1. Introduction

Materials exhibiting structural colors are prevalent in nature and have been gaining attention as alternatives to chemical colors [1]. Unlike chemical colors that primarily rely on light absorption, structural colors are based on light interference. One advantage of structural colors is their resistance to fading over time. Moreover, they can be made from primary materials such as silica or cellulose, which are non-toxic, widely utilized, and often produced sustainably [2]. However, structural colors can exhibit iridescence in the case of crystalline assemblies [3], or they may lack strong and vibrant color appearance in amorphous structures. Nevertheless, discussions are currently underway regarding applications in coloring everyday objects [4], in the food industry [5], in paint formulations [6], and in displays [7].

Numerous studies have focused on structural colors by assembling colloids made from materials such as silica or polystyrene [4,8], with the particle size being approximately half the wavelength of light. However, achieving a red color tone is challenging through direct colloidal assembly. It has been suggested that inverse structures offer improved structural colors [9–11] and enable the creation of red hues [12,13]. An additional important criterion for practical applications is the ability to use structural colors as paints or inks, which requires dispersing the color as a pigment, for example, in inkjet printing [14] or in preparation of paint formulations. In this study, we present the fabrication of inverse foam-like photonic spheres that exhibit colors in the visible spectrum and could replace chemical color pigments. We employ a recently developed microgel templating method to achieve this, where polymer microgels are initially decorated with much smaller silica nanoparticles [13,15]. The photonic spheres are created through solvent drying, followed by the removal of the polymer through calcination.

2. Experimental

2.1 Synthesis and assembly of silica-coated microgels, and color generation from photonic spheres

An important feature of our samples is their foam-like internal structure, which we accomplish by exploiting the inherent softness of microgel colloids. By compressing the microgels, which are surrounded by silica nanoparticles, we create foam-like silica structures with flat walls between the cavities. To achieve this, we synthesized thermoresponsive microgel particles composed of Poly-(N-isopropyl acrylamide) (PNiPAm), crosslinked with N,N’-methylenebis-(acrylamide) (BIS), and decorated with much smaller silica nanoparticles. For detailed synthesis protocols, we refer to [13,15]. Briefly, the synthesis procedure unfolds in two stages. Initially, microgel particles are synthesized using NiPAm and BIS monomers, and a water-soluble initiator, Potassium Persulfate (KPS), as described in [16]. Subsequently, a positively charged co-monomer, N-(3-aminopropyl)methacrylamide hydrochloride (APMA), is introduced along with negatively charged silica nanoparticles (LUDOX AS-40, approximately 20 - 24 nm in diameter, dispersed in an aqueous medium). The electrostatic attraction between the co-monomer and the silica nanoparticles drives the binding between microgel particles and silica nanoparticles, resulting in silica-decorated microgels, Fig. 1(a). The microgel particles decorated with silica nanoparticles are mentioned as primary particles in the later part of the text.

 

Fig. 1. Fabrication of photonic spheres. (a) Microgel particles coated with silica nanoparticles are dispersed in water, and emulsified in mineral oil using a vortex mixer, aided by the surfactant Span 85 (HLB-2). The water phase contains approximately 5 - 10 wt% microgel particles. The water slowly diffuses into the mineral oil, causing droplet contraction and the creation of densely packed particle arrays. Upon drying and subsequent calcination at 500 $^\circ$C, photonic spheres are produced. (b) Scanning electron microscopy (SEM) image of a dried microsphere before calcination. (c) Enlarged view of the microsphere, revealing the features of individual microgels with compact assembly of silica nanoparticles on the surface. (d) SEM image of a photonic sphere post-calcination. (e) Interior view after focused ion beam (FIB) milling of a calcined photonic sphere.

Download Full Size | PDF

We generate a suspension of microspheres through emulsification. In this process, we combine 15 $\mu L$ of an aqueous dispersion containing silica-coated microgel particles with a concentration of 5 - 10 wt.%, and 4 $\mu L$ of non-ionic surfactant (Span-85) with 3 $mL$ of mineral oil. Emulsification and microsphere production are carried out using a vortex mixer (Vortex-Genie 2, Scientific Industries, USA) set at speed No. 10, corresponding to approximately 3200 RPM. We chose Span-85 as the emulsifier due to its low hydrophilic-lipophilic balance (HLB) value of 1.8, which means that it has a high affinity towards the oil phase, allowing effective emulsification with high mineral oil content. We transfer some of the emulsion onto a high surface area substrate to facilitate water removal. During drying, which takes 12 - 15 hours, water diffuses slowly from the microspheres into the excess mineral oil. This gradual diffusion of water causes the microgel-laden droplets to contract, resulting in densely packed microgel/silica structures, Fig. 1(a). The excess mineral oil is removed by washing with heptane and drying, which leads to the formation of the dried microspheres. The microspheres produced are in the size range of 8 - 20 $\mu m$. The next step involves calcining the microspheres in a tubular oven at 500 $^\circ$C. Heating at this high temperature removes the organic polymer, resulting in the formation of solid silica spheres commonly known as photonic spheres or photonic balls. These terms are often used to describe colloidal microspheres specifically designed to exhibit structural coloration. We suppose that no visually relevant traces of carbon remain in the sample following the calcination process. The photonic spheres are only slightly larger than the scattering mean free path. This prevents internal multiple scattering [17] and allows them to exhibit color without needing carbon as a color-enhancing absorber [18].

We utilize primary particles of three distinct sizes to fabricate three different types of photonic spheres, namely $\text {PB}_1$, $\text {PB}_2$, and $\text {PB}_3$. All the particles possess the same crosslinking density as we maintained the same NiPAM:BIS ratio. The main particle hydrodynamic radii $R_H \pm$, determined through dynamic light scattering (DLS) at 20 $^\circ$C, are $394 \left ( \pm 20\right )$ nm, $581 \left (\pm 38\right )$ nm, and $633 \left (\pm 46 \right )$ nm, respectively, with the standard deviation provided in parentheses. To compare, when fitting static light scattering (SLS) data using the fuzzy-sphere model [19], which characterizes the form factor of microgels (Supplement 1), we derive sizes of $397 \left ( \pm 28\right )$ nm, $537 \left (\pm 64\right )$ nm, and $623 \left (\pm 50 \right )$ nm, respectively.

To summarize the naming of the different colloidal building blocks:

2.2 Characterization techniques

We employed several techniques to characterize the materials involved in fabricating structurally colored photonic spheres: light scattering to determine the hydrodynamic radius of the primary silica-coated microgel particles, Scanning Electron Microscopy (SEM) to explore the surface aspects of photonic spheres, Focused Ion Beam (FIB) milling to assess the internal nanoscale features of the photonic spheres, and optical microscopy along with spectrometry to determine the color spectra of the photonic spheres. All these techniques are detailed in the Supplement 1.

3. Results and discussion

Figure 1(b) shows a dense packing of silica-coated microgels into a spherical assembly. In the enlarged view of the sphere, Fig. 1(c), individual silica nanoparticles are clearly seen on the surface of the microgel particles. In our understanding, during the calcination of the spheres, the organic polymer is removed and the silica nanoparticles are pushed towards the periphery, thus forming the cell walls of the foams. This process gives our photonic spheres the inverted and porous structure, which is exhibited in Fig. 1(d). The internal nanoscale characteristics of the porous photonic spheres are explored using FIB milling and imaged using SEM. The cross-section reveals a porous foam-like arrangement, Fig. 1(e).

We note that to achieve silica networks assembled from relatively small silica nanoparticles, a high osmotic pressure is usually required. Prior work [13] adopting this method to larger volumes has revealed notable irregularities and color gradients due to this challenge. Confining the colloids within a spherical geometry initially overcomes and intrinsically simplifies this process, resulting in well-shaped structures.

The calcined photonic spheres $\text {PB}_1$, $\text {PB}_2$ and $\text {PB}_3$ cover structural colors with green and reddish hues as shown in Fig. 2(a)-(c). To get a quantitative understanding of the color perceived by the eye, we performed polarization-dependent reflection measurements of the photonic spheres, shown in Fig. 3. Single and low-order scattering contribute to the Co-polarised state ($C_p$) (Fig. 3(a)), maintaining all or most of the polarisation state of the incident, linearly polarised light. The cross-polarised state ($C_r$) (Fig. 3(b)) contains almost exclusively multiply scattered or diffuse light. The normalized difference of these reflectance spectra, ($C_p - C_r$) (Fig. 3(c)), is a rough measure of single scattering from the photonic balls.

 

Fig. 2. Reflection images of photonic spheres made using three different primary microgel particles observed using optical microscopy under bright field illumination with a 50$\times$ objective (0.8 NA). Panels: (a) $\text {PB}_1$, $R_\text {H}=394$ nm (b) $\text {PB}_2$, $R_\text {H}=581$ nm, and (c) $\text {PB}_3$, $R_\text {H}=633$ nm. The colors in the insets are determined using the experimental reflection spectra shown in Fig. 3 as described in the text. The calculated colors correlate with the colors seen under an optical microscope. (d) Chromaticity diagram utilizing the CIE 1931 2-degree Standard Observer. The color coordinates obtained from the color calculation are depicted on the chromaticity diagram. The circles on the diagram correspond to the calculated colors, with the associated structural lengths $d$ from Fig. 4 indicated.

Download Full Size | PDF

 

Fig. 3. Reflection spectra of the three different structurally colored photonic spheres $\text {PB}_\text {1}$, $\text {PB}_\text {2}$ and $\text {PB}_\text {3}$, recorded using bright field microscopy with a 50$\times$/0.8 NA objective averaged over five spots for each sample. (a) Spectra for co-polarized light reflection, $C_p$. (b) Spectra for cross-polarized light reflection, $C_r$. (c) Difference between co-polarized and cross-polarized reflection. (d) Peak-normalized reflection spectra $C_p - C_r$ plotted against the ratio of the wavelength $\lambda$ to the structural length scale $d$ extracted from the radial distribution functions shown in Fig. 4(d).

Download Full Size | PDF

To understand the origin of the structural colors quantitatively, we relate the structure-induced reflection peak to the length scale of the structural correlations, $d$. By examining the FIB cross-sections for our samples, shown in Fig. 4(a)-(c), we gain insight into the structural correlations present. Through image analysis, we extract the radial distribution function $g(r)$ (see Fig. 4(d) and Supplement 1), wherein the peak position represents a structural length scale, denoted as $d$. Prior research, Ref. [18] and others, has demonstrated that this length scale is anticipated to correlate with the measured structural color by considering single scattering and the effective refractive index ($n_\text {eff}$) of the material [20]. We binarized the FIB cross-sections and calculated the area fraction of bright to dark pixels as a measure of the filling fraction of the silica walls $(\Phi _{silica})$. From the $g(r)$ and the filling fraction analysis, the [d $\pm$ SD, $\Phi _{silica}$] values we obtained are [291 $\pm$ 1.8 nm, 0.45], [298 $\pm$ 16 nm, 0.46], and [328 $\pm$ 10 nm, 0.48] for the photonic spheres $\text {PB}_\text {1}$, $\text {PB}_\text {2}$ and $\text {PB}_\text {3}$ respectively. We observe that the $d$ and $\Phi _{silica}$ values of $\text {PB}_1$ and $\text {PB}_2$ are quite close, which explains why both have colors belonging to a green hue. It is intriguing that even though the primary particles employed in forming photonic spheres $\text {PB}_1$ and $\text {PB}_2$ differ significantly, their ultimate colors are remarkably alike. We attribute this phenomenon to the variations in densification during solvent drying, influenced by factors such as waiting time and silica coverage.

 

Fig. 4. Analysis of the interior composition and structure of three photonic spheres after calcination, using three distinct sizes of primary particles in the template: (a) to (c) Electron micrographs obtained through Focused Ion Beam (FIB) milling of the spheres, followed by imaging using Scanning Electron Microscopy (SEM). The insets display images of the corresponding photonic spheres, copied from Fig. 2. (d) Radial distribution functions around the pore centers, denoted as $g\left (r\right )$, derived from cross-sectional images of the photonic spheres. The legend displays values corresponding to the mean structural length obtained from the peak position of $g(r)$.

Download Full Size | PDF

We plot the reflection spectra $C_p-C_r$ divided by their respective peak values against $\lambda /d$ in Fig. 3(d) where we extract the structural correlation length $d$ from the data shown in Fig. 4(d). These normalized spectra collapse at a $\lambda _\text {max}/ d$ value of $\approx$ 1.9. We compare the $\lambda _\text {max}/ d$ value obtained from the spectra to the calculated $\lambda _\text {max}/ d$ from the structure factor peak of a disordered dense packing of spheres, $q_\text {max}$, using $q_\text {max} \approx 2.2\pi /d$ [21] and $\lambda _\text {max} = 4\pi n_\text {eff} / q_\text {max}$, where $n_\text {eff}$ is the effective refractive index of the material. Combining these equations we derive $\lambda _\text {max}/ d$ $\approx$ 2.03, where $n_\text {eff}\approx 1.12$ is computed using the two-step Maxwell Garnett equation as outlined in [13], using our measurement of the volume fraction of the silica walls described above, $\Phi _{silica} \sim 0.45$, and assuming a volume fraction of nanoparticles within the silica walls of 64%. Hence, our measurements align well with the theoretical forecast for the connection between structure and color.

All specimens display noticeable structural coloration. The presence of larger primary particles results in a reflection peak that shifts towards the red color spectrum. Although the color responses in the green from sample $\text {PB}_\text {2}$ and red from sample $\text {PB}_\text {3}$ in (Fig. 3(c)) are similar, the green reflection spectrum from sample $\text {PB}_\text {1}$ exhibits a more distinct peak. This supports previous findings that obtaining vibrant structural colors at longer wavelengths poses greater difficulties. Concerning red tones, there is frequently an unintentional introduction of blue color components due to increased scattering of individual particles at shorter wavelengths [11]. Contamination from wavelengths surpassing the peak wavelength usually poses no issue due to transparency within this wavelength range. This finding is connected to the increasing homogeneity (or hyperuniformity) of densely packed particle assemblies [22]. Despite a slightly diminished color response compared to the green sample, our red response remains distinct, providing further support that network structures can yield superior structural colors compared to direct arrangements of colloidal spheres in a matrix of lower refractive index (e.g., air).

We use the recorded reflection spectra to calculate RGB colors [23]. These simulated colors align closely with our empirical (visual) observations, as shown in the top right insets of Fig. 2(a-c). Furthermore, utilizing the recorded reflection spectra, we determine the chromaticity coordinates and position them on the chromaticity diagram following the CIE 1931 standard (Commission Internationale de L’Eclairage) [24], as shown in Fig. 2(d). The RGB values obtained from the photonic spheres $\text {PB}_1$, $\text {PB}_2$, $\text {PB}_3$ are (106,177,155), (167,161,149), (182,155,139), and the colors are named as tradewind, dawn, and del Rio according to a freely available software [25]. The corresponding HSV (Hue, Saturation and Value) values are (81,102,177), (20,27,167) and (11,61,181) for $\text {PB}_1$, $\text {PB}_2$, $\text {PB}_3$ respectively, calculated using python with the colorsys package of the openCV library [26].

3.1 Not just spheres: Influence of the water removal kinetics on the shape of the photonic sphere

Although our aim was to create spherical photonic structures by varying the solvent composition, we also encountered other intriguing shapes resembling a doughnut and a fried egg pattern. We believe our observations are linked to research on the drying process of colloidal droplets on diverse surfaces [27,28]. To assess how the kinetics of water removal affect the ultimate shape of the photonic spheres, we modified the composition of the oil phase to regulate its hydrophilicity (refer to section 2.1). For an accelerated drying process, we combined decanol and mineral oil in a 1:5 ratio with primary particles as used for fabricating PB₃ and span-85 surfactant. After overnight drying, the photonic spheres adhered to the coverslip, but with a central dimple, resulting in doughnuts (see Fig. 5(a)) or a fried-egg pattern (see Fig. 5(c)). The colorful doughnuts are illustrated in Fig. 5(b), observed under a dark-field microscope. A circular or fried-egg-shaped pattern appeared when the concentration of primary particles PB₃ was less than 1 wt%, showcased in Fig. 5(d) under a bright-field microscope. The formation of similar anisotropic patterns in organic and inorganic materials is also explored by Velev et al. in [29].

 

Fig. 5. Examples of alternative photonic structures we identified. (a) Scanning electron micrograph displaying calcined micro doughnut-like assemblies. (b) Dark-field optical image of (a) captured using a $50\times$ objective. (c) SEM image illustrating a calcined fried-egg pattern formed from a dilute suspension of primary silica-decorated microgel particles in mineral oil. (d) Optical micrograph depicting a flat disc-like assembly.

Download Full Size | PDF

4. Summary and conclusions

This work showed how to readily fabricate photonic spheres with an internal silica network structure on micrometer-length scales. The spheres show structural coloration, as observed by optical microscopy and spectroscopy. To this end, we have applied a recently introduced method to fabricate silica foams via soft colloid templating, where pNIPAM microgels take the role of soft and compressible nanospheres [13]. Soft colloid templating allows us to create a foam-like structure and topology, resulting in beneficial performance for structural color applications. We show that we can produce photonic balls of different colors depending on the size of the primary microgel particles. Additionally, we show a modified way to design colloidal morphologies that look like toroids and discs.

The advantages of fabricating network structures in spherical shapes are manifold. Unlike previous research, such as the monoliths from thin films that we have previously described [13], the shape and size of the spheres are well-defined. This small spherical form allows us to apply a high osmotic pressure throughout the samples, facilitating the compression of the sacrificial microgel particles and the creation of foam-like topology. Furthermore, in future work, one could explore the possibility of using microfluidic techniques to produce spheres of the same size. This finite size, comparable to or slightly larger than the scattering length, can be exploited to tune the color response according to the needs of different practical applications, using our knowledge of scattering from a single photonic ball [17] and from photonic ball assemblies [30]. Prefabricated photonic spheres can be characterized and standardized before being used as color pigments. We thus can separate the fabrication of photonic spheres from their deposition, i.e. from the application of their color. Last but not least, photonic spheres offer high flexibility: for instance, they are suitable for direct inkjet or 3D printing, and they can be manipulated with electric or magnetic fields in display applications [4,31–33]. Our discoveries could contribute to the progress of utilizing structural colors in various applications such as material coatings, printing, and more.