Among the different coloration strategies of nature, structural coloration, which uses material morphologies with sufficient optical contrast and structural regularity in micro- and nanoscales to cause light interference without the need for pigmentation, has drawn significant attention because it is associated with many intriguing physical principles and properties [1–3]. Efforts to clarify the details of the structural coloration mechanisms have revealed that nature produces colors by using various photonic principles, such as photonic crystal (PhC) effects [4–7], coherent scattering [8,9], and color mixing [10]. Understanding such structural coloration mechanisms and advances in micro- and nano-fabrication technologies enable us to create what we believe are novel photonic structures by mimicking various biological systems [11–15]. Of particular interest are periodic structures that exhibit iridescence, an optical phenomenon that appears to change color hues with a changing viewing angle, because their properties can be understood, customized, and realized by PhC theories.

Many avian feathers are known to exhibit iridescence, which has been attributed to submicron structural patterns inside barbules—the smallest constituent structures of a bird feather. A barbule is typically composed of a central core and a cortex layer surrounding the core. Barbules exhibiting iridescence typically have a common internal structure in the cortex layer: a periodic PhC arrangement of melanin rods (or melanin tubes or air columns) within a keratin matrix [16–19]. It is interesting to note that these three constituents of barbules have considerably different refractive indices: $n_m\approx 2.0$, $n_k\approx 1.5$, and $n_0\approx 1.0$ for melanin, keratin, and air, respectively [20]. (We note a recent report that the refractive index of melanin is measured to be somewhat lower: $n_m\approx 1.7–1.8$ [21].) The structural geometries of the cortex layer vary between species (for example, hexagonal PhCs in duck wings [18] and square/rectangular PhCs in peacock feathers [16]), creating their own unique iridescence properties as they selectively reflect light of certain wavelengths in certain directions, according to their detailed band structures and associated photonic band gaps (PBGs).

However, unlike highly ordered PhC structures and their resultant distinct iridescence (as seen in duck wings [18] and peacock feathers [16]), some bird feathers (such as those of black-billed magpies or jungle crows) have a relatively low structural regularity and thus exhibit much weaker iridescence [22]. While the detailed mechanisms of the strong iridescence have been clarified to a large extent, the origins of the weak iridescence remain unresolved. We want to know why the bird feathers with weak iridescence (or less distinct color hues) exhibit iridescence and what structural parameters are decisive factors for a given iridescence. In this study, we have two aims: One is to clarify the coloration mechanism of the black-billed magpie feathers, and the other is to realize similar artificial barbules based on the identified coloration mechanism.

Figure 1(a) shows a photograph of a feather from a black-billed magpie. As the figure shows, one side of the feather exhibits iridescence, with its color hue changing from yellow-green to violet to navy blue. Figure 1(b) shows magnified images of the areas that exhibit these three colors. To experimentally confirm that our color perceptions are correct, we used a commercial spectrometer (LAMBDA 950, PerkinElmer, Inc.) to measure the reflectance spectra of those feather sections, which are shown in Fig. 1(c). Indeed, broad reflectance peaks are located in the wavelength ranges of the corresponding colors: ${\lambda }_p\approx 575\mathrm{nm}$, 410 nm, and 470 nm for the yellow-green, violet, and navy blue, respectively. Note that the absolute reflectance values are very small ($<1\%$) because a real feather consists of many individual barbs and barbules and thus scatters light very strongly, allowing only a small fraction of incident light to be reflected and collected in reflectance measurement.

 

Fig. 1. Structural colors in the black-billed magpie feather. (a) Photograph of a black-billed magpie feather showing a gradual color variation. (b) Magnified images of three different areas of the feather that exhibit colors of yellow-green (left), violet (center), and navy blue (right). (c) Reflectance spectra measured for the three sections. (d–e) Cross-sectional TEM images of barbules obtained from the three areas (presented in different magnifications).

Download Full Size | PDF

To determine the origin of such coloration in the feather, we performed transmission electron microscope (TEM) measurements for barbules acquired from the three differently colored areas. The obtained cross-sectional TEM images [Fig 1(d)] display the internal microstructures of the barbules. In general, melanin (black) tubes containing an air (white) column are dispersed in a keratin (dark gray) matrix. More importantly, the melanin tubes congregate near the barbule surface to form a cortex layer with internal substructures. Magnified TEM images [Fig. 1(e)] reveal structural discrepancies between different barbules. First, the number of melanin tube layers in the direction perpendicular to the barbule surface varies: 3–5 for the yellow-green barbules, 2–3 for the violet ones, and 1–2 for the navy blue ones. This variation of the layer number is reflected on their relative iridescence strengths, as shown in Fig. 1(c), where the maximum reflectance varies accordingly. On the other hand, when we consider the ordering of the melanin tubes in the lateral direction (parallel to the barbule surface), the degree of ordering (or the regularity) is degraded as the color changes from yellow-green to violet to navy blue. In the case of the navy blue barbule, we observe hardly any periodic arrangement of the melanin tubes; nevertheless, it still exhibits weak, but clear, iridescence.

Based on these observations, we propose that the origin of the weak iridescence of the black-billed magpie feather is the globally layered cortex structure that determines the iridescence characteristics of an individual barbule. We use the word “globally” to emphasize that the microscopic details of the lateral arrangement of the melanin tubes have no importance. According to our interpretation, the average optical thickness of an individual cortex layer determines the color. In fact, this hypothesis is consistent with the TEM images in Fig. 1(e). For example, we observe that while the physical thickness of an imaginary melanin tube layer in the yellow-green barbule is rather similar to one in the violet barbule ($\sim 170\mathrm{nm}$), but the average diameter of the air columns are distinctly smaller for the yellow-green barbules. This results in a larger optical layer thickness—and therefore a resonance at a longer wavelength—for the yellow-green barbules than for the violet ones. Regarding the navy blue barbules, it is difficult to confirm the existence of any layered structure. Nevertheless, the layer thickness determined from the parts where the melanin tubes locally form double layers seems to be slightly larger ($\sim 200\mathrm{nm}$) than that of the violet barbules, which explains why the navy blue barbule has a longer-wavelength color than the violet one. To confirm that the lateral ordering is not an influential factor for coloration, we theoretically compared two types of PhCs in terms of their band structures. Although both types of PhCs had the same interlayer distance, they had different lateral ordering: hexagonal and rectangular lattice structures (see Supplement 1). The simulation results showed that the two structures exhibit nearly identical PBGs and angular dependences of the reflection spectrum (thus, similar iridescence characteristics), which supports our hypothesis.

Naturally, our next goal is to experimentally mimic artificial barbules that have the proposed coloration effect. This was realized by wrapping a flexible thin film with line corrugations around an optical fiber (core) to form a structured cladding (cortex). Figure 2(a) schematically shows the fabrication steps (see Supplement 1 for a detailed description). The thin-film-rolling technique has been previously demonstrated [23], where an index-contrasted polymeric bilayer was rolled to form a Bragg reflector with cylindrical symmetry. In the present work, we rolled a single surface grating film around an optical fiber core to produce a nano-structured cortex layer. During the rolling process, we maintained the grating lines parallel to the optical fiber axis to produce a two-dimensional (2D) array of densely packed air columns in cross section, which represented the hollow melanin tubes of the magpie barbules. Note that if the rolling process is performed in a conformal manner, the interlayer distance remains constant while the alignment of air columns between different layers is arbitrary, therefore complying with the arrangement scheme of melanin tubes in the magpie barbules. Figure 2(b) displays a scanning electron microscope (SEM) image of a photoresist grating fabricated by the laser interference lithography (LIL) technique, while Fig. 2(c) shows a cross-sectional SEM image of a fully fabricated artificial barbule, which comprises 11 rolled layers of approximately 350 nm thick polymer film with a nominal grating period of $\mathrm{\Lambda }=300\mathrm{nm}$. Figure 2(d) shows a photograph of three artificial barbules (of the same grating period but different polymer layer thicknesses) that exhibit red, green, and blue iridescence colors.

 

Fig. 2. Fabrication of artificial barbules. (a) Schematics of fabrication steps. PVA and PMMA stand for polyvinyl alcohol and poly-methyl-methacrylate, respectively. (b) SEM image of a one-dimensional photoresist grating generated by the LIL process. (c) Cross-sectional SEM image of an artificial barbule. (d) Photograph of three artificial barbules that exhibit red, green, and blue colors.

Download Full Size | PDF

The reflection spectra of the fabricated samples were measured using a compact optical-fiber-based micro-reflectance setup (see Supplement 1). As shown in Fig. 3(a), a measured reflectance spectrum (red curve) exhibits a distinct reflection band centered at $\lambda \sim 520\mathrm{nm}$, which is consistent with the bluish-green color observed in the sample (see the inset photograph). For comparison, the figure also displays the simulated reflectance spectra for two model structures, both obtained using the finite-difference time-domain (FDTD) method. The first simulated model structure is a realistic one; that is, a multiple stack of surface grating layers with a sinusoidal grating profile [left, Fig. 3(b)]. Note that the gratings in the model are intentionally misaligned between layers to imitate the configuration of a real sample. The layer thickness, grating period, and groove height were $t=360\mathrm{nm}$, $\mathrm{\Lambda }=300\mathrm{nm}$, and $h=60\mathrm{nm}$, respectively. The second model structure was a simplified one, in which the grating region that contained both air and polymer was replaced with a single homogeneous layer [right, Fig. 3(b)], whose dielectric constant was approximated using the effective medium theory [24].

 

Fig. 3. Reflectance spectra of artificial barbules. (a) Reflectance spectra measured (red) for a fabricated artificial barbule and simulated for realistic (blue) and simplified (black) model structures. Inset shows an optical micrograph of a fabricated artificial barbule. (b) Schematic of the two simulation model structures: the realistic one composed of multiple stacks of surface grating film (left) and the simplified one with effective medium layers for surface gratings. (c) Simulated reflectance spectra for the realistic model structure with different numbers of stacked layers. (d) Simulated reflectance spectra for various extinction coefficients of the composing film material.

Download Full Size | PDF

All three reflectance spectra (one measured and two simulated) are quite similar, exhibiting high reflection bands at nearly identical wavelength ranges centered at $\lambda \approx 510–520\mathrm{nm}$. We attribute the smaller reflectance maximum in the measured spectrum to the inefficient capture of reflected light from the round sample surface, which was assumed to be flat in the simulations. The agreement between the experimental and simulation results indicates that the structural color is mainly determined by the optical thickness of individual layers in the surface normal direction, whereas structural details in the lateral direction is of no importance, as anticipated. Figure 3(c) shows simulated (using the realistic model structures) reflectance spectra for various numbers of stacked layers ($N$), which illustrate how the reflectance spectrum evolves as the structure gains periodicity. As expected, the height of the reflectance peak increases as $N$ increases. However, $N$ does not affect the spectral position of the reflectance peak significantly (except for $N=1$ and 2), which implies that the resultant color is not sensitive to the degree of periodicity. Figure 3(d) shows another simulated reflectance spectra; here the polymer that composes the grating thin film is assumed to have an extinction coefficient ($k=0–0.05$) to effectively account for the absorptive nature of melanin. Although melanin is known to have an extinction coefficient of $k_m=0.2–0.6$ [25,26], we assume the values reduced by an order of magnitude because melanin covers only a small fraction of the cross-sectional area of the cortex region. The figure indicates that the absorption by melanin would not affect the reflectance peak position significantly, either.

A generally accepted notion is that the concept of the effective medium is valid only in the long-wavelength limit, i.e., when the internal feature sizes are much smaller than the relevant photon wavelengths. In the present situations, however, the structural dimensions are comparable to the PBG wavelength of the underlying PhC structure; for instance, for the artificial barbule shown in Fig. 3(a), the grating period $\mathrm{\Lambda }=300\mathrm{nm}$ is approximately 3/5 of the reflectance peak wavelength $\lambda \approx 520\mathrm{nm}$. Nevertheless, the effective medium concept still seems to hold, as seen in the simulation results of the same figure. To identify the limit of the effective medium (thus, the limit in our structural design of artificial barbules), we fabricated samples with different grating periods. For the grating period up to 600 nm, we could still observe structural coloration. Simulations revealed that for stable coloration, the structural randomness between layers is more important than the small feature size imposed by the effective medium concept (See Supplement 1 for detail).

To demonstrate the flexibility of the described structural coloration scheme, we designed and fabricated artificial barbules of different colors. Figure 4(a) shows the simulated reflectance spectrum as a function of the polymer film thickness. During the simulation, we varied only the polymer layer thickness ($t$) while the grating period and height were fixed at $\mathrm{\Lambda }=300\mathrm{nm}$ and $h=60\mathrm{nm}$, respectively. There is a strong linear relationship between the reflectance peak wavelength and the polymer film thickness. This indicates that the structural color can be tuned simply by changing the polymer layer thickness, which can be accomplished experimentally by controlling the polymer spin-coating speed. Figure 4(b) shows a photograph of four artificial barbules prepared at different spin-coating speeds: 2500, 3000, 3500, and 4000 rpm. It is evident that the color changes from bluish to reddish as the spin-coating speed decreases (or the polymer layer becomes thicker). Figure 4(c) presents the measured reflectance spectra of the four artificial barbules, which demonstrate a monotonic shift in the reflectance peak wavelength in response to the change in the polymer layer thickness.

 

Fig. 4. Variation in structural color. (a) Calculated reflectance spectra as a function of polymer layer thickness. (b) Photographs of artificial barbules prepared at different polymer spin-coating speeds: 4000, 3500, 3000, and 2500 rpm. (c) Measured reflectance spectra for the artificial barbules shown in (b).

Download Full Size | PDF

In conclusion, we propose that the color of the black-billed magpie barbule is determined by the optical thickness of the melanin tube layers in the surface normal direction. Artificial barbules with various structural parameters were fabricated by rolling a polymeric grating film around an optical fiber, which complied with real magpie barbules in terms of their cross-sectional microstructures. The measured reflectance spectra were highly consistent with computer simulation results obtained using the corresponding model structures. By controlling the single and most important structural parameter, the polymer grating film thickness, we could systematically change the color of the artificial barbules. We believe that our findings will not only help understand the physics of weak iridescence in nature, but could also be used in implementing new photonic materials and devices using biomimicry.