Pixelating Structural Color with Cholesteric Spherical Reflectors

While structural color is a powerful means of obtaining saturated and durable pigments that minimize absorption, scattering, and negative environmental impact, appearing naturally in animals and plants as well as in carefully designed artificial composites, it is fundamentally limited to spectral colors, leaving white and other mixed colors elusive. It also normally suffers from a strong viewing angle dependence, making color definition difficult. Herein, it is demonstrated that these challenges can be overcome by using cholesteric spherical reflectors (CSRs), spheres of polymerized cholesteric liquid crystal with radial alignment of the self‐assembled helical structure. Exhibiting omnidirectional selective retroreflectivity of well‐defined color, CSRs are discrete “packages” of structural color. This allows them to be used as pixels for generating nonspectral colors, following the principle of digital displays. A method of creating densely packed monolayers of CSRs with red (R), green (G), and blue (B) retroreflection is developed. Mixing them in equal proportions gives a white surface. By embedding the CSRs in an index matching transparent medium, nonselective specular reflections and scattering are avoided. The approach can be used to create arbitrary colors, including nonspectral ones, without any absorption or nonselective scattering, opening doors to decorating surfaces as desired while minimizing light loss.


Introduction
Because our world is so colorful, the ability to control the color by which an object appears is of great importance, the impact ranging from the survival and proliferation of animals and plants to proper function and market success of human-designed industrial products. Most often, color is generated by a combination of absorption across parts of the visible spectrum (except for black, where all visible light should be absorbed) and strong scattering of all colors, to produce white or to enhance colors defined by absorption. [1,2] While the biological world manages this very well, in a fully sustainable manner, the industrial use of human-made dyes and pigments has severe negative environmental impact. [3][4][5] Concerns have also been raised about the health impact of the TiO 2 particles [6] frequently used to scatter light for strong whites. Moreover, many inorganic pigments rely on rare earth elements or toxic chemical compositions. [2] Structural color, [7] arising from wavelength-selective interference phenomena (Bragg diffraction) in periodic structures with characteristic lengths on the order of visible light wavelengths, is an alternative method to create color, often resulting in vibrant and strongly saturated spectral colors. [2,5,[8][9][10][11][12][13][14][15][16][17][18] It is frequently used in nature [17][18][19] -typically in combination with absorbing pigments-and in some artificial products such as glittery foils. Even in artificial materials, it can be produced fully sustainably from renewable biosources like cellulose or chitin. [8,9,11,[20][21][22][23] Structural color has two important limitations, however. First, the wavelengths of constructive and destructive interference depend on the incidence angle with respect to the symmetry axis of the periodic structure, meaning that a flat structurally colored surface will change in color depending on how the sample is illuminated and observed (iridescence). Second, structural color is purely spectral, hence it does not cover the entire perceivable color space which includes mixed colors, most notably white.
The first problem is much reduced if the structurally colored material is produced not in the form of flat films-as is usually the case-but rather in the shape of spheres, with radial orientation of the symmetry axis. [24] This is achieved particularly easily using cholesteric liquid crystals (CLCs), [19,22] since these longrange ordered and birefringent liquids can be processed into spherical droplets just like with any other liquid, [25][26][27][28][29][30][31][32] for instance using microfluidic or other emulsification techniques, and since they spontaneously organize with a helical organization of the optic axis that modulates the refractive indices with a period p that is easily adjusted from %100 nm to several tens of microns by varying the CLC mixture composition. The helical structure gives rise to Bragg diffraction that is selective not only in wavelength but also in polarization: only the circularly polarized component of light with the same handedness as the cholesteric helix is Bragg-diffracted. [33] Effectively, this turns cholesterics into selective reflectors that are active in a wavelength band of constructive interference Δλ = pΔn, where Δn is the birefringence of the CLC in the absence of helix, centered around a wavelength λ r (as measured in air) given by Bragg's law for cholesterics Here n is the average refractive index experienced by light entering the CLC and θ is the incidence angle of light with respect to the helix axis.
By choosing the right stabilizer of the droplets, one can easily get the cholesteric helix to develop along the desired radial direction. [28] The resulting cholesteric spherical reflectors (CSRs) [24,34,35] are selective retroreflectors tuned to λ 0 ¼ np as retroreflection (θ = 0) wavelength and the circular polarization defined by the helix handedness, which is typically selected by chosing a certain enantiomer of chiral dopant in the CLC mixture. The fascinatingly rich optical properties of CSRs have attracted much attention over the last decade, exploring peculiar aspects from multicolored patterns due to photonic cross communication between adjacent CSRs [30,[36][37][38][39][40][41][42][43][44][45][46][47][48][49][50] to radically different internal reflection paths of CSR droplets [24,38,41] and shells, [24,51] or the change in reflection behavior upon mechanical compression. [41,52,53] Most relevant for the present study is that the omnidirectional selective retroreflectivity turns CSRs into discrete "packets" of structural color that can be tuned anywhere we wish within the visible spectrum (and beyond). [24,35,48] Importantly, since CLCs can be made from chemically reactive molecules, the droplets can be polymerized into solid CSR beads once the CLC self-assembly has reached its equilibrium state. [30,35,40,41,44,49,50,[54][55][56] With optimized parameters, the radial helix structure is sufficiently conserved for retained omnidirectional selective retroreflectivity from the final beads. [24,41] The ease of manipulating solid-state CSR beads and the reduced viewing angle dependence of their apparent color open for a strategy to address the second short coming of structural coloring, namely to produce any apparent color, including white. Thanks to the ease in tuning p via the CLC composition (usually by varying the concentration of the chiral dopant), we can readily produce CSRs with red (R), green (G), and blue (B) retroreflection, and since polymerization turns CSRs into discrete beads that can be distributed into any arrangement, we can use the standard method of the digital display technology for generating full color, namely to mix R-, G-, and B-pixels in a proportion that produces the color we want. Such mixing of structural color elements that each generate one of the three primary additive colors to produce white has actually been found in nature, specifically in the Giant Clam [57] : a microscopic investigation of regions of the animal that appear white to the eye reveals that the white color arises as a result of R-, G-, and B-"pixels" of %5 μm diameter in roughly equal proportions. To the best of our knowledge, this powerful strategy has not yet been utilized in manmade materials to produce nonspectral structural colors.
Here we demonstrate that this color mixing principle can be exquisitely applied to CSRs, a randomly arranged cluster of R-, G-, and B-CSRs in equal proportions appearing white to the eye. We also analyze the CSR optics in great detail in air as well as in an index matching binder, revealing that the latter minimizes specular reflection, indiscriminate scattering, and blue-shifted Bragg diffraction under ambient illumination, meaning that we can generate white while keeping both absorption and broad-band scattering to an absolute minimum. The resulting films with broadly tunable color can be extremely powerful in contexts like photovoltaic cells and solar heat collectors, [58][59][60][61] where arbitrary coloring and decoration are highly desirable for esthetic and architectural reasons, yet the light loss that comes with traditional dyes, pigments, and scatterers would be detrimental to the energy conversion/heat storage performance.

Making Solid Cholesteric Spherical Reflector Beads Tuned for Red, Green, and Blue Retroreflection
We produce monodisperse CSR droplets in a microfluidic setup (see Experimental Section), using two types of reactive CLC mixtures with tunable p, and thus tunable λ 0 of the final CSRs. Most experiments are made with CSRs based on two commercial CLC mixtures (Merck Chemicals), RMM1037 (red λ 0 ) and RMM1035 (near-ultraviolet λ 0 ). R-CSRs are prepared from pure RMM1037, while G-and B-CSRs are made by mixing RMM1037 and RMM1035 in 3:1 and 1:3 mass ratio, respectively. For some experiments, we need p longer than that of the R-CSRs, and here we use an in-house-designed mixture described in the Experimental Section. All CLCs have right-handed helix. Unless otherwise noted, data shown are obtained with CSRs derived from the commercial mixtures. All mixtures are rather viscous and have high clearing temperatures (around 80°C for the commercial mixtures), rendering them unsuitable for microfluidic droplet production on their own. Therefore, a small amount (<10 wt%) of tetrahydrofuran (THF) is added to each CLC mixture to obtain solutions that are isotropic at room temperature and have much lower viscosity.
Each CLC-THF solution is streamed in the microfluidic set-up as disperse phase into an aqueous continuous phase consisting of 5 wt% of polyvinylalcohol (PVA) in water, see Figure 1a and the Supporting Video. Since THF is miscible with water, it is extracted almost instantly from the CLC solution once the latter comes in contact with the aqueous phase, as visible from a birefringent texture developing during droplet production.
The time to obtain the required radial helix orientation in a droplet diverges with its radius, and beyond a certain droplet size, which depends on CLC viscosity as well as elastic constants, well-aligned radial CSR droplets are never obtained. Thus, a compromise is sought where droplets are small enough to ensure reasonably rapid annealing into radial CLC helix alignment, yet large enough to allow easy optical microscopic analysis of individual CSRs with good resolution. The act of polymerizing the CLC may often give further reduction in alignment quality, [40] especially for large droplets. In our case, we find a droplet diameter of 40-60 μm to be suitable. This is also sufficiently large to ensure that the confinement effect on the pitch can be safely neglected. [62] Microscopic investigations of droplets collected directly after production show poor LC alignment, but the alignment progressively improves toward radial helix after an incubation period of 48 h at 35°C. During this time, the droplets are suspended in the aqueous PVA solution. In Figure 1b-d, we show the hexagonally close-packed arrangement of highly monodisperse CSR droplets after annealing. The Maltese cross seen in each droplet in (c) and the corresponding central bright spot in (d) confirm the near-uniformly radial configuration of the CLC helix. After this verification of alignment, the droplets are irradiated with UV light to initiate the polymerization of the reactive monomers and thus solidify the droplets into solid CSR beads with largely retained optical properties. Photopolymerization of acrylate monomers induces shrinkage, [63] with corresponding reduction of CLC reflection wavelength, and indeed we notice a non-negligible reduction in diameter and a slight blue shift in λ 0 of our CSR beads compared to the droplets from which they are formed. After polymerization, the beads are rinsed with pure water several times to remove PVA from their surface. Figure S1 in the Supporting Information shows the reflection polarizing optical microscopy (POM) and reflection spectra of solid R-, G-, and B-CSR beads suspended in water and contained in a flat glass capillary. Each CSR bead features a bright central retroreflection spot with a color that is true to λ 0 under these imaging conditions. As the CSRs are denser than water, and the capillary internal height and the glass wall of the capillary each are about six times the CSR diameter, illumination/reflection light must pass significant distance through air, glass, and water between objective and CSRs, with the effect that strongly inclined paths are excluded from imaging. This leads to minimum blueshift of retroreflection, and thus the central spots appear red, green, and blue, respectively. However, we see in the spectra that the peak of the G-CSRs is significantly closer to that of the B-CSRs than that of the R-CSRs, suggesting that their mixture composition is not quite optimized, yielding a pitch after polymerization that is slightly too short.
Moreover, for the R-CSRs, a pattern of blue-green lines surrounding each central spot is also visible. These lines are due to photonic cross communication between CSRs at wavelengths shorter than λ 0 . [36,38,42] Such visible reflections at wavelengths shorter than what corresponds to red make the R-CSRs poorly color-saturated when viewed macroscopically, approaching pink rather than red. However, since the effect is perfectly predictable, this can largely be compensated for by adjusting p, as we will show toward the end of the paper. Also in G-CSRs, the crosscommunication pattern is seen, albeit much weaker and only with blue lines, reflecting the shorter λ 0 . For the B-CSRs, in contrast, no further reflection spots are visible to the human eye since the cross communication occurs in the UV region when the pitch is as short as in these CSRs.

Collection at a Water-Air Interface
The best way of using CSR beads to make colored surfaces is to transfer them from the 3D aqueous suspension onto a target substrate in such a way that they form a 2D monolayer sheet of close-packed CSRs. Our solution is based on a phenomenon Figure 1. a) Hybrid schematic representation of the microfluidic set-up for CSR droplet generation; the grayscale photo is a frame from a high-speed video obtained during production, while the remaining components are drawn. b-d) Unpolymerized G-CSRs (green retroreflection, λ G 0 % 490 nm after polymerization, see Figure S1, Supporting Information) observed in transmission without analyzer (b), between crossed polarizers (c) and in reflection between crossed polarizers (d); white double arrows indicate polarizer and analyzer directions for c and (d).
www.advancedsciencenews.com www.adpr-journal.com first observed by Pieranski in 1980, [64] in which spherical beads are trapped at an air-water interface, forming a close-packed colloidal crystal stabilized by a balance of capillary and electrostatic forces. This principle was recently refined by Anyfantakis et al. [65] using aqueous suspensions of nano-to micrometer range spheres with density greater than that of water: by filling the suspension into a glass vial with opening smaller than the capillary length and flipping the vial upside down, the spheres sink to the concave meniscus, which remains in place due to capillary forces, where a monolayer of spheres gets trapped at the air-water interface. Flipping the vial back to right-side up, excess particles sink back down from the meniscus, leaving only the trapped monolayer in which the spheres move toward the meniscus center by gravity, forming a close-packed 2D colloidal crystal. As our CSR beads have density greater than that of water, we use the same principle. Starting with CSR beads sedimented at the bottom of a waterfilled glass vial with small opening, we gently flip the open vial upside down until the beads reach the concave meniscus and get trapped, whereafter the vial is gently flipped back. The process is repeated several times until a sufficient number of CSRs is collected at the air-liquid interface. Leaving the vial right-side-up, the CSR beads collectively sink toward the center of the meniscus due to gravity until they reach a close-packed arrangement, forming a densely packed lightly curved 2D monolayer sheet of CSRs. We do this with each of the R-, G-, and B-CSR suspensions (Figure 2a-c) as well as with a suspension where all three CSR types are present in equal proportions (Figure 2d-f ), which would be expected to give a film with effective white appearance according to additive color mixing principles.
We analyze the color appearance of each CSR sheet trapped at the meniscus, initially focusing on only R-, G-, and B-CSRs, respectively, under different illumination conditions. In all cases, we take photographs at about 45°angle with respect to the long axis of the glass vial. We use this photography angle throughout this paper as a standard reference condition, which is reasonably representative of general viewing conditions. As is clear to see in Figure 2, the apparent color depends strongly both on the type of CSR in the sheet and on the illumination. Under regular office lighting (artificial ambient light), the CSRs are illuminated from many directions due to multiple diffuse light sources and reflections on surfaces. While the R-, G-, and B-CSR sheets are clearly colored toward red, green, and blue under these conditions, the color saturation is poor, see Figure 2a. The washed-out colors have several reasons, the most prominent being nonselective scattering and specular reflection of white light due to refractive index mismatch between the water and the CSRs. We will come back to the elimination of these effects below, but first we discuss the second important effect, namely that for G-, and even more for R-CSRs, each sheet selectively reflects also other visible colors than λ 0 . This is due to the cross communication between CSRs [36,38,42,50] (see Figure S1a,b, Supporting Information) and also to the fact that we here have illumination along directions far from the observation direction, yielding visible Bragg diffraction reaching the eye at wavelengths blueshifted compared to λ 0 . [35] This gives the G-CSR sheet a blueish green tint, and the R-CSR sheet appears pink. In contrast, it is not much of an issue for the B-CSRs, since the only additional visible color is violet, all other Bragg diffraction being in the invisible UV range. For this reason, the B-CSR sheet appears with the purest color, although the white scattering and specular reflection still reduces its color intensity significantly.
By introducing an additional ring-shaped light source around the camera (Figure 2g), ensuring illumination nearly along the observation direction, it is immediately apparent that retroreflection starts dominating, giving the CSR sheets more distinct and Figure 2. Densely packed 2D sheets of CSRs with red (left), green (middle), and blue (right) λ 0 , assembled at water-air interfaces, photographed with an ordinary camera at 45°to the interface normal. Panel a) shows the appearance under ambient illumination (light entering from many directions), panel b) adds retroreflection by using a ring-shaped LED around the camera objective, and c) shows retroreflection only, i.e., the ambient light is turned off. A randomized arrangement of R-, G-, and B-CSR beads in 1:1:1 ratio assembled at the water-air interface, photographed at a distance using ambient light d) and without ambient light but with retroreflection lighting e) shows an apparent white color. The individual colored CSRs are visible in the close-up image f ), taken with retroreflection illumination. The set-up for retroreflection imaging is shown for the case of substrate-supported films (see below) in g).
www.advancedsciencenews.com www.adpr-journal.com saturated colors (Figure 2b). By turning off the ambient light and observing the samples under retroreflection illumination only, we see maximally saturated colors (Figure 2c) because the reflections originating from illumination directions other than along the observation direction are absent. However, except for the B-CSR sheet, the colors are still far from pure because the photonic cross-communication contribution is still active. In fact, for sheets of close-packed CSRs of a single type, we effectively get two types of selective retroreflection: on the one hand, we have the direct retroreflection from the center of each CSR, along the helix, at maximum wavelength λ 0 . But in addition, there is now a sheet retroreflection [48] due to photonic cross communication between CSRs, occurring mainly at a wavelength λ s % λ 0 cos45°, since this retroreflection component originates from light hitting one CSR at 45°to the helix axis, being reflected in-plane to the surrounding CSRs, which in turn reflect the light back to the source via a second reflection at 45°angle to their helix axis. [36] Compared to normal incidence, many cross-communication channels will be deactivated, since the inclined illumination direction breaks the symmetry and renders the cross-communication plane non-coplanar with the CSR sheet, but there may be indirect retroreflection pathways involving multiple CSRs that contribute as well. [42] All in all, this sheet retroreflection is strong enough to significantly alter the retroreflection color of the R-CSRs sheet such that it appears pink, and the G-CSR sheet appears blueish green even under this ideal retroreflection illumination condition. Interestingly, the absence of scattered ambient light now reveals the granularity of the B-CSR sheets quite well, while the R-CSR sheet appears almost continuous; the G-CSR sheet has an intermediate character. These differences are not due to different CSR diameters (they are all about 50 μm in these samples) but can be traced back to the sheet retroreflection component and its gradual shift from the visible to the invisible part of the spectrum from R-to G-to B-CSRs. In B-CSRs, we thus have a ring with no visibly colored retroreflection around every central retroreflection spot, while in R-and G-CSRs, the central spot is surrounded (in certain directions) by reflections due to cross communication. Of course, the stronger reflection and smoother appearance of the R-CSR sheet come at the cost of reduced color saturation.
In the mixed sheet, all three CSR bead types are present with almost identical numbers and sizes, and by virtue of the preparation procedure, their location in the 2D sheet is randomized, ensuring that large red, green, or blue clusters are avoided. Indeed, the overall appearance at a distance is quite close to true white, not only under ambient illumination (Figure 2d), where this is more expected due to the nonselective scattering, but even under retroreflection conditions without ambient illumination (Figure 2e) where the colored reflection of each individual CSR is maximally saturated. Since the beads are small enough for the sheet to appear uniform at a distance, the individual colored reflections are not seen but instead mixed to give a white impression. Only by taking a close-up photography (Figure 2f ), where each individual CSR occupies a significant fraction of the image, do we see the break-up into red, green, and blue "pixels." While the color of the mixed sheet is slightly off from pure white, exhibiting a slight blue tone as can be expected, given the slightly too short λ 0 of the G-CSRs and the blue-shifted Bragg diffraction from R-and G-CSRs under ambient light, the offset is not as strong as one might expect when seeing how pink the sheet containing only R-CSRs is. This can be understood by considering that the sheet retroreflection is less prominent in the mixed sheet, since efficient photonic cross communication requires near matching of helix pitch between adjacent CSRs. [41] With the random arrangement of the three CSR types, much of the cross communication is thus interrupted due to significant mismatch in p between adjacent CSRs that are not of the same type (this effect is confirmed at microscopic scale below), giving a quite good balance between red, green and blue reflections. A direct translation from RGB-based color generation cannot be applied, however, since some cross communication remains and since the reflection from R-CSRs under ambient light still occurs across the entire visible spectrum. The balance between the three different CSR types thus needs to be changed with respect to that of traditional RGB pixels to create a certain desired color. Nevertheless, the fact that we can reproduce nearly pure white using solely the structural color of CSRs is highly encouraging, giving us confidence that any color can eventually be produced by tailored mixtures, following a detailed analysis of the overall color generation under varying illumination conditions. If light absorption is not a problem, it may also be advantageous to incorporate dyes in R-and G-CSRs, chosen as to attenuate all colors but that corresponding to λ 0 . [50] To reach deep red colors without dyes, an interesting alternative is to incorporate also CSRs with λ 0 in the near infrared, which will give ambient light reflection and sheet retroreflection in the deep red part of the spectrum. [34] We will come back to this option below.

Substrate-Supported 2D Sheets of Cholesteric Spherical Reflectors Surrounded by Air
The next step is to transfer the CSR sheet from the water meniscus to a solid substrate and evaporate the water. We do so by gently tilting the vial and pouring over the suspension into a glass cylinder, being careful to minimize the detachment of CSRs from the 2D sheet trapped at the air-water interface, see Figure 3 steps 1-2. The cylinder is placed on the target substrate (in our case, a black-painted microscope glass slide) and fixed using a sealing agent that can later be removed, applied around the lower outer rim of the cylinder to prevent water leakage. After transfer of the suspension into the cylinder, the sample is left stable for a few minutes in order for the CSRs to again pack closely as gravity pushes them toward the center of the new meniscus ( Figure 3 steps 2-3). The water is allowed to evaporate slowly such that the CSR sheet moves down until it coalesces on the target substrate to form a flat dry 2D sheet of CSRs once all water has evaporated (steps 3-4). The sealing agent and cylinder are now removed, leaving only the uncovered CSR sheet on the substrate. The small fraction of CSRs that detached from the air-water interface in the transfer process end up below the deposited CSR sheet, hence the final dried films have a few nonmonolayer regions, but these are in minority.
We study the resulting dry CSR sheets under the three illumination conditions used above, and we compare appearance without polarizer with appearance through left-and right-handed circular polarizer, respectively, see Figure 4. We first note that Step 1: The aqueous suspension with polymerized CSR beads trapped at the meniscus is gently poured into the cylinder, forming a monolayer of CSRs trapped at the new meniscus that is initially sparse.
Step 2: Gravity packs the CSRs into a densely packed monolayer around the deepest point of the meniscus.
Step 3: By evaporating the water, the close-packed CSR monolayer is gently brought down to the target substrate, where it remains as a coating after the cylinder is removed (Step 4). For each case, the left column shows appearance without polarizer, the middle column through a left-handed, and the right column through a right-handed circular polarizer. In all cases, the sheet is photographed along a direction 45°with respect to the sheet normal. Scale bar in bottom right photo applies to all photos. Sketch d) shows two paths (blue and green) that yield substrate-mediated nonselective retroreflection of all light and one (red) that does not contribute, because its substrate reflection is directed away from the camera. Sketch e) reproduces the retroreflection path with 57°incidence angle, illustrating how an R-CSR with λ 0 % 0.6 μm will divert most of the retroreflected light via Bragg diffraction during its passage through the CSR, such that it never reaches the observer.
www.advancedsciencenews.com www.adpr-journal.com under ambient light only, a pronounced blue shift is observed for all CSR sheets compared to the situation with the CSRs trapped in the water meniscus. This is particularly clear for the R-CSRs (Figure 4a), which appear green in the top row and the G-CSRs (b) which appear blue in the top row. An even greater surprise awaits when we add illumination for retroreflection (middle row): while the colors of the R-and G-sheets now recover the expected retroreflection colors, in particular when observed through a right-handed circular polarizer, the sheet made from B-CSRs now looks predominantly white. The effect is amplified by the camera tending to overexpose easily, so the photos are taken with manually reduced exposure, still leaving the color saturation of the B-CSR sheet weak. It appears grey rather than the expected blue. When the ambient light is turned off (bottom row), the sheet remains white with only a weak blue tone, showing that the white retroreflection dominates. The strong variation of the effective colors of naked CSR sheets on black glass slides constitutes a striking illustration of the importance of the refractive index of the surrounding medium. We believe the reason for the strong blue shift under ambient light is that the CSR beads are no longer surrounded by a thin layer of water, as was the case when they were trapped at the water meniscus. In that situation, the illuminating light rays were subject to refraction at the air-water interface before they reached the CSRs, reducing the maximum angle θ at which light could enter with respect to the helix axis. As explained by Geng et al., [35] this refraction prior to entering the CSRs removes a significant part of the blue-shifted Bragg reflections occurring when the angle between illumination and observation directions is large, explaining why there was a less severe blue shift upon ambient illumination when the CSRs were surrounded by water.
Concerning the strong white retroreflection of the G-and B-CSRs, we conjecture that this is due to so-called "cat's eye retroreflection" enabled by the CSR acting as a focusing lens for incoming light, [66] possibly supported by a reflection event in the underlying substrate, as illustrated in Figure 4d. Also internal reflections at the back CSR-air interface should contribute to varying extent, but for the sake of clarity, these paths are not drawn. The three drawn paths comprise two (blue and green: these colors are used only for distinguishing different paths; they have no relation to the wavelength of light) that give rise to nonselective retroreflection, whereas the third (red) enters too close to the CSR center, leading to the substrate reflecting the light away from the CSR rather than back into it. All paths are calculated for an average CSR refractive index n ¼ 1.6 and n = 1 outside, since the surrounding medium is air. The blue path is the furthest from the center that can enter the CSR, as its incidence angle with respect to the CSR-air interface normal is 90°.
Why do we not see strong cat's eye retroreflection from the R-CSRs? The reason is that with these CSRs, the light reflected from the substrate is subject to Bragg diffraction from the cholesteric structure and, as illustrated in Figure 4e, all visible wavelengths shorter than about 0.55 μm (estimated assuming λ 0 % 0.6 μm; see Figure S1d, Supporting Information) are reflected away from the retroreflection path if they have the appropriate polarization. These wavelengths are therefore not retroreflected. Red light, in contrast, is fully retroreflected by the cat's eye retroreflection paths, and regardless of polarization. The overall effect is the predominantly red retroreflection appearance seen in Figure 4a, with poor polarization contrast (although a slight bluish tint appears with left-handed polarizer that is not seen through a right-handed polarizer).
The removal of visible light from the retroreflection path by cholesteric Bragg reflection does not happen in the B-CSRs since their pitch is too short. All paths with 90°or smaller incidence angle, down to somewhat greater than 28°(the red path in Figure 4d), thus contribute to retroreflection, for all wavelengths and polarizations. This gives rise to the white appearance of the B-CSRs under retroreflection conditions. For G-CSRs, an intermediate situation arises, giving a slight green-yellow tint, arising from the strong red and green retroreflection of all polarizations, and blue for the polarization that is not Bragg-reflected.
Also at microscopic scale, we see differences in the optical behavior of CSRs surrounded by air as compared to water, as revealed in the polarizing optical microscopy photos in Figure S2, Supporting Information. The most striking aspect is that the central reflection spot of R-CSRs is strongly blueshifted when imaged using a high-magnification objective, appearing yellow-green through a 50x objective rather than the orange-red seen with 20x or 10x objectives. This is most likely due to the high numerical aperture of this objective, leading to illumination and imaging over a large range of incidence angles, blueshifting the reflection according to Bragg's law. The reason that the central spot is orange at low magnification instead of the red seen in Figure S1a, Supporting Information, is the absence of water and glass over the CSRs, which ensured that almost only vertical reflections, at λ 0 , were imaged under the conditions of Figure S1a, Supporting Information. With the CSRs surrounded by air, all inclined illumination and reflection paths allowed by the numerical aperture of the objective participate in the imaging, blueshifting the central spot to a more orange color. The effect is present also for the B-and G-CSRs, but there it is not as striking, since much of the blue-shifted wavelengths are in the UV range and thus not visible.
A second effect seen best for the R-CSRs (weakly visible also for G-CSRs) is that the cross communication does not give rise to radial lines, as when the CSRs were embedded in water ( Figure S1, Supporting Information) but only to spots (with some azimuthal extension), regardless of magnification. This change may be explained by the greater degree of refraction that now takes place at the direct air-CSR interface. Only at the periphery, does the light reflected at 45°to the helix axis of one CSR traverse in-plane to reach the adjacent CSRs for back reflection. The light entering the core of a CSR is refracted due to its higher refractive index than air, changing its course and thus disrupting the cross communication between CSRs at internal locations within each bead. [38] As during macroscopic photography, the nonselective cat's eye retroreflectivity strongly affects the character of the B-CSRs, less the G-CSRs, and least the R-CSRs. Comparing the photos obtained with circular polarizers (top set of micrographs in Figure S2, Supporting Information) with those without any polarizer (bottom), the B-CSRs lose their intense blue color and appear with a washed-out violet. This can be explained by superposition of selective Bragg reflection and nonselective cat's eye retroreflection. With the R-CSRs, the main consequence is an overall red color, also outside the central spot, when imaging with the 10x objective. As shown in Figure 4e, all shorter wavelength light that was cat's eye retroreflected is Bragg-reflected away from the path for imaging during the passage through the CSRs, but the red light is not, as this light path is never along the helix, hence it reaches the camera.

Substrate-Supported 2D Sheets of Cholesteric Spherical Reflectors Embedded in Binder with Matched Refractive Index
In order to avoid the cat's eye retroreflectivity, to minimize the blue shift upon ambient illumination, and to truly saturate the color, we need to embed the CSRs in a binder with as high refractive index as feasible, yet it must remain close to n in order not to cause scattering at the CSR-binder interface. Indeed, the ideal is to approximate index matching [35,48,56] (complete index matching is impossible since the CSRs are optically anisotropic), which means that the binder should have a refractive index near n ¼ 1.6 for the CSRs studied here. Following the procedures of CSR assembly and transfer to a solid substrate described above, we thus proceed by finally embedding the CSRs in the transparent and colorless UV adhesive NOA160 (Noland Optical adhesive) which has a refractive index n = 1.6.
The samples shown in Figure 5 were prepared in the same way as those in Figure 4, with the additional step that each sheet was finally covered by NOA160 which was then UV cured into solid state. The increased color saturation is apparent, in particular for the G-and B-CSRs. For all illumination conditions, without polarizer or through a right-handed polarizer, the G-CSR sheet now appears clearly green, albeit with a blueish tint, and the B-CSR sheet appears clearly blue. This demonstrates the efficiency of the index matching binder in eliminating the nondiscriminate cat's eye reflectivity. The polarization contrast is excellent for the B-CSR sheet, which is almost dark through a left-handed polarizer for all illumination conditions, and very good also for the G-CSRs, even if here some colored reflection can still be seen through the left-handed polarizer. The second striking difference is the much lower sensitivity to the lighting conditions, again the most apparent for G-and B-CSRs. This can be explained by the beneficial impact of the refraction that light undergoes at the air-NOA160 interface, blocking light at high incidence angle from entering the film, thus not reaching the CSRs. [35] Because the indiscriminate cat's eye retroreflection is now absent, all Bragg reflections, including the contributions from cross communication, are now strong when they are in the visible domain. This is particularly apparent in the R-CSR sheet, which now appears with pink color under retroreflection conditions without polarizer or through right-handed polarizer, while it appears purple through left-handed polarizer under these lighting conditions. When the illumination is only ambient light, the θ > 0 reflections covering almost the entire visible spectrum give the sheet an almost white appearance when no polarizer is present, slightly blueish through left-handed polarizer while it is a warmer pink-yellow through right-handed polarizer. The longer p also means that even the light passing the air-NOA160 interface at non-retroreflection conditions gives rise to significant blue-shifted visible reflection, rendering the R-CSR sheet the only one where significant color change is still seen between ambient and retroreflection illumination conditions, respectively.
At microscopic scale, the color consistency is now excellent, see Figure S3, Supporting Information. Regardless of magnification, and with little effect of adding or removing polarizers, the central reflection color remains constant in color: blue for B-CSRs, green for G-CSRs, and orange for the R-CSRs. The latter are closer to red than when they were surrounded by air, but they are still distinctly more orange than when surrounded by water and enclosed in a capillary in Figure S1a, Supporting Information. The CSRs studied here are covered only by a minimum excess amount of NOA160, hence inclined illumination and reflection paths do participate somewhat more than under the conditions of Figure S1, Supporting Information, explaining the slight remaining blueshift. The R-CSRs provide one further difference compared to the situation in water inside the capillary: the blue-green radial cross-communication lines Figure 5. 2D sheets of a) R-CSRs, b) G-CSRs, and c) B-CSRs embedded in cured NOA 160 glue, on a black-painted glass substrate. The top row shows appearance under ambient illumination, the middle under mixed ambient þ retroreflection from a ring light at the camera, and the bottom under retroreflection only. For each case, the left column shows appearance without polarizer, the middle column through a left-handed and the right column through a right-handed circular polarizer. In all cases, the sheet is photographed along a direction 45°with respect to the sheet normal. Scale bar in bottom right photo applies to all photos.
www.advancedsciencenews.com www.adpr-journal.com are largely absent. It appears that the optimum binder refractive index for obtaining the line retroreflection at these short wavelengths is somewhere in between that of air (n = 1) and that of the average CSR refractive index (n % 1.6), such that water (n % 1.3) gives ideal cross communication throughout the CSR core. This suggests that the initial analysis of internal cross communication between CSR droplets by Fan et al., [38] which ignored refraction effects, needs to be refined. Quite likely, it is not only refraction at the interface of CSRs and their surroundings that needs to be taken into account but also internal refraction, since the effective refractive index changes during passage through the CSR, thanks to the continuously changing helix orientation. Given the almost white appearance of the balanced mixture of R-, G-, and B-CSRs trapped at the water meniscus shown in Figure 2, it is interesting to see how this appears in the dry state, encapsulated by index matching NOA160 binder. The RGB-CSR mixture was thus transferred into a dry film and then encapsulated by NOA160 that was finally UV-cured, following the same procedure as for the pure R-, G-, or B-CSR suspensions, the result is shown in Figure 6. For all types of illumination, a blueish tone of white is seen from the monolayer areas, while some regions with stacked CSRs (due to detachment from the meniscus in the transfer process) appear more white. With ambient light only (upper row in panel a), the blueshift is weakened by the warm color of the artificial lighting active in the lab used for this experiment, but it is clearly visible when adding illumination for retroreflection (middle and lower rows). The blue tone is not primarily due to the sheet retroreflection, as might first be expected based on the experiments with single-CSR-type sheets, because the random distribution of the three CSR types cancels most cross communication. This is confirmed in the microscopic investigation (panels b-d). When comparing the appearance without polarizers (c) with that using right-handed (b) and left-handed circular polarizers (d), we also note that right-handed polarizers maximize the contrast and color saturation, but some color remains reflected by some CSRs even when using left-handed polarizers. This is most likely due to an imperfect alignment within these beads. Interestingly, the color in the left-handed channel is less blueshifted than in the right-handed channel, and rather than uniformly filled spots, a concentric ring pattern appears. The elucidation of these (weak) signals is outside the scope of this study.
We believe there are two main reasons for the blueish character of the RGB sheet under retroreflection conditions. Foremost, we attribute it to the fact that our G-CSRs have too short λ 0 , being significantly closer to λ 0 of B-CSRs than that of R-CSRs, as seen in Figure S1d, Supporting Information. Second, even the ring light does not give illumination exclusively along the imaging direction, hence the retroreflection spot of R-CSRs is blueshifted, appearing more orange than red, as confirmed microscopically in Figure S3, Supporting Information. To achieve a deeper red effective appearance, we proceed as follows. Since the appearance of the R-CSR sheet in Figure 5a is so strongly affected by the sheet retroreflection at cross-communication wavelength λ s < λ 0 , and by ambient light giving rise to similar blueshifted reflections, we conjecture that an effectively red-appearing sheet may be achieved by extending p until λ 0 is in the infrared, such that the apparent color is defined almost only by λ s , because λ 0 is too long for our eyes to see. We thus prepare one CSR sheet with λ 0 % 880 nm after polymerization ( Figure 7c) and investigate it both with POM and macroscopically. As expected, the CSRs no longer have a visible retroreflection spot in the center, neither before ( Figure 7a) nor after (b) polymerization, but the crosscommunication reflections are red-orange. They appear slightly deeper red before than after polymerization, which reflects the shrinkage of the pitch induced by the polymerization. www.advancedsciencenews.com www.adpr-journal.com As seen in Figure 7d, the red appearance is confirmed also on macroscopic scale provided that the illumination for retroreflection is turned on (mid and lower row). The polarization contrast is weak, the sheet appearing the most saturated red through right-handed polarizer, while it appears a brownish red through left-handed polarizer and a pinkish red without polarizer. When only ambient light is used to illuminate the sheet, the sample now appears predominantly green, almost independent of polarizer. This suggests that p needs to be expanded further to produce even higher λ 0 in order to make the sheet appear red under random illumination directions, but the strategy is clearly promising.
We are confident that further refinement of the pitch of IR-CSRs will allow the generation of sheets that appear effectively red also under ambient illumination, but the optimum p will depend on whether color mixing as in Figure 6 is intended, since the mixing of different CSRs largely stops the cross communication, or whether a pure red color is desired. Most likely, the ideal CSR color mixing palette will consist of more than three types, with several values of λ 0 differently far into the infrared complementing CSRs with red λ 0 , the exact proportion between all types to be calculated following a complete analysis of the different phenomena uncovered in this study. The fact that the CSRs for different primary colors operate quite differently renders the analysis more complex than for standard RGB color mixing.

Conclusions and Outlook
By systematically analyzing the reflections of polymerized cholesteric spherical reflector (CSR) beads under ambient, retroreflection, and mixed illumination conditions, without polarizers and through left-as well as right-handed circular polarizers, in water, air, and in index matching binder, and by varying the cholesteric pitch from short enough to produce blue retroreflection to long enough to move the direct retroreflection into the IR range, we have demonstrated the potential and complexity of color generation with CSRs. Most importantly, we have for the first time emphasized the importance of an index matching binder to avoid nondiscriminate cat's eye retroreflectivity, the presence of which quenches the selective reflection colors of CSRs, and we have shown that even red retroreflection color can be obtained at macroscopic scale by utilizing the sheet retroreflection at λ s , shifting the direct retroreflection wavelength λ 0 to the near-infrared. We also demonstrated that the sheet retroreflection due to cross communication is largely suppressed when mixing CSRs with sufficiently different pitch, as in an RGB-CSR sheet designed to approximate white appearance. While much fine-tuning and in-depth analysis remain to be done, the combination of CSRs with different tailored pitch appears to be a promising strategy to produce arbitrary structural color even without absorptive elements. For commercial use, the method for creating the monolayer used here may need to be replaced with one that is easier to scale up, but for lab scale experiments this method is reliable. The main challenge is to avoid CSR detachment from the water meniscus during transfer, leading to CSRs stacking on top of each oher.
With random arrangements of CSRs with red, green, and blue retroreflection, respectively, a good white appearance results at macroscopic scale, even though our not fully optimized mixtures led to a slight blue shift. By fine-tuning λ 0 of each CSR type, possibly incorporating also CSRs with λ 0 in the IR range, we believe an excellent white appearance can be achieved. This requires reworking the color mixing principle compared to standard RGB displays, given the more complex optics of CSRs. To modulate intensity and produce grayscale, CSRs with λ 0 in the ultraviolet range will produce "pixels" that are effectively black, since the background of any structural color-based pigments should be black. This is the case for many of the application scenarios that may be considered, such as inorganic photovoltaic cells or optimally designed solar water heaters, hence they are ideal backgrounds for strong structural color generation. By decorating them with CSRs with tailored λ 0 , in carefully chosen proportions at each location, complex patterns that can have any color should be possible, while letting through all light that is not selectively reflected, thus having minimum impact on the performance of the device. This would be of great value for depositing such devices in, e.g., building facades, with much greater public acceptance. Compared to flat film structural color materials, the use of CSRs greatly reduces the problem of viewing angle-dependent image colors. A further future step of value is to generate the  Figure 7. POM investigation (polarizer and analyzer directions indicated by perpendicular white double arrows in (a)) of IR-CSRs (made using the in-house-formulated reactive cholesteric mixture) in aqueous PVA solution directly after production, a) before polymerization and b) after polymerization, embedded in NOA160 and c) retroreflection spectrum obtained at the top of the CSRs in NOA160. d) Macroscopic appearance of a 2D sheet of IR-CSRs embedded in cured NOA 160 glue on a black-painted glass substrate, investigated in the same way as in Figure 5. The photo in (b) has many artifacts due to air bubbles trapped in the NOA glue.

Experimental Section
Microfluidic Droplet Production: Reactive CLC droplets were produced using microfluidic devices following the design of Utada et al. [67] This flow-focusing technique was based on two tapered cylindrical glass capillaries (Drummond, 1 mm outer diameter and 0.58 mm inner diameter) inserted into a square capillary (1.05 mm inner side length, 1.5 mm outer side length). Both inner capillaries were tapered using a Sutter P-100 pipette puller and cut using a microforge to obtain a 35 μm diameter orifice for the injection capillary (for injecting CLC mixture as disperse phase) and a 175 μm diameter orifice for the collection capillary (for collecting a dispersion of CLC droplets in aqueous continuous phase). The continuous phase was flown through the interstitial spaces between collection and square capillaries, in the direction opposite to that of the disperse phase flow.
All capillaries were cleaned with deionized water (Sartorius arium pro DI) and dried with compressed air. The capillary set-up was assembled on a clean microscope slide in such a manner that the two capillary tips face each other, are well aligned and centered, and with the injection capillary eventually inserted approximately 100 μm inside the collection capillary, as shown in Figure 1a. A Sterican Blunt 21G cannula (B. Braun) was fixed on top of one end of the square capillary using two-component epoxy glue (Pattex) creating an inlet for the aqueous continuous phase. At the other end of the square capillary, the interstice is sealed with UV reactive epoxy NOA81 (Norland Optical Adhesive). The flow rates of the disperse THF-dissolved CLC phase and the aqueous continuous phase were separately controlled using a 1034 mbar Fluigent MFCS-EZ pneumatic control system, pressuring vials containing the respective liquids. Typical flow rate values were 7.5 and 20 mL h À1 for inner and outer phases, respectively.
During droplet production, the microfluidic device was kept on a hot stage at a 60°C. Inside the collection capillary, disperse phase droplets broke up just after leaving the injection capillary orifice (dripping mode). The resulting dispersion was collected into a glass vial and later incubated for 48 h at 35°C to obtain good alignment of the CLC droplets before optical characterization. After the incubation period and after ensuring proper LC alignment, the droplets were irradiated for 15 min by a UV lamp operating at 365 nm with an intensity of 32 mW cm À2 .
Prior to transferring CSR beads from the meniscus in a vial to the glass cylinder, the bottom glass slide was painted using Maston black matt paint. A single coat was applied following the written directions provided by the manufacturer and left for at least 1 h to dry. Afterward, the glass cylinder was placed on top and the bottom part was sealed to prevent any leakage using using two-component silicon-glue Twinsil (Picodent).
Optical Characterization: An Olympus BX51 polarizing optical microscope equipped with a digital camera (Olympus DP73) was used for microscopic characterization. For illumination with circularly polarized light and analysis using a circular polarizer, a λ/4-plate was inserted in the light path, and the handedness for polarization and analysis were adjusted independently by rotating the polarizer and analyzer, respectively. Macroscopic images were acquired with a Canon EOS 100D digital camera. Commercial right-and left-handed circular polarizers (Thorlabs) were used in front of the camera for confirmation of circularly polarized reflection during imaging. The reflection spectra were obtained using unpolarized white illumination with an AvaSpec-2048 (Avantes) spectrophotometer coupled to the microscope.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.