Effect of graphene oxide inclusion on the optical reflection of a silica photonic crystal film

In this study, the inclusion of graphene oxide in silica photonic crystals was found to affect optical reflectance intensity and reflectance peak broadening. The quantitative relationship between weight percentage and the reflected light intensity and corresponding wavelength shift of light GO-decorated photonic crystals was studied, providing a useful parameter in the rational design of antireflection coatings for GO-based photonic crystal films. Comparison of the experimental results with a pure SiO2 particle film shows that a SiO2 particle surface layer incorporated with a fixed graphene oxide weight percentage results in broadening of the peak and a decrease in reflectance intensity. The percentage of the reduction in reflectance intensity is a function of particle size, as indicated by the structured color film surface, demonstrating the possibility of estimating the effect of different graphene oxide inclusion percentages on the antireflection properties of photonic crystal films.


Introduction
Photonic crystals in colloidal systems are arranged in periodic structures. These periodic structures are commonly accompanied by color reection, depending on the periodicity. 1,2 The colors generated from periodic physical structures are important, not only in nature but also for understanding the optical band gaps of photonic crystals. [3][4][5][6] Visible light of specic wavelengths can be prevented from propagating in the photonic crystal structure by controlling the photonic band gap in the visible light range of 380 to 780 nm. Selective light with a specic wavelength reected via the interaction between physical structure arrays and incident light is dened as structural color. Structural color reection can be formed by dispersion, scattering, interference and Bragg diffraction, 7-10 without any chemical colorants or pigments.
Recently, black materials have been used as additives for fabricating photonic crystals with interesting optical properties. 11,12 General photonic crystals made from polystyrene (PS), polymethylmethacrylate (PMMA) or silicon dioxide (SiO 2 ) nanoparticles used as building blocks exhibit iridescent structural colors with limited potential applications. Black materials like carbon-based materials provide a scattered light-absorbing background that can contribute to the increased coloration and antireection properties of the material. In the case of opal-type structures, it has recently been shown that the incorporation of a light absorbing agent can enhance the structural coloration of these materials. [13][14][15][16][17][18] The published literature provides evidence that carbonbased nanomaterials can be used as effective scattered light absorbers in a photonic crystal matrix. The inclusion of this absorber species prevents the scattering of stray light and limits the reected light to coherent light generated by the stop band alone. 19 Pursiainen et al. reported that sub-50 nm carbon nanoparticles uniformly incorporated in the interstices of highly ordered polymeric colloidal crystal lms enhanced the color strength of these elastomeric lms. 20 Wang et al. observed an enhancement in interference colors aer applying a thin carbon layer onto an anodic aluminum oxide lm with well aligned nanochannels. 21 Aguirre et al. observed that white opalescent poly(methyl methacrylate) colloidal crystals became uniformly and strongly colored aer the incorporation of carbon black nanoparticles which acted as absorbers within the structure. 15 The color reection could be tuned by changing the size of the colloidal spheres, and its intensity depended on the carbon dopant loading, up to a controlled weight percentage of carbon, above which the materials became more optically absorptive.
Among the carbon-based dopants used for the improvement of optical properties, an important derivative of graphene, graphene oxide (GO), has become one of the most studied nanomaterials over the past decade, due to its stability under ambient conditions 22 and controllable reection or transmittance in the wide electromagnetic spectrum, ranging from UV to the near IR region. 23 Compared with graphene, GO has many unique advantages. First of all, the optical properties of GO can be tuned dynamically by manipulating the content of oxygen-containing groups, through either chemical or physical reduction methods. 24,25 Secondly, GO can be chemically synthesized and is solution processable. This is benecial for low cost and scalable integration methods, such as spin coating and spraying methods, applicable for the proposed antireection or light harvesting layer formation. Therefore, it is expected that a high performance, simple and low cost antireection coating with selected colors can be developed using a selfassembled SiO 2 particle lm with incorporated GO.
Li et al. reported that GO-based sheets in an aqueous solution can form photonic liquid crystals, which display tunable colors at various GO concentrations. 26 In addition, Eda and coworkers observed that ultrathin lms of reduced GO (thicknesses ranging from a single monolayer to several layers, i.e. around 1-5 nm) appeared almost transparent. 27 Structural coloration, especially color reection in a GO dispersion, is quite unusual, considering that graphene oxide is highly polydispersed, irregularly shaped, curved in solution 28,29 and two orders of magnitude thinner than photonic band-gap materials. The basic color reection mechanism of GO-doped photonic crystals at a submicron scale, however, is not well understood. From a practical point of view, silica-based photonic crystals potentially allow the use of graphene oxide in various optical applications, as long as an easy method is provided for the manipulation of optical reection. 30 Inspired by these interesting observations, we speculated about which factors might cause a change in color intensity with the adjustment of the percentage of graphene oxide and any signicant change in color reection intensity upon the incorporation of GO into the photonic crystal array. The effect of GO inclusion on the anti-reection properties of photonic crystals is still in the theoretical stage and the fabrication of graphene oxide-based photonic crystals with tunable optical reection and reected wavelength shi properties remains of interest for further investigation.
The optical reectance results of GO-doped silica photonic crystal lms with various particle sizes have conrmed that the amount and distribution of graphene oxide as a scattered light absorber on a SiO 2 photonic crystal lm reduces the color reection intensity and reected peak broadening in terms of the Full Width at Half Maximum (FWHM) value as well as the shi in reection wavelength, rather than the grain size effect. Due to a lower extinction coefficient value compared to other carbon-based nanomaterials, the reectance intensity in terms of the percentage of graphene oxide inclusion can be adjusted and is applicable in a range of particle sizes.

Materials
The chemical reagents used in these experiments are: tetraethyl orthosilicate (TEOS) (99.0%) purchased from Sigma-Aldrich Co., LLC; ammonia (NH 3 , 25% in H 2 O) and ethanol (EtOH, 99.9%) from Fisher Scientic Co., Ltd., UK; graphene oxide (4 mg mL À1 dispersion in H 2 O) purchased from Sigma-Aldrich Co., LLC; distilled water (H 2 O, distilled by a USF-ELGA water purier), which was dispensed from the laboratory facility. All of the materials were used as received without any further purication.

Preparation of the silica photonic crystal lm
Uniform silica nanoparticles (SNPs) with diameters ranging from 207 nm to 350 nm were synthesized based on the Stöber Scheme 1 Fabrication procedure of the ordered silica particle films with graphene oxide inclusion. Graphene oxide (GO) contains a mixture of epoxy, hydroxyl, carbonyl and carboxyl groups on a carbon plane framework.
method. [31][32][33][34][35] The study on the effects of TEOS, ammonia water and water on the synthesized particle sizes was described in the ESI. †

Preparation of the silica photonic crystal lm with graphene oxide inclusion
The stock solution of graphene oxide (GO) used in the preparation of the silica photonic lms mixed with various weight percentages was obtained as follows: 0.3127 mL of GO solution (4 mg mL À1 as purchased) was dispersed in 2.5 mL of distilled water, to obtain a 0.04447 wt% GO dispersion. For the preparation of various weight percentages of GO in a silica emulsion, 53.94 mL, 89.9 mL, 125.86 mL and 153.5 mL of freshly prepared GO stock solution were mixed with 666.7 mL of 12 wt% silica emulsion. Additional quantities of distilled water were added into the mixtures containing GO and the silica emulsion to reach a nal volume of 1 mL of graphene oxide containing silica solutions of 0.03%, 0.05%, 0.07% and 0.1%, respectively, and these were subsequently used to prepare the photonic crystal lms. The GO-silica solution mixture was then dropped on preheated glass in an oven which was maintained at 80 C. Aer 25 min, the solvent was evaporated and the GO-silica photonic crystal lm was obtained (Scheme 1).
Characterization of the silica and graphene oxide-silica photonic crystal lms Scanning electron microscopy (SEM, JSM-6490, JEOL Co. Japan) was used to observe the morphology of the silica photonic crystal lms. To reduce charging effects, the samples were sputter-coated with a thin layer of gold ($5 nm) prior to observation.
FT-Raman spectra were recorded between 1800 cm À1 and 300 cm À1 using a Bayview Raman spectrometer equipped with an optical microscope. SiO 2 photonic crystal lms, with and This journal is © The Royal Society of Chemistry 2018 without 0.05 wt% graphene oxide inclusion, were irradiated with a CW green laser at 532 nm and 20 scans were captured for each sample. The spectra of silica nanoparticle lms were acquired using the micro-sampling mode with a 10Â objective lens and an output laser power of 150 mW. The scattered radiation was collected at 180 and the spectral resolution was set to 4 cm À1 .
The mean particle size and polydispersity of the synthesized silica nanoparticles were measured and calculated using a particle size analyzer (Zeta Plus, Brookhaven Instruments Corp., USA). The color reection and chromaticity coordinates (L*a*b* color chart) of the lm samples were determined and analyzed using an SF 650 spectrophotometer (DataColor International, USA). Fig. 2 Particle size distribution (number count vs. particle size (diameter)) measured using a zetasizer with the mean particle sizes of (a) 198. 8

SEM observation of the silica nanoparticle lm
The drop-coating method enabled the silica colloids to assemble into highly periodic arrangements with a facecentered cubic (fcc) crystal driven by capillary force under slow water evaporation. It was observed that the SNPs were spherical in shape and uniform in size (the polydispersity index value was close to that of a monodispersed particle). Using the evaporation-induced self-assembly method, the arrangement of SNPs represents the face-center cubic (fcc) structure in the h111i growth direction, 1,15 which can be seen in Fig. 1(a-e).
The size (diameter) of the silica nanoparticles measured using both the particle size analyzer and the SEM micrographs is expressed in Table 1. The mean particle size was obtained aer the measurement of the diameters of at least 20 particles or more from the micrographs. However, the differences in size determination using SEM and the zetasizer should, in practice, be considerably smaller because diameters measured using electron microscopy are an average between the major axis and the axis perpendicular to it, and the size data obtained using the zetasizer is orientation-averaged due to the rotational motion of the particles. However, the hydrodynamic diameter measured using the zetasizer is inuenced by interactions between the particles and the dispersion media. This indicates that a few aggregates were present in the solution. The aggregates due to particle interaction observed using SEM could be attributed to the drying process during sample preparation. However, the particle size values measured using the particle size analyzer showed varied size distribution with an uncertainty of approximately 40 nm (Fig. 2).
Effect of optical reectance with respect to graphene oxide weight percentages We further investigated the color reectance of graphene-oxide on a SiO 2 layer with a mean particle size of 198.8 nm. Fig. 3 shows the reectance of a 198.8 nm diameter particle as a function of graphene oxide percentages. However, the reectance of the graphene oxide exhibited observable oscillations in intensity.
Oscillations of the reectance intensity, which accompany the main Bragg peak in various dopant concentrations (Fig. 3), appear because of the interference of the light between the color reection from the photonic crystal surface and changes in the layer periodicity of the graphene oxide sheet. We interpret these oscillations, which are observable for the aperture size of 6.6 mm in diameter, as evidence of good ordering of the measured color reectance intensity.
The silica particle size (diameter) used for GO is 198.8 nm. The optical reectance peak values were obtained from the weight percentage of GO, ranging from 0 wt% to 0.1 wt%. The reectance intensity decreased when the GO dopant percentage increased, up to 0.05 wt%. This attenuation in reectance intensity implies that the graphene oxide is involved in the absorption of scattered light at the stop band region and the arrangement of the graphene oxide sheets is more likely to be in the isotropic phase. When the GO weight percentage was above 0.05%, the color reection intensity increased to 60%, 20% more than that of the pure SiO 2 photonic crystal lm, indicating that graphene oxide tends to be aligned from the isotropic to the nematic phase. This indicates that as the dopant weight percentage reached 0.07 wt%, the quality of the lamellar structure of graphene oxide distributed in the photonic crystal lm became critical for brilliant color with maximum reectance intensity. As the dopant weight percentage further increased to 0.1%, the reectance intensity became lower, indicating the deterioration of the periodic lamellar structure of graphene oxide. Overall, the peak became broader in terms of the FWHM and the corresponding reectance intensity became lower, implying that periodicity can be adjusted by the graphene oxide (GO) dopant percentage in the photonic crystal lm. Consequently, the peak width and color reection intensity decreased in the stop band close to the level of the passband due to the intensive absorption of light in the whole visible region. Fig. 3 reveals that the optimum GO weight percentage for the highest color strength (lowest color reectance intensity) was 0.05 wt%. Apparently, the optical reection property of the crystal lm can be effectively manipulated by adding an appropriate amount of graphene oxide. This GO weight percentage, based on the result of minimal reection (maximum white light absorption), was used as the optimal dopant amount to study the color reection of various sizes of silica nanoparticle.
Effects of optimal graphene oxide content on the reectance percentages and peak proles of photonic crystal lms with various particle sizes To evaluate the possibility of GO functioning as an anti-reection lm, optical reectance measurements were carried out using an SF 650 spectrophotometer (DataColor International, USA). Fig. 4 shows the reection spectra of a silica photonic crystal lm with and without graphene oxide The relative stop-bandwidth (Dl/l 0 ), where Dl is the full width at half maximum (FWHM) of the peak from the Bragg resonance at normal incidence and l 0 is the center wavelength value of the peak, can be determined. All measured reectance peaks are baseline tted before measuring the FWHM (Dl) and peak values (l 0 ). The relative stop bandwidth (Dl/l 0 ) for various particle sizes ranged from 4.7% to 6.7% ( Table 2). The bandwidth for the particles with mean diameters of 198.8 nm and 224.3 nm matches closely with the 6.3% FWHM of the [111] crystalline plane with a relative gap width calculated by the plane wave method. 27 For particles with diameters of 232 nm to 288.2 nm, the stop bandwidth is larger than the nominal value of the hexagonal crystalline plane (6.3%). This large deviation in relative stop bandwidth could be due to the dispersion of the colloidal silica suspension, leading to contribution from defects, cracks and disorder occurring during the self-assembly of the photonic crystal structure, thus limiting the domain size of the single-crystalline nature of the lm. 27 The reectance spectra show that the graphene oxide doped silica photonic crystals result in a reduction in the reectance value as a function of mean particle diameter along the 400-700 nm wavelength range. For a photonic crystal lm with a larger mean particle size, more crystal defects and disorder would have been induced during self-assembly, which reduce the efficiency of the photonic stop band. Less ordered particle alignment results in a reduction of the peak reectance value at higher percentages and a broadening of the stop band peak width. Graphene oxide in a SiO 2 particle lm with particle sizes from 198.8 nm to 288.2 nm reduces the reectance intensity by approximately 17-45%, within the 400-700 nm range (Table 2). It is feasible that the inclusion of GO reduces the loss of light from the silica photonic crystal, resulting in better light harvesting properties. Photonic crystal lms with a larger particle size, for example 288.2 nm, have different color reection behavior compared to that with a smaller particle size, such as 198.8 nm. The volume occupied by particles of a larger size is signicantly smaller than the volume occupied by smaller particles. In effect, larger particles are packed like a polydispersed sample with more voids and occupy a lower percentage of the substrate surface compared to smaller particles. However, particles with a larger size (average diameters of 258.7 nm and 288.2 nm) are less likely to be packed in a photonic crystal array with long range order. This reduction in reectance values and broadening of the peak may be attributed to the imperfect alignment of graphene oxide distributed on the photonic crystal. The smaller particle size provided a better packing density with less disorder than that of the larger particle size, which was identied from the FWHM value of the reected peaks ( Table 2). The asymmetry in the shapes of the reection spectra is due to domain defects in the crystalline structure of the photonic crystal. Due to light scattering at these defects, they also lead to the broadening and attening of the spectral features. 43 This investigation indicates the decisive role of graphene oxide in the photonic crystal structure, and thereby of the deposition scheme, in the exploitation of the color reection characteristics of silica photonic crystals, with and without dopant.
From Table 2, it can be inferred that 0.05 wt% graphene oxide doped SiO 2 particle lms absorb even more scattered photons with respect to increasing particle size, exhibiting enhancement of their anti-reection properties.
However, the GO layer was not very smooth and may be wrinkled, as observed using TEM. Consequently, the ARC effect of the GO could be improved upon homogeneous mixing in the SiO 2 particle lm. It is also plausible that the rough surfaces may contribute to internal reections, which may trap light and contribute to variations in the reectance.
It is notable that the ARC effect of GO persists over a relatively large wavelength range (from 300-1000 nm). Such a broadband response may arise due to the ake-like structure of GO mimicking a broad range of surface roughnesses with a wide range of optical parameters present in graphene oxide inclusion. However, the refractive index of GO would be expected to change with the surface roughness, which would be a function of the thickness. The inuence of GO distribution on the optical response would be a promising topic of investigation, with relevance to its inuence on optical limiting efficiencies for applications such as antireection coatings.
Raman spectroscopy, a powerful tool, was used to identify graphene oxide inclusion in a non-destructive strategy, to examine the chemical signature of the graphene oxide in the silica photonic crystal lm. As shown in Fig. 5, the most pronounced band at 470 cm À1 originated from oxygen atom vibrations with identical distortions of neighboring Si-O bonds, 36 i.e. symmetric stretching vibrations. 37 The band at 800 cm À1 is also associated with the symmetric stretch vibrations of oxygen atoms, but these vibrations involve a substantial amount of surrounding Si atoms. The band at 1058 cm À1 is similar to those associated with the Si-O transverse and Table 2 Full Width at Half Maximum (FWHM) and reflectance peak intensity (%) of pure SiO 2 and 0.05% graphene oxide doped SiO 2 particles at a stopband with particle sizes ranging from 198.8 nm to 288.2 nm   This journal is © The Royal Society of Chemistry 2018 longitudinal optical modes. 37,38 The peaks identied as the longitudinal optical mode are characteristic of the Si-O network rather than defects or impurities. It was found that the Raman prole of the GO-decorated silica particle lm exhibited sharp peaks around 1346 cm À1 and 1598 cm À1 , which correspond to the well-dened D and G bands of carbon-based materials. 39 The D band is a defect-induced Raman signature observed due to disorder or defects at the edge of graphene oxide. The G band is known to be due to sp 2 carbon networks in the graphene oxide doped silica photonic crystal lm. 40 In addition, a 2D peak at 2700 cm À1 was observed. 41 The 2D peak originates from a second-order Raman process and can be used to determine the thickness of graphene oxide layers. The intensity ratio (I 2D /I G ) determined from the spectrum, which is much lower than 1.9, indicates the formation of a multi-layered graphene oxide sheet structure in the sample. 42 When 0.05 wt% of graphitic oxide is incorporated into the lm, the spectrum of the silica nanoparticle lm reveals not only the strong vibrational bands of the Si-O-Si network in the lm but also the D and G bands of graphene oxide inclusion, indicating that the graphene oxide sheet has been distributed near the silica particle lm surface.
More quantitatively, color reection due to graphene-oxide inclusion can be evaluated by calculating the difference between the L*a*b* parameters of pure SiO 2 and SiO 2 doped with 0.05 wt% graphene oxide. Table 3 summarizes the L*a*b* values of the crystal lms with silica spheres of 198.8, 224.3, 232, 258.7 and 288.2 nm in average hydrodynamic diameter with 0.05 wt% of graphene oxide dopant. In order to identify the color difference aer the doping of graphene oxide, the L*a*b* color space was modeled aer a color-opponent theory, stating that two colors cannot be red and green at the same time or yellow and blue at the same time. As shown below, L* indicates lightness, a* is the red/green coordinate and b* is the yellow/blue coordinate.
As illustrated in Table 3, the values for the particles with average diameters from 198.8 nm to 232.0 nm increased towards the positive x-axis in the color chart, indicating a tendency towards red shi. However, a green shi (towards the negative x-axis) was observed as the average particle diameter further increased from 232.0 nm to 288.2 nm. The b values for the particles with average diameters from 198.8 nm to 288.2 nm decreased towards the negative y-axis, representing a blue shi. This blue shi of the Bragg resonance in the reectance spectra occurs due to the decreasing of the lling ratio of the graphene oxide as particle size increased. This factor serves to increase the photonic stop-bandwidth (Dl/l 0 ) as the particle diameter increased, as indicated from the measured FWHM data shown in Table 2.

Conclusions
In this study, we found that GO can improve antireection and color strength properties and can therefore serve as a promising material for the manipulation of structural color reection within photonic materials. By adjusting the percentage of graphene oxide inclusion and various silica particle sizes, we successfully produced GO-modied silica photonic crystals with tunable antireection of colors, which is totally different from the traditional view that GO can only appear as transparent or dark brown. More importantly, the results demonstrate the potential applications of GO in color related elds such as light harvesting materials, textiles, paint and thin lm devices with optical limiting properties.

Conflicts of interest
There are no conicts to declare. The L* values of the crystal lms do not change signicantly with respect to particle size. This indicates that lms of all particle sizes exhibit a darker color compared with the pure SiO 2 particle lms which have no obvious change in lightness upon the incorporation of graphene oxide. The a* and b* values indicate the chromatic change in terms of the designated color coordinates and the percentage distribution of the three primary colors.