Structural colors are ubiquitous in nature. These are observed in a wide variety of living organisms, for example, from the wings of Morpho species of butterflies, the scales of Lamprocyphus augustus weevil beetles or peacock feathers , , , . Unlike dyes, where colors are produced by the selective absorption and emission of light by the dye molecules, the structural colors are manifestations of different forms of behavior of light in micro- and nano-structured materials and depend on the geometry of the structures and their spatial arrangements. The optical phenomena contributing to the structural colors can be resonant scattering, structure and material dependent absorption, diffraction or multilayer thin-film interference , , . Dyes and pigments are vulnerable to photo bleaching and they can degrade over time. In this regard, structural colors are advantageous as their optical properties are mostly governed by the physical and material parameters of the structured media , , , , . Research on the structural colors both in reflection and transmission has led to a stream of potential real-life applications that include high-density printing , , color filters , sensors , biomimetic devices , , artificial camouflage  and colored solar cells , . Precise control over the way by which the incident light interacts with nanostructures is necessary to obtain the desired spectral characteristics over the entire visible wavelength range. This can be achieved through specific choices of the material properties (e.g. refractive index, absorption) and geometrical designs for the nanostructures (e.g. dimensions, shape and spatial arrangement). One of the simplest ways to achieve colors is by exploiting the optical interference at the interfaces of multilayer or ultra-thin films , . Nanostructures on both high and low index dielectric substrates also exhibit a wide range of colors through wavelength-dependent light scattering and/or coupling to resonant modes. For example, semiconductor nanopillar/nanodisk arrays display reflective colors that are mainly determined by the resonant modes in the structure and, in some cases, together with the material absorption , , , , . Color filters have been fabricated by randomly texturing polymers with wavelength-dependent light-scattering properties that depend on the topography of the texture . Vibrant colors have also been achieved using metallic nanostructures, such as nanodisks and gratings, by enhancing localized plasmon resonances , , , .
Most of the cases mentioned above are nanostructures designed for the visible wavelength regime, and the structures and their spatial separations are in the few tens to few hundreds of nanometers. Achieving precision in the fabrication of these optical nanostructures is important as the optical properties are sensitive to the shapes and spatial distributions of the nanostructures. Top-down methods (nanopatterning and etching), bottom-up methods (self-assembly and solution synthesis) or their combinations have been used extensively to achieve the desired nanostructures. E-beam lithography (EBL) is most commonly used for patterning small areas due to its flexibility and high precision. Alternative methods for low-cost and large area nanopatterning, such as imprinting or colloidal lithography, are also used, but these often involve multi-step procedures , , , . The ion-sputtering-assisted formation of self-organized nanostructures, as reported in this work, is an interesting phenomenon that can be used to fabricate disordered nanopillar arrays (DNPA) in a single-step wafer scale process. With this procedure, it is possible to obtain uniform color all over the wafer without additional processing steps, such as pattern generation, metal deposition and subsequent etching.
The ion bombardment of III-V surfaces, such as InP and GaSb, results in the formation of self-organized nanopillars on the surface , , , . The preferential sputtering of the group V species leads to excess group III species (here, Ga or In) on the surface, which diffuse and coalesce to form self-assembled nano-islands. These (metallic) nano-islands function as self-generated etch masks for nanopillar formation, resulting in metalo-dielectric DNPA. In addition, the spatial arrangement of the metal-capped nanopillars can be disordered, thereby reducing coherent scattering. Unlike other topics, such metalo-dielectric nanopillar arrays with spatial disorder have received much less research attention in the context of structural color generation. However, the fabrication of hybrid Au(Cr)-Si metalo-dielectric nanostructures has been reported, and the results demonstrate unique optical properties by re-shaping the metal nanodisk on top of a Si nanocone . The high refractive index contrast between the nanostructures and the surrounding ambient/matrix is advantageous in tuning the optical properties. High-index nanostructures can efficiently confine electromagnetic fields and support different kinds of resonant modes depending on their dimensions. Therefore, tuning the shapes and sizes of individual structures can lead to sharp changes in the reflection features , , . The III-V semiconductor materials are promising candidates in this regard as most of these materials have refractive indices above 3 in the visible wavelength region. Even though there have been ample studies on silicon-based color generation , , , , very few attempts have been made in the context of III-V materials, which are important for several optoelectronic applications, such as photodetectors and solar cells , , .
In this work, we report on the generation of reflective colors from metalo-dielectric (In/InP) DNPA. The In/InP nanopillars are fabricated by a thermally driven, ion-sputtering-assisted, self-assembly process. We demonstrate that it is possible to generate, in a simple one-step process, spatially disordered nanopillars with sub-wavelength dimensions and average spacings. Vivid colors are obtained by tuning the nanopillar dimensions. By detailed optical measurements and electromagnetic simulations, we investigate the structural color generation from In/InP nanopillars, including the role of their dimensions and spatial arrangements. The spectral characteristics of the specular and diffused reflectance are separately investigated to identify the modal and scattering contributions to each of these components. We analyze the size-dependent absorption in the InP nanopillars and demonstrate how it plays an important role in determining the dips in the specular reflectance. The light-scattering properties of individual nanopillars are also examined in detail to explain the spectral characteristics of the diffused reflectance.
2 Experimental methods
To generate the In/InP nanopillars, the InP substrates (typically 1 cm2 in size) at 270°C were sputtered using 400 eV nitrogen ion beam. Sputtering times varied between 30 s to 4 min, which increased the nanopillar height and diameter from ~240–650 nm and ~140–220 nm, respectively. For these nanopillar dimensions, a range of vivid colors as shown in Figure 1A were obtained. The physical mechanism of the nanopillar formation and the process details are discussed in detail in Ref. . A representative cross-sectional scanning electron micrograph (SEM) showing the dimensions and shapes of the nanopillars formed by 2 min sputtering is shown in Figure 1B. From our previous studies, detailed material analysis reveals that these nanopillars contain an In-rich ball-like cap on top of the InP pillars . This is shown schematically in the inset of Figure 1B. As can be seen, all nanopillar samples have similar properties, albeit different in their dimensions. The relations between the nanopillar radii (i.e. of the In-rich cap) obtained from the top view SEM images and their heights for different sputter times are plotted in Figure 1C. With increasing sputter times, their heights increase along with the average radii of the In-cap due to continued buildup of In during the process. Within a given sample, the heights and radii of the pillars show small variations (Figure 1C). However, a few coalesced pillars could also be found (Figure 1D). The spatial arrangement of the nanopillars appears disordered. The 2D fast Fourier transform (FFT) of the image is shown in Figure 1E. The isotropic distribution of spatial frequencies and the lack of discrete peaks in FFT image clearly show that the nanopillars are spatially disordered, thereby implying that coherent scattering is unlikely. In all the nanopillar samples, the estimated average nearest neighbor distances (~250–320 nm) are in subwavelength range (Supplementary material, Table S1).
Reflectance spectra from the nanopillar assembly were measured using Perkin Elmer Lambda 950 UV/Vis/NIR spectrophotometer. The total reflectance was obtained by restricting the reflected light within an integrating sphere, and the diffused reflectance was measured from the light remaining inside integrating sphere after directing the specular light out. The specular reflectance was then calculated by subtracting the diffused from the total reflectance. The total reflectance measured for samples with different nanopillar heights (Figure 2A) clearly shows distinct reflectance spectra corresponding to the observed colors (Figure 1A). For larger pillar heights (>420 nm), the total reflectance is reduced in the entire spectral region. Interestingly, the colors disappear in all the samples when the top portions of the pillars (amorphous InP and the In-cap) are removed using a hydrogen fluoride (HF) treatment. The representative results are shown in the Supplementary material, Figure S2, where one can note the associated featureless reflectance spectrum. These observations indicate the importance of the dimensions, shape and material composition of the In/InP nanopillars in generating colors.
3 Results and discussion
To gain insights into the origin of spectral colors, we simulated the optical properties of the In/InP nanopillars by the finite difference time domain (FDTD) method using the Lumerical commercial software . As discussed previously, material analysis on self-assembled In/InP nanopillars has shown In-rich capping on top of the InP pillars composed of amorphous and crystalline regions . Accordingly, we modelled the material composition of the In/InP pillars. To keep the model simple, we approximate the In-rich capping as a sphere and the InP part to be composed of amorphous InP (a-InP). The shape parameters (r, ht, hb, rb1 and rb2) are illustrated in the inset of Figure 2B. The quantitative details of the shape parameters and the nearest neighbor distances are provided in the supplementary information section (Supplementary material, Table S1). As a simple simulation setup, we arranged pillars in a square lattice. The average nearest neighbor distance and the average In-cap radius, as determined by the top-view SEM images, were used for the period and the In sphere radius (r), respectively. A plane polarized light source (400–850 nm) was incident normally in the z direction. Total reflectance was calculated from a field monitor placed behind the source. The refractive indices for the In and InP substrates were taken from Palik  and that for the a-InP substrate came from the Adachi database . Perfectly matched layers (PML) were used on the z boundaries and periodic boundary conditions (BC) on the x and y boundaries. Figure 2B shows the simulated reflectance data for an In/InP nanopillar (height: 340 nm; In-cap radius: 80 nm) square array (period: 280 nm). Although the reflectance dip and peak positions are in qualitative agreement with the experimental data, the intensities show significant deviations. Changing the period from 250 to 400 nm does not have a dramatic effect on the position of the reflectance dip, while the overall reflectance is influenced to a certain extent (Supplementary material, Figure S3). However, changing the pillar shape and material composition does play a major role in determining the reflectance spectra (Supplementary material, Figure S4).
In order to obtain accurate reflectance spectra from the simulations, variations in the diameters and the heights of the pillars along with their disordered arrangement have to be considered. Therefore, we adopted a supercell approximation to be able to use periodic BC in the simulations without losing the disordered nature of pillar arrangement. We took the supercell of ~6 μm2 for this purpose. Within this boundary, the pillars are positioned at coordinates extracted from the top view SEM images along with their respective In-cap radii (r). For this supercell, periodic BCs were used on the x and y boundaries and PML on the z boundaries. The above approach closely represents the disorder in the nanopillar assembly. As shown in Figure 2A–C, the reflectance data obtained from these simulations are in quantitative agreement with the measured spectra for all the samples. As the simulations of the disordered nanopillar assemblies are computationally intensive, as discussed earlier (Figure 2B), it might still be useful to consider periodic arrangements in predicting the qualitative trends, such as reflectance dips and peak positions. Figure 2D shows the chromaticity positions on the CIE 1931 color map of the reflective colors (measured and simulated) from the In/InP nanopillar assemblies with different pillar heights. The details of the x, y chromaticity parameter calculations are provided in the Supplementary material. Though the colors reside within the standard RGB range, they are not as extended in the color gamut as presented in recent studies , . The chromaticity values estimated from the simulated and the measured reflectance spectra are in reasonable agreement.
It is important to determine the extent to which the diffuse and specular components contribute to the reflected color. For example, the color and glossiness can vary depending on the spectral intensities of the specular and diffuse reflectances . Therefore, it is necessary to study them individually and understand their physical origins. As seen on Figure 3A and B, the diffused and specular reflectances dominate in different spectral ranges, except for the short pillars for which the latter is consistently higher in the full visible range. For shorter pillars (height <420 nm) measurements show that the specular part dominates (Figure 3A) over the diffused part (Figure 3B) resulting in a glossy appearance. Although the diffused reflectance has different spectral characteristics depending on the nanopillar height, the intensity variation is within ~5%–15%. From the simulated total reflectance (Figure 2D), the specular component was calculated by extracting only the zeroth grating order from the field monitors and the diffused component by summing over the rest of the orders. These results, together with the measured data, are shown in Figure 3A and B. The similarity in the simulated and experimental spectra strongly supports the considered physical model for the nanopillars and the approximations used in the simulations to model the spatial disorder of the nanopillars.
It is interesting to analyze the role of a spatially disordered nanopillar arrangement on the specular and the diffused reflectance. To address this issue, one approach is to introduce spatial randomness to a periodically arranged nanopillar system . The disorder was obtained by displacing individual pillars in a square lattice to a new position (Prand) by a constant distance from their original positions (Po); however, the direction in which they are displaced is random. Quantitatively, this can be represented in Cartesian coordinates as
The above observations clearly suggest that the diffused reflectance increases with the disorder and can be attributed to the scattering from individual pillars. To establish this, we investigated light scattering from a single nanopillar on the InP substrate. The simulations were performed using a total field scattered field (TFSF) source incident normally in the z direction. The results are shown in Figure 3D. The inset shows the schematic of the simulation set-up. The scattering cross-section was calculated using field monitors (X, Y and Z normal) placed outside the TFSF source. The Z-normal monitor was placed only above the pillar. This ensured that only the scattered light not coupled to the substrate was collected. Scattering cross-section (σ) was estimated for each monitor by normalizing the transmission power measured by each of the monitors with the source intensity. The total scattering cross-section was obtained by summing the cross-section values obtained from all the monitors. As shown in Figure 3D, the measured diffused spectrum closely matched the normalized scattering cross-section of the individual pillars. Similar results were obtained for other nanopillar heights (Supplementary material, Figure S5). Thus, the diffused reflectance from the DNPA is primarily determined by the scattering properties of individual pillars. Moreover, the analysis on field intensity profiles of pillars within TFSF boundary (Supplementary material, Figure S6) shows that field intensity maxima occur near the In cap for wavelength close to peak wavelength in the diffused reflectance spectrum. We attribute the diffused reflectance peak to the scattering resonances associated with the In sphere on top of the high-index dielectric nanopillar. Furthermore, simulations comparing the In/InP and InP (both having the same shape) show that the In/InP model (Figure 2B) is in excellent agreement with the measurements (Supplementary material, Figure S5).
The scattering cross-section peaks match well with the peaks in the diffused spectra (Figure 3B, Figure S4). Importantly, the peaks in the diffused reflectance (Figure 3B) do not overlap with the reflection dips observed in the total (Figure 2A) and in the specular (Figure 3A) reflectance spectra. We also performed the TFSF simulations (considering a non-absorbing substrate but with the same refractive index as InP) to determine the scattering into the substrate. Here, the scattering cross-section peak again occurs at the spectral peak position of the diffused reflectance. Considering the above, we believe that the spectral dip (Figures 2A and 3A) must have originated from a different resonant phenomena. The specular reflectance level (Figure 3A) is also clearly higher than the diffused reflectance level (Figure 3B). In the total reflectance (Figure 2A) from the pillars with heights of ~340 or 420 nm, the spectral dip position observed in both the specular and total reflectance has greater influence on the overall coloration. In the metalo-dielectric nanopillars, such phenomena as size-dependent resonant coupling (including absorption) in the dielectric and/or plasmonic nanostructures can be expected to influence the reflectance characteristics. For the metallic nanostructures, the reflectance dips may correspond to the excitation of localized surface plasmon resonances (LSPR) for specific structural dimensions , . Although the self-organized pillars have metal (In)-rich capping, the plasmonic resonances of In fall in the ultraviolet wavelength region , . Thus, for the In/InP DNPA, the direct plasmonic origin for the specular structural colors is unlikely. The observed specular reflectance characteristics are likely due to the size-dependent resonant absorption in the semiconductor nanopillars , , . As the pillar height increases, their diameters also increase from 140 to 220 nm (Figure 1C). As shown in Figure 3A, no clear reflectance dips/peaks are observed for the shortest pillars, whereas a clear dip is seen for the other samples. Pillars taller than 500 nm invariably show more than one reflectance dip (Supplementary material, Figure S7). Studies on silicon vertical nanopillars have shown that hybrid modes (HE11) can be excited and, consequently, the resonant absorption can occur at specific wavelengths depending on pillar diameter . For wider pillars, the resonant absorption also occurs due to the higher order modes and the absorption peak can also broaden for the tapered/conical pillars , . Therefore, we expect such modal properties to be present even in the case of the In/InP DNPA.
To verify the role of the nanopillar size in the resonant absorption, an analysis was performed to investigate the absorption in the In/InP nanopillars. Power absorption in a material depends both on electric field intensity |E|2 and the absorption co-efficient of the material at a given frequency (ω). The total power dissipation in a material within a given volume of V can be evaluated as
We have demonstrated structural colors in reflection from metalo-dielectric (In/InP) DNPA. We realized the DNPA array on the InP substrates using the ion beam sputtering method. With this method, wafer scale uniform color from the In/InP DNPA can be obtained in a single step, as opposed to using wafer scale patterning, metal deposition and subsequent etching. However, to realize the macro-scaled color variations in a given wafer, a multi-step fabrication procedure, which might include masking steps, is necessary. The fabricated nanopillars have In-rich capping on top of the InP nanopillars. The FFT analysis shows that the arrangement of the nanopillars is subwavelength and disordered. The reflective colors from pale blue to dark red can be generated from the DNPA by varying only the sputter duration and within a small range from 30 s to 4 min. The optical measurements show that both specular and diffused reflectance contribute to the color generation. A higher specular contribution in the pillars of smaller heights (<420 nm) leads to glossy colors from these samples. The FDTD simulations on the In/InP pillar model, with supercell approximations on the arrangement of pillars, matches well with the experimentally observed reflectance spectra. Though simulations with supercell approximations is a better approach, simulations using periodic boundary conditions can also be considered to estimate qualitative spectral features in the reflectance profile. By introducing spatial disorder to the periodic arrangement of pillars, we observe that diffused reflectance increases with increasing disorder. Scattering studies using the TFSF simulations show that the spectral peaks in the diffused reflectance are related to the scattering cross-section of the individual pillars. In comparison, the spectral features in the specular reflectance originate from the modal (HE1m-like) coupling of light into the pillars and its absorption within these pillars. Overall, the studies on the In/InP DNPA show that the color features can be characterized by describing the interactions of light with an individual pillar. By controlling the shape and the size of individual pillars, we can effectively tune the spectral behavior of the diffused and the specular components of the total reflectance.
Furthermore, the analysis can be extended to other types of metalo-dielectric DNPA from which spectral colors are expected. Our future studies will focus on improving the range of colors by optimizing the process parameters for pillar formation and by appropriately selecting the refractive indices of the substrate and the surrounding medium. For example, the In/InP DNPA can be fabricated from InP thin films provided on a low-index substrate instead of using the InP substrates. Similarly, the nanopillars can be embedded in a low-index (e.g. 1.5) transparent medium.
Included in the supplementary material are the details regarding dimensions of the nanopillars considered for the simulations (Table S1), the effect of removing the top portion of the nanopillars on the colors (Figure S1), relevant FDTD simulation results (Figures S2–S7) and the chromaticity calculations for the reflective colors.
The authors acknowledge Yohan Désières for useful discussions.
Park JG, Kim SH, Magkiriadou S, Choi TM, Kim YS, Manoharan VN. Full-spectrum photonic pigments with non-iridescent structural colors through colloidal assembly. Angew Chem Int Edit 2014;53:2899–903. CrossrefGoogle Scholar
Park CS, Shrestha VR, Yue W, et al. Structural color filters enabled by a dielectric metasurface incorporating hydrogenated amorphous silicon nanodisks. Sci Rep 2017;7:2556. PubMedCrossrefGoogle Scholar
Kawamura A, Kohri M, Morimoto G, Nannichi Y, Taniguchi T, Kishikawa K. Full-color biomimetic photonic materials with iridescent and non-iridescent structural colors. Sci Rep 2016;6:33984. CrossrefPubMedGoogle Scholar
Kanamori Y, Katsube H, Furuta T, Hasegawa S, Hane K. Design and fabrication of structural color filters with polymer-based guided-mode resonant gratings by nanoimprint lithography. Jpn J Appl Phys 2009;48:06FH04. Google Scholar
Le Roy S, Barthel E, Brun N, Lelarge A, Søndergård E. Self-sustained etch masking: a general concept to initiate the formation of nanopatterns during ion erosion. J Appl Phys 2009;106:094308. CrossrefGoogle Scholar
Murdoch BJ, Portoles JF, Tardio S, Barlow AJ, Fletcher IW, Cumpson PJ. Visible wavelength surface-enhanced Raman spectroscopy from In-InP nanopillars for biomolecule detection. Appl Phys Lett 2016;109:253105. CrossrefGoogle Scholar
Lee WJ, Senanayake P, Farrell AC, Lin A, Hung CH, Huffaker DL. High quantum efficiency nanopillar photodiodes overcoming the diffraction limit of light. Nano Lett 2015;16:199–204. PubMedGoogle Scholar
Lumerical Inc., https://www.lumerical.com/, (Accessed June 8, 2019).
Palik ED. Handbook of optical constants of solids. Cambridge, MA, USA, Academic Press, 1998. Google Scholar
Adachi S. Optical constants of crystalline and amorphous semiconductors: numerical data and graphical information. New York, USA, Springer Science & Business Media, 1999. Google Scholar
The online version of this article offers supplementary material (https://doi.org/10.1515/nanoph-2019-0171).
About the article
Published Online: 2019-08-14
Funding Source: Swedish Research Council
Award identifier / Grant number: 349-2007-8664
Award identifier / Grant number: 621-2014-5078
Funding Source: Swedish Energy Agency
Award identifier / Grant number: 45199-1
The funding for this work came from the Swedish Research Council (Vetenskapsrådet; Grant Nos.: 349-2007-8664, Funder Id: http://dx.doi.org/10.13039/501100004359 and 621-2014-5078, Funder Id: http://dx.doi.org/10.13039/501100004359) and the Swedish Energy Agency (Energimyndigheten; grant: 45199-1, Funder Id: http://dx.doi.org/10.13039/501100004527).
Citation Information: Nanophotonics, Volume 8, Issue 10, Pages 1771–1781, ISSN (Online) 2192-8614, DOI: https://doi.org/10.1515/nanoph-2019-0171.
© 2019 Ajith P. Ravishankar and Srinivasan Anand et al., published by De Gruyter, Berlin/Boston. This work is licensed under the Creative Commons Attribution 4.0 Public License. BY 4.0