Polarization-sensitive color in iridescent scales of butterfly Ornithoptera

Abstract

Diverse biological microstructures that result in advanced optical effects have been systematically investigated. However, the mechanisms of polarization-sensitive color have not been fully understood. Here, we report a combined architecture comprising upper deep grating and bottom multilayer in butterfly iridescent scales that leads to polarization-sensitive color related with scale azimuth. The polarization process is unraveled through detailed investigations on the green scales in butterfly Ornithoptera priamus poseidon and the orange scales in butterfly Ornithoptera croesus lydius. The bright green and orange scales result from the interaction between the transmitted 0th order diffraction of upper grating and the 1st order interference of the bottom multilayer. Combining experimental results with calculation, we clarify the structural origin and the mechanism of the polarization conversion. Tapered grating achieves form-birefringence rotating polarized light and the multilayer selectively reflects the incident light dominating the reflection color. The height of the grating, which determines the phase difference of the two decomposed polarized lights, is essential to polarization conversion. To our knowledge, it is the first study to provide a structural prototype comprising deep grating and multilayer for achieving polarization-sensitive color. These findings show significant promise in biosensing, anticounterfeiting and optical-material design.

---

Introduction

Many natural biological structures achieve structural colors, the mechanisms of which inspire biomimetic photonic materials.^1–7 Natural organisms usually adopt typical ordered structures, such as grating, multilayer, and photonic crystals, to produce ordinary color signals such as iridescence.^8–10 However, more elaborate microstructures are required to achieve sophisticated optical effects such as polarization-sensitive color.^11–14 The Chrysina gloriosa beetle adopts the chiral and periodic helical structures to selectively reflect the left-circularly polarized light to have a striking iridescence.¹¹ The fruit of Pollia condensate has an intriguing pointillist appearance because of the helical structure in its surface that leads to a specific color.¹² The wings of the Papilio palinurus butterfly and Madagascan sunset moth display a color mixing effect because of the retroreflection structure that rotates the linearly polarized light.^13,14

The polarization-sensitive color is caused by the polarization-dependent reflection and can increase the perceived visual contrast of light with different brightness and wavelengths.¹⁵

Previous studies show that many organisms having polarization-sensitive photoreceptors can potentially discriminate objects through the polarization-dependent reflectance.^16,17 The polarized light signals contribute to biological signalling functions in mating, feeding, and oviposition.^18–20 For example, the Papilio aegeus and Heliconius butterflies use polarized light in mate recognition.¹⁸ Utilizing the polarized light signals is highly useful for species living in complex light environments,²⁰ where illumination can dramatically change in brightness and spectral properties.

Here, we report a polarization-sensitive phenomenon in the butterfly Ornithoptera iridescent scales. Detailed investigations on the green scales of butterfly Ornithoptera priamus Poseidon (O.p.poseidon) and the orange scales of butterfly Ornithoptera croesus lydius (O.c.lydius) reveal the structural origin of the polarization-sensitive color. Combined architecture comprising upper deep grating and bottom multilayer is extracted from scale structure for unraveling the polarization mechanism. Through experiments and calculations, we clarify that the combined architecture selectively reflects and rotates incident polarized light.

Material and methods

Butterfly samples

The investigated O.p.poseidon and O.c.lydius male butterflies, two species of the bird-wing butterfly living in Indonesia, were provided by the Shanghai Entomological Museum, China.

Morphology characterization

Digital photographs were captured by a Canon EOS 350D digital camera. Optical microscopic images were obtained with a digital microscope (VHX-100, KEYENCE), revealing the arrangement of butterfly wing scales. The polarization effect was investigated by the Zeiss Axio Scope A1 Microscope equipped with collinear and crossed polarizers. The anatomical microstructure of wing scales was investigated with a field emission scanning electron microscope (FESEM; FEI NOVA NanoSEM 230) and transmission electron microscopy (TEM; FEI Tecnai G2 spirit Biotwin). Samples for TEM imaging were prepared by first embedding the wing scales in epoxy resin at 60 °C for 48 hours, then cutting with a low temperature sectioning system (Leica ULTRACUT UC6) into ultrathin sections (70 nm), and finally staining with lead citrate for 6 minutes.

Spectra

The reflectance spectra of intact wings were measured with a normal UV-vis-NIR spectrometer (Varian Cary 500 infrared-visible-ultraviolet spectrometer with an integrating sphere) and 1 cm² area of forewing color patches were subjected to incident light. To obtain the absorbance spectrum of the absorbing pigments within the wing scales with a microspectrophotometer (Craic QDI 2010), color scales were removed with adhesive tape. A single scale was immersed in bromoform, the refractive index (RI) of which is 1.6,²¹ and 10 μm² area of a single scale was illuminated by incident light. Angle-dependent reflectance spectra were collected using an UV-vis-IR fiber optical spectrometer (Ideaoptics Instruments Ltd., China). The incident light illuminated on the intact color scales and the diameter of the circular light spot was about 2 mm. The incident light and detector were rotated in the plane perpendicular to the scales to measure the mirror angles. To investigate the polarization-dependent reflection, we equipped the fiber optical spectrometer on the Zeiss Axio Scope A1 Microscope. The diameter of the circular light spot was about 40 μm. Upon the crossed polarizers and collinear polarizers, we rotated the sample azimuth to obtain the polarization-dependent reflectance spectra.

Chromaticity diagram

The chromaticity coordinates x and y were calculated by the formula defined by the CIE 1931 convention.²

Simulated calculation

Theoretical simulations were performed on software OptiFDTD from Optiwave Systems Inc. According to the parameters obtained from the FESEM and TEM images, we established the simplified scale model. For the finite difference time domain (FDTD) simulation, a Gaussian modulated light source was adopted. An anisotropic perfect matching layer (APML) boundary condition was employed. The resulting calculation data was transferred from time domain to frequency domain by discrete Fourier transform method, generating the simulated spectra.

Results and discussion

Wing morphology and optical spectra

The two sister species of butterfly Ornithoptera, O.p.poseidon and O.c.lydius, have different dorsal wing patches. The butterfly O.p.poseidon exhibits striking green patterns, whereas the butterfly O.c.lydius has orange patterns (Fig. 1a and b). The O.p.poseidon and O.c.lydius forewings both have cover scales of about 140 μm in length and 80 μm in width, which closely pack on the black ground scales (Fig. 1c and d).

Fig. 1 Morphology and spectra of O.p.poseidon green scales and O.c.lydius orange scales. (a) Photograph of O.p.poseidon dorsal surface. (b) Photograph of O.c.lydius dorsal surface. (c) Optical microscopic image of green scales. (d) Optical microscopic image of orange scales. (e) Reflectance spectra tested with an integrating sphere. (f) Normalized absorption spectra of single scale immersed in bromoform. (g) FESEM image of green scale from the top view. (h) FESEM image of orange scale from the top view. (i) TEM image of the cross section of green scale. (j) TEM image of the cross section of orange scale. Scale bars: (a and b) 2 cm; (c and d) 100 μm; (g–j) 1 μm.

Reflectance spectra were measured on a spectrophotometer with an integrating sphere under normal incident light. The green and orange scales have a reflectance peak at 550 nm and 670 nm, respectively, as shown in Fig. 1e. To identify the pigments in the scales, we investigated the absorption spectra of the pigments in green and orange scales. A single green scale and a single orange scale were immersed in bromoform (RI is 1.6), approximately matching the RI of chitin (RI is 1.56).²¹ The absorbance spectra of both the wing scales have similar main bands in the violet-blue wavelength ranging from 350 nm to 480 nm (Fig. 1f), suggesting that the pigments in both the green and orange scales have similar light-absorbing effects.

FESEM and TEM images reveal the similar structures of the green and orange scales (Fig. 1g–j). Tapered ridges, with tiny inclination cross-ribs, run parallel to the longitudinal axis of the scale. The adjacent distance of the ridges is 0.54 μm for green scale and 0.76 μm for orange scale (Fig. 1g and h). Images of the cross section perpendicular to the longitudinal axis of scale show anatomical architecture consisting of upper tapered ridges and bottom multilayer (Fig. 1i and j). In the green scales, the tapered ridge structure is 1.90 μm and the multilayer is 2.50 μm in height. In the orange scale, the tapered ridge structure is 2.60 μm and the multilayer is 2.60 μm in height. The bottom multilayer structure is composed of alternative chitin and hybrid layers (hybrid layer is composed of chitin columns and air). The thickness of the chitin layer is about 0.10 μm, and the thickness of the hybrid layer is about 0.11 μm for green scale, and 0.12 μm and 0.14 μm, respectively, for orange scale (Fig. S1, ESI†). A schematic model of the scale structure was established, the average dimensions of which were statistically determined based on FESEM and TEM images. (Fig. S2a and Table S1, ESI†).

This model was further simplified to facilitate analysis on the optical effects in the green and orange scales (Fig. S2b, ESI†). The tapered ridge structure was simplified into tapered triangular grating structure, and the bottom multilayer structure (alternative chitin and hybrid layers) was simplified into an ideal multilayer structure (Fig. S2b, ESI†). The optical mechanism was analyzed based on the simplified architecture composed of an upper grating structure and a bottom multilayer structure.

The RIs of chitin and air are 1.56 and 1.00, respectively.²¹ According to the parameters of the scales (Table S1, ESI†) and the effective RI theory,¹⁰

where n_c and n_air are the RIs of chitin and air, and f_c is the volume ratio of the chitin. The calculated effective RIs of every part in the combined architecture are shown in Table 1 (Fig. S2b, ESI†).

Table 1 Effective RI of the grating and the hybrid layer

Butterfly	O.p.poseidon		O.c.lydius
Structure	Hybrid layer	Grating	Hybrid layer	Grating
Chitin volume ratio	0.11	0.9	0.06	0.80
Effective RI	1.08	1.50	1.04	1.48

---

When the normal incident light illuminates the combined architecture, the upper triangular grating should cause the grating diffraction and the bottom multilayer should cause the multilayer interference.²² The grating diffraction and multilayer interference orders were calculated to explain the optical phenomenon in the O.p.poseidon and O.c.lydius butterflies.

It has been shown that the tapered triangular grating acts as an antireflection layer that reduces the reflected diffraction order.²³ In our model, the grating has adjacent distance comparable with the wavelength of visible light; therefore, the modal method is effective for interpreting the diffraction behavior.^24,25 According to the modal method, only the transmitted 0th order and ±1st orders diffraction lights (0th and ±1st mode lights) will be nonevanescent at normal incident light.^24–26 The 0th and ±1st mode lights pass through the upper grating and reach the bottom multilayer in the 0th and ±1st transmitted diffraction angles, respectively.

In the case of the 0th mode light, the 0th transmitted diffraction angle is zero. The 0th mode light normally reaches the bottom multilayer and causes the optical coherent interference. The constructive interference equation of the multilayer is given by²²

where (n_c), t_c(t_h), and θ_c(θ_h) are the RI, the thickness, and the refractive angle of the chitin layer (hybrid layer), respectively, λ is the incident light wavelength, and m is an integer.

The calculation peaks of the multilayer are at 550 nm and 670 nm (m = 1) in the green and orange scales, respectively. The experimental reflectance peaks (Fig. 1e) are approximately the same with the results of the 1st order multilayer interference of the 0th mode light, indicating that the 0th mode light contributes to the striking color of the green and orange scales.

In the case of the ±1st mode lights, incident lights reach the multilayer in the diffraction angles, , respectively.²⁴ The relevant calculation is shown in Text S1, ESI.† According to the multilayer interference theory, the obtained calculated peaks of the multilayer are at 410 nm and 550 nm (m = 1) in green and orange scales, respectively, which are shorter than the experimental results (Fig. S4, ESI†). Therefore, the ±1st mode lights should have little contribution to the main bands of the reflectance spectra. The reflected lights of the transmitted ±1st diffraction order of the green scales are probably suppressed by the pigment, which has a main absorbance band from 350 nm to 480 nm. The reflected lights of the transmitted ±1st diffraction orders in the orange scale slightly widen the reflection range.

Theoretical calculation based on our model indicates that the main reflection bands of the green and orange scales mainly result from the interaction between transmitted 0th order diffraction of the grating and the 1st order interference of the multilayer.

Iridescence of the green and orange scales

In nature, iridescence is a structural color widely observed in birds, beetles, and butterflies. Iridescence is known to depend on the incident light and viewing angle. Multilayer interference iridescence has been found in the elytra of the jewel beetle Chrysochroa fulgidissima and the breast-feather barbules of the bird Parotia lawesii.^27–29

Fig. 2 shows angle-dependent optical microscopic images and reflectance spectra in the plane perpendicular to the ridges. The incident light and detector are rotated in the plane perpendicular to the scale ridges to achieve different mirror angles. The observed color of the green scales gradually changes from yellowish green to bluish green with increasing viewing angle, and becomes yellow at 60°. In the case of the orange scales, the observed color gradually changes from orange to yellow (Fig. 2a). The peaks of angle-dependent reflectance spectra of both the butterflies undergo blue shift with increasing viewing angle and vanish at 60° (Fig. 2b). The chromatic diagram confirms the relation between the color variation and reflectance spectrum, revealing that the yellow color at 60° is caused by the disappearance of the spectrum peaks. The tendency of color variation in the chromatic diagram is the same as that in the experimental microscopic images (Fig. 2c).

Fig. 2 Iridescence property of green scales and orange scales. The incident light and detector rotate in the plane perpendicular to the scale ridge to measure the mirror angles. (a) Angle-dependent optical microscopic images of green and orange scales. (b) Angle-dependent reflectance spectra of green and orange scales. (c) Chromaticity diagram of the angle-dependent reflectance spectra in green and orange scales. The square represents the corresponding color of green scales, and the circle represents the corresponding color of orange scales. “⊥” represents that incident light and detector rotated in the plane perpendicular to the scale ridges.

Both the angle-dependent spectra show blue shift tendency with increasing incident angle (Fig. 2b). Their reflectance amplitudes decrease with increasing incident angle, which is similar to the angle-dependent spectra of the breast-feather barbules in Lawes' parotia.^28,29 However, for the classical multilayer, the reflectance amplitude will increase with the increasing angle of incidence.^27,28 Thus, the upper tapered ridges probably affect the reflectance of the bottom multilayer under oblique incidence and lead to the reflectance amplitude different from that of the classical multilayer. The vanishing of peaks at 60° in reflectance spectra may be caused by imperfect structures of the wing scales such as disordered array and curvature. The valley from ultraviolet to about 480 nm in reflectance spectra, which exhibits low reflection, is strongly suppressed by the pigment filter.³⁰

Structural origin and mechanism of polarization

Optical microscopic images under the polarization microscope reveal that the green and orange scales have the polarization-sensitive effect dependent on the scale azimuth (Fig. 3). The scale under the crossed and collinear analyzers was observed with the orientation of the input linear polarizer fixed. The scale was rotated to investigate its polarization-sensitive property. The input linear polarizer and the collinear analyzer are parallel to the y coordinate axis, and the crossed analyzer is parallel to the x coordinate axis. Under the crossed analyzer, the reflected light almost extinguishes when the scale ridges are perpendicular or parallel to the input linear polarizer (Fig. 3a and e), while it reappears when the scale rotates ±45° to the input linear polarizer (Fig. 3c and g). Under the collinear analyzer, the reflected light fades when the scale ridges are at ±45° azimuth (Fig. 3d and h). The reflected light transmits mainly through the collinear analyzer, when the scale ridges are at 0° or 90° azimuth (Fig. 3b and f). Notably, the scales under polarized light exhibit colors differently from those under nonpolarized light.

Fig. 3 Polarization property of green scales and orange scales. (a) Green scales at 90° azimuth under crossed polarizers. (b) Green scales at 90° azimuth under collinear polarizers. (c) Green scales at 45° azimuth under crossed polarizers. (d) Green scales at 45° azimuth under collinear polarizers. (e) Orange scales at 0° azimuth under crossed polarizers. (f) Orange scales at 0° azimuth under collinear polarizers. (g) Orange scales at −45° azimuth under crossed polarizers. (h) Orange scales at −45° azimuth under collinear polarizers.

A fiber optical spectrometer was equipped on the Zeiss Axio Scope A1 Microscope to collect reflectance spectra to investigate the polarization-dependent reflection of the scales. The input linear polarizer is parallel to the y axis, and the angle between scale ridges and input linear polarizer was 45° (Fig. 4a). Here, the polarized orientation of p-polarized light is parallel to the y coordinate axis and the polarized orientation of s-polarized light is parallel to the x coordinate axis. Under the crossed polarizers, the s-polarized reflection bands are at 500–600 nm and 650–730 nm in the green and orange scales, respectively. Under the collinear polarizers, the p-polarized reflectance of the green and orange scales is approximately less than 5% (Fig. 4b). The polarization-dependent reflected spectra reveal that when the scale is at the 45° azimuth, it selectively reflects the incident light and rotates the p-polarized incident light into s-polarized light.

Fig. 4 Experimental and simulated polarization-dependent reflectance spectra of the 45° azimuth scale. (a) The 45° azimuth scale model. (b) The experimental polarization-dependent reflectance spectra of green and orange scales. (c) The simulated polarization-dependent reflectance spectra of green and orange scales.

Previous studies show that polarized light can be caused by differential reflection or differential scattering, double reflection, or differential absorption.^31,32 To demonstrate that the polarization conversion is caused by the anisotropic ridge structure rather than the pigments, we set the imaginary part of the material's RI to zero to eliminate the influence of pigments.

The normal incident light is a p-polarized light, and the combined architecture is at 45° azimuth (Fig. 4a). Simulation was performed with FDTD, and the calculated reflectance spectra are shown in Fig. 4c. The simulated s-polarized reflectance spectra show a reflection band at about 460–640 nm with the green scale model and a reflection band at about 570–730 nm with the orange scale model. Both the calculated p-polarized reflectance with the green and orange scale models is approximately less than 10%. Although the calculated s-polarized reflection bands with the green and orange scale models are wider than their experimental counterparts, both the calculated and experimental s-polarized reflection bands exhibit the same polarization conversion property, suggesting that the polarization conversion should be mainly caused by the structure of anisotropic ridges.

To further clarify the polarization conversion progress caused by the scale model (45° azimuth), we analyzed the optical process using the form-birefringence. Here, we only analyzed the 0th mode light because it is the main contributor to the reflection band. Previous studies show that the triangular grating achieves form-birefringence, leading to nonsymmetric effective RIs.^33,34

The form-birefringence effective RI theory³³ is given by.

where n_∥ and n_⊥ are the effective RIs in the parallel and perpendicular grating orientation, respectively; f is the volume fraction, n_air is the RI of air, and n_r is the effective RI of the ridge.

The different effective RIs along parallel and perpendicular grating orientations should cause a phase difference. We decomposed the incident p-polarized light at normal incidence into two polarized light components, the parallel and perpendicular grating lights (Fig. S6, ESI†). The phase difference, δ_i, is described with the following equation:

where λ_i is the wavelength of light, is the infinitesimal segment in the tapered ridge, d_h is the distance from the infinitesimal segment to the top of ridge, h_r is the height of the ridge, and n_∥ and n_⊥ are the effective RIs along the parallel and perpendicular grating orientations. The relevant calculation is shown in Text S2, ESI.†

The s-polarized reflection intensity I_iR(s) and the p-polarized reflection intensity I_iR(p) are given by³⁵

where λ_i is the wavelength, I_i0 is the intensity of incident p-polarized light, and R_i is the reflectance of the multilayer. The relevant calculation is shown in Text S3, ESI.† The obtained results show that I_iR(s) = I_i0R_i sin²(0.25π/λ_i) for green scale and I_iR(s) = I_i0R_i sin²(0.32π/λ_i) for orange scale. The values of the sin²(δ_i/2) approximately match those of R_i in both the green and orange scales, suggesting that the height of the grating may be essential for the polarization-dependent reflection. These calculated results can well explain our experimental observations that the color of the 45° azimuth scales is intense under the crossed analyzer (Fig. 3c and g) whereas it dims under the collinear analyzer (Fig. 3d and h).

When the linearly polarized light illuminates on the 45° azimuth quarter-wave plate, the transmitted light will be circular polarized.³⁶ In our model, the upper triangular grating acts as a quarter-wave plate. When p-polarized light illuminates on the 45° azimuth simplified model, the transmitted light will be circular polarized. Our study reveals that the reflected light will be s-polarized. Therefore, the entire combined architecture at 45° azimuth selectively reflects and rotates the incident p-polarized light into the s-polarized light (Fig. 5).

Fig. 5 Schematic diagram of polarization conversion process. When the p-polarized incident light illuminates on the 45° azimuth scale model, the reflected light is s-polarized light, and the transmitted light is circular-polarized light.

Conclusions

The iridescent scales of two Ornithoptera butterflies, O.p.poseidon and O.c.lydius, exhibit a polarization conversion phenomenon at 45° azimuth, which is caused by an elaborate architecture consisting of tapered ridges covering a multilayer in the scale. The tapered ridge structure serves as triangular grating, which causes form-birefringence and hence different phases in different component lights, parallel and perpendicular grating lights, depending on the height of the triangular grating. The multilayer structure selectively reflects the incident light, contributing to the reflection band and deciding the intense color. Hence, the entire scale architecture consisting of the tapered ridges and the multilayer can selectively reflect and rotate the incident polarized light, leading to the polarization-dependent reflection and polarization-sensitive colors. These findings provide the design guide for biomimetic optical materials with polarization-sensitive structural colors, showing significant promise in biosensing and anticounterfeiting.

Acknowledgements

The authors are grateful for the financial support from the National Natural Science Foundation of China (51172141) and the Research Fund for the Doctoral Program of Higher Education (20100073110065 and 20110073120036).

References

1. E. Yablonovitch, J. Mod. Opt., 1994, 41, 173–194.
2. M. Srinivasarao, Chem. Rev., 1999, 99, 1935–1961.
3. A. R. Parker, Philos. Trans. R. Soc., A, 2004, 362, 2709–2720.
4. I. B. Burgess, J. Aizenberg and M. Lončar, Bioinspiration Biomimetics, 2013, 8(4), 045004.
5. M. Kolle, Photonic structures inspired by nature, Springer, Heidelberg, 2011, pp. 30–55.
6. P.-Y. Chen, J. McKittrick and M. A. Meyers, Prog. Mater. Sci., 2012, 57, 1492–1704.
7. K. L. Yu, T. X. Fan, S. Lou and D. Zhang, Prog. Mater. Sci., 2013, 58, 825–873.
8. M. Kolle, A. Lethbridge, M. Kreysing, J. J. Baumberg, J. Aizenberg and P. Vukusic, Adv. Mater., 2013, 25(15), 2239–2245.
9. A. R. Parker, J. Exp. Biol., 1998, 201, 2343–2347.
10. V. Saranathan, C. O. Osuji, S. G. J. Mochrie, H. Noh, S. Narayanan, A. Sandy, E. R. Dufresne and R. O. Prum, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 11676–11681.
11. V. Sharma, M. Crne, J. O. Park and M. Srinivasarao, Science, 2009, 325, 449–451.
12. S. Vignolini, P. J. Rudall, A. V. Rowland, A. Reed, E. Moyroud, R. B. Faden, J. J. Baumberg, B. J. Glover and U. Steiner, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 15712–15715.
13. P. Vukusic, J. R. Sambles and C. R. Lawrence, Nature, 2000, 404, 457.
14. S. Yoshioka and S. Kinoshita, Opt. Express, 2007, 15, 2691–2701.
15. J. M. Douglas, T. W. Cronin, T. H. Chiou and N. J. Dominy, J. Exp. Biol., 2007, 210(5), 788–799.
16. T. W. Cronin, N. Shashar, R. L. Caldwell, J. Marshall, A. G. Cheroske and T. H. Chiou, Integr. Comp. Biol., 2003, 43, 549–558.
17. S. Kleinlogel and A. G. White, PLoS One, 2008, 3, e2190.
18. A. Kelber, Nature, 1999, 402, 251.
19. A. Kelber, C. Thunell and K. Arikawa, J. Exp. Biol., 2001, 204, 2469–2480.
20. A. Sweeney, C. Jiggins and S. Johnsen, Nature, 2003, 423, 31–32.
21. H. L. Leertouwer, B. D. Wilts and D. G. Stavenga, Opt. Express, 2011, 19(24), 24061–24066.
22. S. Kinoshita, S. Yoshioka and J. Miyazaki, Rep. Prog. Phys., 2008, 71, 076401.
23. S. Lou, X. M. Guo, T. F. Fan and D. Zhang, Energy Environ. Sci., 2012, 5, 9195–9216.
24. T. Clausnitzer, T. Kämpfe, E. B. Kley, A. Tünnermann, U. Peschel, A. V. Tishchenko and O. Parriaux, Opt. Express, 2005, 13(26), 10448–10456.
25. X. F. Jing, S. Z. Jin, J. C. Zhang, Y. Tian, P. Liang, H. B. Shu, L. Wang and Q. M. Dong, Opt. Lett., 2013, 38, 10–12.
26. A. V. Tishchenko, Opt. Quantum Electron., 2005, 37, 309–330.
27. D. G. Stavenga, B. D. Wilts, H. L. Leertouwer and T. Hariyama, Philos. Trans. R. Soc., B, 2011, 366, 709–723.
28. D. G. Stavenga, H. L. Leertouwer, N. J. Marshall and D. sorio, Proc. R. Soc. B, 2011, 8, 2098–2104.
29. B. D. Wilts, K. Michielsen, H. De Raedt and D. G. Stavenga, Proc. Natl. Acad. Sci. U. S. A., 2014, 111(12), 4363–4368.
30. B. D. Wilts, T. M. Trzeciak, P. Vukusic and D. G. Stavenga, J. Exp. Biol., 2012, 215, 796–805.
31. G. P. Konnen, Polarized light in nature, Cambridge University Press, 1985, pp. 131–143.
32. T. W. Cronin and J. Marshall, Philos. Trans. R. Soc., B, 2011, 366(1565), 619–626.
33. W. J. Yu, T. Konishi, T. Hamamoto, H. Toyota, T. Yotsuya and Y. Ichioka, Appl. Opt., 2002, 41, 96–100.
34. F. Wang, M. Xiao, K. Sun and Q. H. Wei, Opt. Express, 2010, 18, 63–71.
35. Y. B. Liao, Polarization optics, Beijing Science press, 2003, pp. 101–109.
36. C. Hitzenberger, E. Götzinger, M. Sticker, M. Pircher and A. Fercher, Opt. Express, 2001, 9, 780–790.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra07988d

---