is calculated exemplary for the sample d1500. This illustrates that the gap in the maximum achieved encircled energy between surface structured diffuse reflective fused silica and Lambertian reference can be minimized by increasing substrate thickness of the samples.

The encircled energy plots for the surface structured highly absorbing fused silica differ to the black Spectralon (Spectralon SRS-02) more clearly (see Fig. 9(b)). A first essential difference concerns the offset at 0° which is related to the specular reflected beam. Here, for the four samples z500…z1500 the offsets are different, which corresponds to the wavelength dependent specular reflection at the measurement wavelength of 1000 nm (see Fig. 5(a)).

Consequently, the offset exhibits a minimum for sample z1000 and a maximum for z500. Additionally, towards larger off-specular angles, the scattering intensity is only increasing marginally for the black fused silica samples and much less than the black Spectralon. Thus, the highly absorbing fused silica differs to a Lambertian scatterer. But for an application as an absorber this is beneficial as the primary focus lies on the absorption properties and not necessarily on the shape of the scattering distribution. Furthermore, the encircled energy curve for the sample z1000 is well below the corresponding curve for the reference Spectralon for off-specular angles larger than 20°, demonstrating the high absorption capabilities of the surface structured fused silica substrate.

4. Conclusion

In conclusion, we equipped volume modified fused silica substrates with a nanostructured surface to create anti-reflective properties. The volume modification is realized by adding filler particles or micrometer sized cavities during the sol-gel fabrication process of fused silica which results in highly absorbing or diffuse reflecting samples, respectively. The spherical, nearly equal sized cavities in the fused silica lead to strong scattering and thus white appearance, offering a homogeneous diffuse reflection across a wide spectral range. In contrast, carbon black particles in the fused silica are absorbing most of the incoming light and the respective samples appear black. The antireflective behavior is based on artificial moth-eye structures, which are created by a BCML step and subsequent RIE etching. The moth-eye-structures reduce significantly specular reflection effects at the samples surface over a large spectral range and for large incidence angle variations. Because the volume and the surface of the material in essential consists of pure fused silica, it offers a high potential for applications under harsh environmental conditions, such as expected e.g. for space applications, strong temperature changes or for the use with high power lasers. The desired characteristics of the material have to be tested and investigated in following research. The idea of volume modified fused silica by adding particles during a sol-gel-process offers a wide field of additional changes and optimization steps to tailor its optical characteristics. For example, a replacement of the micrometer sized spheres by ellipsoidal particles and in addition an oriented implementation of these particles into the fused silica volume, will introduce anisotropic scattering effects, which allow to reshape the diffuse scattering to a desired distribution.