Brochosome-inspired binary metastructures for pixel-by-pixel thermal signature control

Abstract

Microscale thermal signature control using incoherent heat sources remains challenging, despite recent advancements in plasmonic materials and phase-change materials. Inspired by leafhopper-generated brochosomes, we design binary metastructures functioning as pixel twins to achieve pixelated thermal signature control at the microscale. In the infrared range, the pixel twins exhibit distinct emissivities, creating thermal counterparts of “0-1” binary states for storing and displaying information. In the visible range, the engineered surface morphology of the pixel twins ensures similar scattering behaviors. This renders them visually indistinguishable, thereby concealing the stored information. The brochosome-like pixel twins are self-emitting when thermally excited. Their structure-enabled functions do not rely on the permittivities of specific materials, which distinguishes them from the conventional laser–illuminated plasmonic holographic metasurfaces. The unique combination of visible camouflage and infrared display offers a systemic solution to microscale spatial control of thermal signatures and has substantial implications for optical security, anticounterfeiting, and data encryption.

Bionic metastructures achieve pixelated emissivity control while retaining similar visual appearances for information concealment.

INTRODUCTION

Infrared radiation from objects determines their thermal signatures. Because of the incoherent, isotropic, and broadband characteristics of thermal radiation from bulk materials, precisely engineering thermal signatures, including directional (1, 2) and spectral (3, 4) control by metastructures, is critical for a wide range of energy management and conversion technologies, such as thermophotovoltaics (5–7), radiative cooling (8–11), and smart windows (12). Meanwhile, the wide bandwidth and imperceptibility to human ocular system of thermal infrared light have enabled applications in optical security, camouflage, and encryption. With carefully designed boundary conditions (13), the isothermal contours of heated metal plates, and thus the location-dependent thermal radiation, can be manipulated to encrypt letters or patterns for viewing under infrared cameras only, which are covert in the visible range. However, thermal cross-talk between the hot and cold regions due to the diffuse heat conduction hampers the miniaturization of temperature-based spatial control of thermal signatures. To overcome this drawback, by leveraging the scattering fields from nanostructures (14–16), metasurfaces have been designed to tailor (infrared) wavefronts to reveal the concealed patterns and images when illuminated by lasers. Nevertheless, the requirement of external laser sources limits their applications. While advancements have been made in spatial modulation of emissivity, via phase-change materials (17, 18), nanostructured plasmonic materials (19, 20), and gated graphene (21), a universal and systematic solution for heat-driven, microscale thermal signature control is still lacking.

Inspired by leafhopper-produced brochosomes (Fig. 1A and inset), here, we introduce an infrared signature management solution by pixelating surfaces with a spatial resolution down to the micrometer level with a pair of microsized pixel twins. Brochosomes are hollow spherical structures with distributed open pores interconnected by a hollow cavity (22–25). On the basis of the brochosome-inspired pixel twins, we demonstrate a dual effect of concealing binary information/images in the visible range but displaying them in the infrared range via thermal excitation, without relying on external light sources such as lasers. To mimic the binary states in digital electronics, these brochosome-like pixel (BLP) twins are engineered with either open pores (op-BLPs; Fig. 1B, top) or closed pores (cp-BLPs; Fig. 1B, bottom) and perform as the building blocks for spatial emissivity manipulation. As illustrated in Fig. 1C, BLP twins show distinct emissivities when thermally excited, which form infrared counterparts of the binary states 0 and 1. Therefore, images can be encoded using a “bitmap” consisting of the pixel twins, which can then be visualized using infrared imaging systems. This pixel-by-pixel approach uses the unique brochosome geometries, instead of specific material property for emissivity control, and thus is compatible with different substrates. The two types of BLPs share similar appearances when observed under visible imaging systems. This similarity allows the stored patterns formed by one type of BLPs to blend with the background formed by the other. This effect resembles “background matching,” a common camouflage strategy in nature (26, 27). To the best of our knowledge, this is the first time that a pair of microscopic structures has been designed to be distinguishable in the infrared range, while remaining indistinguishable in the shorter visible range, despite the fact that the shorter wavelength typically provides a finer spatial resolution, according to the Rayleigh criterion.

Fig. 1. Thermal signature manipulation by BLPs.

(A) An optical image of a leafhopper Gyponana serpenta. Scale bar, 1 mm. Insets: A scanning electron microscopy (SEM) image of leafhopper-produced brochosomes. Scale bar, 500 nm. (B) Top to bottom: Three-dimensional (3D) models of BLPs with op-BLPs and cp-BLPs, respectively. (C) Schematic of information camouflage and display by BLPs. Information is concealed in the binary array formed by BLPs, which is camouflaged in the visible range but can be displayed under infrared (IR) imaging systems.

RESULTS

BLPs enabled visible camouflage and infrared display

The total irradiance (L) from a BLP reaching the image plane of an imaging system determines its appearance (28–30), and therefore engineering total irradiance in different wavelength ranges is crucial for the design of BLPs. The total irradiance can be expressed as

$$L(\mathbf{x},\mathbf{k})\mathbf{=}\iint f_s(\mathbf{x},{\mathbf{k}}_{\mathbf{i}},\mathbf{k})L_i(\mathbf{x},{\mathbf{k}}_{\mathbf{i}}) \text{cos} {\mathrm{\theta }}_{i}d\mathbf{x}\mathbf{+}{L}_e(\mathbf{x},\mathbf{k})$$ (1)

where L(x, k) represents the total irradiance from the location x on the surface of a BLP with a wave vector k (see Fig. 2A for notations). In Eq. 1, the first term on the right-hand side represents the scattering of the BLP (blue arrow in Fig. 2A) with respect to the light source L_i(x, k_i), and the surface integration encompasses the illuminated surface of the BLP. f_s(x, k_i, k) is the bidirectional scattering distribution function (BSDF) that characterizes the scattering behavior. The second term, L_e(x, k), accounts for thermal emission from the BLP (red arrow in Fig. 2A), which mainly occupies the infrared spectrum when the BLP is thermally excited to a temperature slightly above room temperature, based on Wien’s displacement law. In the visible range, Eq. 1 is dominated by the scattering term due to the negligible contribution from emission, while the emission term L_e(x, k) reveals the key electromagnetic behaviors of BLPs in the infrared range. To facilitate the short-wavelength camouflage and long-wavelength display, it is desired that the two types of BLPs exhibit similar visible scattering while having distinct infrared emission. To demonstrate this, we performed finite-difference time-domain (FDTD) simulations for op-BLPs and cp-BLPs in both the infrared and visible ranges (see Materials and Methods). The diameters of pores (d) and BLPs (D) (see fig. S1, B and D) are the key design parameters. We started the design process by adopting the ratio d/D = 0.28, which is an average derived from natural brochosomes (31) produced by diverse leafhoppers, as shown in Fig. 2B. Subsequently, the ratio was optimized to achieve the desired camouflage and display functions, and then the size of BLPs can be chosen according to the target working wavelengths. Here, the diameter D and pore size d of BLPs are set to be 20 and 3.7 μm, respectively, for our infrared measurements (see Materials and Methods). The material of BLPs is chosen to be nickel (Ni) or other reflective metals, which provide a high-permittivity contrast between the BLPs and the environment.

Fig. 2. Multispectral design of BLPs.

(A) Schematic of light-BLP interactions. (B) Pore diameter to BLP diameter ratio d/D of natural brochosomes produced by various leafhoppers, from which an averaged ratio of 0.28 is acquired and used as the initial design of the BLPs. Data are acquired from (31). (C) Simulated BSDFs in the visible range for both a cp-BLP (solid lines) and an op-BLP (dashed-dotted lines) with respect to normal illumination. a.u., arbitrary units. (D) Simulated absorption cross-sectional areas σ of an op-BLP, a cp-BLP, and a sphere. The outer diameter of the simulated BLPs and sphere is 20 μm, and the material is nickel (Ni). (E) Excess absorption cross-sectional areas Δσ of BLPs compared to the sphere. (F and G) Simulated spatial distribution of the magnitude of the electrical field [λ=4.5 μm, marked by the dashed line in (E)] near an op-BLP and a cp-BLP, respectively.

In the visible range, we simulated the BSDFs of both types of BLPs to demonstrate their scattering behaviors. Conventionally, determining BSDF(θ_o, θ_i) requires the simulations/measurements of the energy distribution at any given scattering angle θ_o with respect to illuminations from various directions θ_i, as marked in Fig. 2A. Given that optical microscopies used for observing microscale BLPs usually have illuminations primarily from the vertical direction, we limited the visible-range simulation of the BLPs to the scattering distribution functions with respect to the normal illumination only, i.e., BSDF(θ_o, ⊥ ). In Fig. 2C, we compared the simulated results for both op-BLPs (solid lines) and cp-BLPs (dashed lines) at various visible wavelengths. The similarity of BSDFs between the two types of BLPs can be attributed to the identical arrangement (20, 32) of surface features (either open pores or closed pores, see Supplementary Note). These similar BSDFs ensure that the pixel twins cannot be easily distinguished under visible imaging systems, thereby establishing an effective visible camouflage effect.

In the infrared range, we calculated the spectral absorption cross sections (σ) for both types of BLPs. The magnitude of σ also indicates the emission capability of the two structures, based on Kirchhoff’s law (30). Figure 2D showed the results for a single op-BLP and cp-BLP. For comparison, we included the results for a Ni sphere with the same diameter (20 μm). Within the wavelength range from 2.5 to 8 μm, the op-BLP shows an enhanced absorption cross section, compared to those of the cp-BLP and the sphere. The contrast of σ between the op-BLP and cp-BLP structures can be further illustrated in Fig. 2E, where we plotted their excess absorption cross section (Δσ = σ_BLP − σ_sphere) relative to that of the sphere. Such an absorption contrast between the BLP twins occurs in a broad wavelength range overlapping with the dominant wavelength of thermal radiation at a temperature slightly above room temperature and thus only requires thermal excitation to be detected. This characteristic differentiates BLPs from other plasmonic metasurfaces, which require external laser illumination. The enhanced σ of the op-BLP within this wavelength range can be explained by the exposed cavity formed by its open pores. As shown in Fig. 2 (F and G), the electromagnetic field penetrates the op-BLP (Fig. 2F), becomes trapped within the exposed cavity, and eventually gets absorbed because of the electron damping effect of Ni. Conversely, this cavity effect remains absent in the cp-BLP (Fig. 2G), which results in the contrast of σ between these two structures. Note that such a high contrast gradually decreases after it reaches the maximum at the wavelength around 5 μm because the long-wavelength electromagnetic waves cannot penetrate BLPs even with the existence of the open pores (fig. S2). Therefore, as a general design principle, one can tune the diameter of the open pores (the diameter of the closed pores also needs to be tuned accordingly to ensure the similar visible appearances between the two types of BLPs) to optimize the BLPs for the targeted working wavelength range. By choosing the metal coatings with different damping rates, the contrast between the two types of BLPs can be further engineered. In addition, the unique multipore configuration of BLPs is crucial to achieve the high contrast because simple spherical structures with fewer or single open pore will substantially reduce the amount of light reaching the cavity, thus leading to diminished absorption cross sections (fig. S3A).

Thermal-optical characterization of BLPs

To quantify the emission contrast, we fabricated arrays of both cp-BLP and op-BLP and measured their spatial distributions of emissivity using an infrared thermal mapping system (see Materials and Methods). In Fig. 3 (A and B), both the nine-by-nine BLP arrays were fabricated using the two-photon polymerization three-dimensional (3D) printing and subsequently coated with ~100 nm of Ni [see fig. S4 for the enlarged scanning electron microscopy (SEM) images and the x-ray element mapping]. The thickness of the Ni coating is more than two times of its skin depth ( ${\mathrm{\delta }}_e=\frac{\mathrm{\lambda }}{2\mathrm{\pi }\mathrm{Im}(\tilde{n})}\approx 40 \text{nm}$ , where $\text{Im}(\tilde{n})$ represents the imaginary part of the refractive index) within the working wavelength range of our infrared imaging system (see Materials and Methods), ensuring efficient shielding of electromagnetic waves. Consequently, the cavity effect is pronounced only for op-BLPs. Figure 3 (C and D) plots the spatial distributions of the measured emissivity for the op-BLP and cp-BLP arrays, respectively, from which the spatially averaged emissivities were calculated on the basis of the central seven-by-seven BLPs (the BLPs located at the edges are excluded because of the edge effect). The measured average emissivities are 0.4189 and 0.3073 for the op-BLP and cp-BLP arrays, respectively. These experimental measurements agree well with our simulation results, 0.449 and 0.301, which are obtained by integrating simulated spectral-directional emissivity ε_{λ, θ} (Fig. 3, E and F) for both arrays, respectively (see Materials and Methods). The spectral contrast in infrared emissivity for both arrays can also be revealed by the Fourier transform infrared measurements (see fig. S3B).

Fig. 3. Emissivity contrast of BLPs.

(A and B) SEM images of fabricated op-BLP array and cp-BLP arrays, respectively. Scale bars, 20 μm. (C and D) Measured spatial distribution of emissivities of op-BLP and cp-BLP arrays, respectively. Scale bars, 50 μm. Insets: Typical local enlarged images of single BLPs from which their emissivities are evaluated. (E and F) Simulated spectral-directional absorptivity (SDA) of op-BLP and cp-BLP arrays, respectively. The averaged emissivities can be calculated according to Eqs. 2 to 4 (see Material and Methods), in which the integration domains are marked by the dashed white boxes.

Notably, the interstices between neighboring BLPs in both arrays show high emissivities (~0.5) because they also act as optical cavities enhancing the thermal radiation from the substrate in the interstices, where the Ni coating may not be conformal (see fig. S4) and, thus, the Si substrate may be partially exposed (see Materials and Methods for the specification of the Si substrate). Such a high background emissivity compromises the emissivity contrast between the two types of BLPs, when estimated at the array level as described above. Here, by calculating the average local emissivities within arbitrarily selected regions containing only individual BLPs (insets of Fig. 3, C and D), the emissivities of single op-BLPs and cp-BLPs are estimated to be 0.3329 and 0.1329, respectively, which gives rise to an emissivity contrast ratio of approximately 2.5:1. In comparison, the optical images (shown in fig. S5, A and B) of the two arrays were analyzed via the quantitative color pattern analysis method (33), which showed well-matched pattern energy distributions (fig. S5C), indicating the visual indistinguishability of the two arrays to human eyes at visible wavelengths.

Experimental demonstration of visible camouflage and infrared display

The BLP twins represent a pair of binary states under infrared imaging systems due to their emissivity contrast, whereas their nearly identical visible scattering behaviors render them almost indistinguishable to human eyes. Such a wavelength-dependent distinguishability lays the foundation for their camouflage and displays functionalities in the visible and infrared ranges, respectively. We optimized the design of BLPs by studying the influence of the pore shape and size on the proposed camouflage and display functions. Considering the naturally occurring hexagonal and pentagonal open pores found in leafhopper-produced brochosomes (inset of Fig. 1A), we designed the BLP arrays composed of op-BLPs and cp-BLPs with polygonal pores in addition to the round ones, as shown in Fig. 4A-a for the particle morphology map and Fig. 4A-b for the SEM images. The array was then examined under both an optical microscope (Fig. 4A-c) and an infrared microscope (Fig. 4A-d) at the same magnification. Our observations indicate that variations in pore shape have a minimal impact on their performance. On the basis of the circular-pore design, we fabricated another array incorporating BLPs with varying pore sizes (Fig. 4B). We found that increasing pore diameter enhances the infrared emissivity contrast between the two pixels. However, increasing pore diameter simultaneously reduces the visual concealment between the op-BLPs and cp-BLPs.

Fig. 4. Visible camouflage and infrared display based on optimized BLPs.

(A and B) Optimization of the shape and size of pores in BLPs, respectively. Scale bars, 60 μm. (A-a) Design of the BLPs array, with P-1 to P-4 representing BLPs with circular open pores, circular closed pores, polygon open pores, and polygon closed pores, respectively. The diameter of the circular pores is 3.7 μm, and diameter of polygon pores is 5 μm. In (B-a), P-1 to P-8 represent op-BLPs with d = 3.7 μm, bare spheres, op-BLPs with d = 4.0 μm, cp-BLPs with d = 4.0 μm, op-BLPs with d = 5.0 μm, cp-BLPs with d = 5.0 μm, op-BLPs with d = 6.0 μm, and cp-BLPs with d = 6.0 μm, respectively. (A and B, b to d) an SEM image, an optical image, and an infrared image of the designed BLP arrays. For the best visible camouflage and infrared display effect, combination of circular op-BLPs/cp-BLPs were used in the demonstration [red and yellow combination in (A-a)]. (C) BLP array forming letters C and P. Scale bar, 100 μm. (D) BLP array forming a QR code (storing the address of the homepage of Carnegie Mellon University). Scale bar, 150 μm. (C and D) (a) SEM images of the BLP array; (b) optical images of the BLP arrays. The patterns are camouflaged through background matching; (c and d) infrared images and calculated spatial distribution of emissivity based on the infrared images, displaying patterns due to the emissivity contrast between different BLPs. The infrared images in (C-d) and (D-d) are taken using the objectives with ×20 and ×12 magnifications, respectively.

Using the optimized BLPs pair (D = 20 μm and d = 3.7 μm, round pores, as those in the top left quadrant of the array shown in Fig. 4A), we demonstrated the pixel-by-pixel camouflage and display functions. Figure 4C-a shows an SEM image of two BLP arrays, where we patterned the letters “C” and “P” using the cp-BLPs within the background formed by op-BLPs. The square-packed 20-μm-diameter BLP pixels result in an effective pixel density of 1270 pixel per inch, which is more than five times denser than typical micro–light-emitting diode displays used in smart devices (34), and the BLPs can also be closely packed into a hexagonal lattice for an even greater pixel density (see fig. S6). Figure 4C-b demonstrates the visible camouflage effect as observed through an optical microscope with a ×20 magnification objective. In this view, the two types of BLPs are indistinguishable, causing the patterns to blend into the background. However, the emissivity difference between the two types of BLPs offers a pronounced contrast to display the hidden patterns in the infrared range, as shown in Fig. 4C-c, which is captured at the same magnification but with an infrared microscope. Figure 4C-d shows the calculated spatial distribution of emissivity from Fig. 4C-c, where the typical emissivities of each individual op-BLPs and cp-BLPs are approximately 0.35 and 0.15, further verifying the theoretical contrast ratio of around 2.5:1. The thermal stability analysis (see Supplementary Note and fig. S7) also indicates that the BLP twins enable applications in a temperature range from 50° to 110°C. Furthermore, because of the highly symmetrical arrangement of pores on the BLP surfaces, the infrared emission contrast between the two types of pixels can be maintained even when viewed from different angles. As shown in fig. S8, the C and P patterns remain discernable under infrared microscopes when tilted by a small angle (~6° achieved by our experimental setup), thus accommodating angular misalignments during the view process. Note that the maximum viewing angle is not limited by the BLPs but rather the focal depth of the infrared microscope. We further quantify the camouflage and display capabilities of the BLPs by counting the pixel fractions at different brightness level (see fig. S9) based on the method in (35). This analysis revealed that the BLPs preserve a certain degree of camouflage ability even under visible-range dark-field imaging systems.

We extended the application of BLPs by crafting a BLP array patterned into a quick-response (QR) code (version 1, ECC level L) containing the web address of Carnegie Mellon University, where the cp-BLPs formed the QR code, while the op-BLPs composed the background, as shown in Fig. 4D. The dimension of the QR code is 21 pixel by 21 pixel with a 2-pixel margin on each sides, resulting in a physical size of 0.5 mm by 0.5 mm. Compared with other nano/microstructure-based QR codes, our pixel size and overall QR code dimensions are much smaller than those based on photonic crystals (36, 37) and are comparable to those derived from imprinted polyvinyl alcohol nanostructures (38). This is particularly notable considering that our QR code is designed to function at a longer wavelength. The QR code remains effectively camouflaged during visible-range scanning, as demonstrated in the visible optical image shown in Fig. 4D-b. Conversely, the infrared images of the BLP pixel array (Fig. 4D-c) and calculated emissivity (Fig. 4D-d) clearly display the encrypted QR code, thus allowing rapid scanning using common QR code scanning applications on smart devices (see movie S1). In practice, slightly blurring the infrared images of the BLP QR code, either by defocusing the infrared microscope or by applying Gaussian filters (39, 40), can mitigate the local emissivity variations within single BLPs. This approach can better reveal the binary digits akin to traditional QR codes, making the BLP QR code easier to scan (see fig. S10).

DISCUSSION

To summarize, inspired by leafhopper-produced brochosomes, we created a pair of BLP twins with either open pores or closed pores for microscale thermal signature control. On the basis of the unique emissive and scattering properties of the BLP twins, we developed a pixel-by-pixel approach for spatial thermal signature control and demonstrated a dual effect of visible camouflage and infrared display via thermal excitation. The BLP-based camouflage effect under visible-range imaging systems arises from their similar visible scattering behaviors, while the high emissivity contrast (2.5:1) between the two types of BLPs enables the thermal display of the concealed information by infrared imaging systems. Although it is well known that the shorter wavelength provides a finer spatial resolution based on the Rayleigh criterion, we designed a microscopic structure pair to be distinguishable in the infrared range but indistinguishable in the visible range. Using the BLP twins as the fundamental building blocks, our pixel-by-pixel approach allows for systemic and scalable control of thermal signatures, which paves the way for various optical security applications, such as anticounterfeiting and encryption.

MATERIALS AND METHODS

Design and modeling of BLPs

The design and 3D modeling of BLPs are conducted in SolidWorks, a computer-aided design software. The designed structures are then exported as standard triangle language files (.stl) for optical simulations and fabrications. The 3D and cross-sectional views of both a BLP with op-BLP and with cp-BLP are shown in fig. S1 (A to D), respectively. Each BLP has six loops of pores evenly distributed among the surface. Each loop contains from top to bottom, 1, 6, 10, 14, 14, and 10 pores, respectively. The diameter of pores d and the diameter of BLPs D (shown in fig. S1, B and D) are the key design parameters. We started the design by adopting the ratio d/D = 0.28, which is the average value acquired from natural brochosomes generated by various leafhoppers (Fig. 2B), and then optimized such ratio for best wavelength-dependent distinguishability between the two BLPs (Fig. 4, A and B). The actual size of the BLPs is designed to ensure strong infrared contrast between the two structures within the entire working wavelength range of the infrared imaging systems (2.5 to 4.2 μm). Results shown in the Results section are achieved with BLPs with D = 20 μm and d = 3.7 μm.

FDTD simulation of BLPs

The FDTD simulations of the BLPs were conducted in Ansys Lumerical FDTD Solutions with the .stl files generated by SolidWorks. The diameter of single BLPs is scaled to 20 μm. The material of the BLPs is set to nickel (Ni); a silicon (Si) substrate (the blue block in Fig. 1a), coated with 100-nm-thick Ni, is also considered in simulations. Both the materials are characterized by their permittivities, which are acquired from (41).

In the simulations of absorption cross section σ of single BLPs, the total-field scattered-filed source is used to excite the BLPs, and σ is calculated by a built-in analysis module, absorption analysis group, of the software. To generate the field profiles, single BLPs are excited by a plane wave source linearly polarized along the x direction and propagating along the z direction, and a frequency-domain field profile monitor is used to capture the field profiles at the central cross section of the BLPs parallel to the z axis (see Fig. 1A for the coordinate system). Perfectly matched layers (PMLs) are used in both simulations as boundary conditions to remove unphysical echoes. For the simulations of scattering distribution function of the BLPs, a plane wave source (polarized along the x direction and propagated along the z direction) with Gaussian profiles is used to illuminate single BLPs, and a 80-μm by 80-μm 2D power monitor parallel to the x-y plane located 13.5 μm away from the center of the BLPs is used to captured the Poynting vectors of the scattered field, which are then projected to the far field through the built-in module, far-field projection, of the software to calculate the angular distribution. In the simulations regarding the spectral-directional absorptivity (SDA) of BLP arrays, the plane wave source with the type of “broadband fixed angle source technique” is used to illuminate one unit cell. The boundary conditions at directions transverse to the source propagation direction are handled by the software, and the rest boundary conditions are set to PMLs. To be comparable with unpolarized thermal absorption/emission, simulated SDAs shown in the Fig. 3 are averaged over both the S and P polarization states. Then, the averaged SDAs are integrated to calculate the overall emissivity (30)

$$\mathrm{\epsilon }=\frac{1}{\mathit{AB}}{\displaystyle {\int }_{{\mathrm{\lambda }}_{\text{min}}}^{{\mathrm{\lambda }}_{\text{max}}}}\mathit{d}\mathrm{\lambda }{\displaystyle {\int }_0^{{\mathrm{\theta }}_{\text{max}}}}{\mathrm{\epsilon }}_{\mathrm{\lambda },\mathrm{\theta }} \sin \mathrm{\theta } \text{cos} \mathrm{\theta } E_{\mathrm{\lambda },b}(\mathrm{\lambda },T) d\mathrm{\theta }$$ (2)

where

$$A={\displaystyle {\int }_0^{{\mathrm{\theta }}_{\text{max}}}}\text{sin} \mathrm{\theta } \text{cos} \mathrm{\theta } d\mathrm{\theta }$$ (3)

and

$$B={\displaystyle {\int }_{{\mathrm{\lambda }}_{\text{min}}}^{{\mathrm{\lambda }}_{\text{max}}}}{E}_{\mathrm{\lambda },b}(\mathrm{\lambda },T)d\mathrm{\lambda }$$ (4)

in which E_λ,b(λ, T) represents the blackbody radiation. θ_max, λ_min, λ_max, and T are set to 10°, 2.5 μm, 4.2 μm, and 323.15 K, respectively, to match with the specifications of the thermal mapping system used in measurement.

Fabrication of BLPs

The BLP arrays were fabricated by two-photon polymerization 3D printing method using the 3D files of BLPs (.stl) designed in SolidWorks. The 3D printing process was executed by a 3D printer Nanoscribe, with a resolution of 200 to 500 nm. Specifically, the photoresist IP-Dip of about 50 μL was added onto a silicon substrate (p-type/boron-doped 〈100〉, a resistivity of 1 to 5 ohm·cm; Purewafer, USA), and the photoresist will be polymerized by a laser beam of 780 nm. The BLP structures were printed, while the focus point of the laser beam moves spatially as directed by the 3D modeling file. Last, the uncured polymer was removed by immersing sample into the SU-8 developer for 10 min, and then the sample was rinsed with isopropyl alcohol for 2 min. After that, a 100-nm-thick nickel layer was deposited to the 3D printed BLPs with the electron beam evaporator (Temescal FC-2000, Ferrotec, USA). The temperature of the samples during the deposition process was maintained around 289.15 K (16°C).

Characterization of BLPs

The morphologies of the fabricated BLP samples are examined under an SEM (Merlin, Zeiss, Germany) for quality control. The infrared images are taken by the thermal mapping system (QFI InfraScope) with a ×20 magnification objective (×12 for the QR code sample unless otherwise specified, thermal stability analysis shown in fig. S7, and the tilted measurement shown in fig. S8). The thermal mapping system captures the infrared emission from the samples within the wavelength range from 2.5 to 4.2 μm and normalizes it to the blackbody emission to calculate the emissivity. During the infrared imaging process, the BLP samples are heated to 323.15 K (50°C) by a Linkam HFS600 thermal stage, unless otherwise specified in the thermal stability analysis shown in fig. S7. The infrared reflection spectrum of the two samples shown in fig. S3B are measured by the Fourier transform infrared microscope (Nicolet Continuum). Note that because of the spherical shape of BLPs, the reflected light is distributed in the upper hemispherical space, so the majority of which cannot be received by the Fourier transform infrared system, resulting in the extremely low measured reflectivity. So, it cannot be used to estimate the absorptivity/emissivity of the structure.

The optical images of the BLP arrays were taken by Nikon digital sight DS-Fi1 with a ×20 magnification objective (×10 for the QR code sample). The similarity/indistinguishability of the images of different BLP arrays is assessed by the quantitative color and pattern analysis method (33), implemented in the Mica Toolbox. Specifically, images of both the op-BLP and cp-BLP (fig. S5, A and B) arrays discussed in Fig. 3 were captured with the same illumination, and camera settings are imported to the software, where the significance/notability (measured in pattern energy) are analyzed for features of different sizes (measured in pixel) in the images. Figure S5C compared the pattern energy distribution of the two images, which overlap with each other, indicating a high degree of similarity between them to human eyes.

Supplementary Materials

This PDF file includes:

Supplementary Note

Figs. S1 to S10

Legend for movie S1

Other Supplementary Material for this manuscript includes the following:

Movie S1