Artwork often needs to be verified for authenticity during transactions. Electronic tags and labels are easily spotted and hence counterfeited. A covert and effective tag needs to be: (1) microscopic, (2) camouflaged into the surroundings, and (3) contain multiple sets of information. Here, we developed aluminum (Al) nanostructures that resonate across the ultraviolet (UV) to infrared (IR) spectra for use in micro-tags with varying colors of similar brightness and containing two sets of information in the visible and IR. Native Al2O3 on Al films was measured to be ~4–7 nm thick, enabling resonances to be supported by Al disks with diameters merely ~1/6th of the wavelength at the fundamental mode. Through accurate modeling of the nanostructures and high-resolution electron-beam lithography, we designed and fabricated a printed micro-tag on silicon. This micro-tag requires image processing to extract a quick response (QR) code in the visible, and 1.2 IR illumination (or visible light darkfield imaging) to extract a covert barcode. The compact and multi-spectral encoding of prints demonstrated here is particularly suited for discreet tagging of art and high-value merchandise.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Colors in the visible spectrum are the basis of human vision, a highly developed part of the nervous system, and serve an important role in nature, signaling, technology and the arts. In addition to pigments, metallic and dielectric nanostructures act as “geometry-tunable colors” due to the excitation of surface plasmons and Mie scattering. Color generation has been demonstrated [1,2] using many types of structures, including metal disks above a perforated back-reflector [3–5], dielectric resonators above dielectric substrates [6,7] and all-metallic structures [8,9]. These designs have expanded the technology of color printing, as the resolution has been increased to 1 × 105 dots per inch , and the attainable color gamut has exceeded that of standard Red Green Blue (sRGB) . There has also been strong interest in designing structures that generate different visual responses by a change in the polarization of light, e.g. using structures with elliptical  or rod shapes .
Conversely, prints in the infrared (IR) spectrum  are less well understood and largely based on IR inks . IR prints would enable a discreet channel of security tagging and encryption, as they are invisible to humans. In this paper, we use gap plasmons in a thin layer of native aluminum (III) oxide (Al2O3) to create micro-tags that can be viewed under visible and infrared illumination, allowing for two sets of images to be printed onto the same area. Only the infrared image can be observed when the print is viewed under infrared illumination, and vice versa, only the visible color image is observable when the print is viewed using visible light. The infrared resonance arises from the fundamental gap plasmon mode, while both fundamental and higher order modes contribute to producing resonances in the visible regime.
Gap-plasmon resonators are formed when a thin layer of dielectric is sandwiched by two layers of electrical conductors [14–17]. The sandwich structure supports a localized surface plasmon mode with tight confinement of the electric and magnetic fields. In particular, if the top layer is comprised of nanodisks, a metasurface is formed that acts as a magnetic mirror  and exhibits resonances with desirable features including perfect absorption, high spectral selectivity and strong wavelength tunability. As the gap decreases, the resonance modes redshift due to the larger capacitive coupling between the top and bottom layers [19,20]. Hence, gap plasmon resonances in structures with sub-10 nm thick dielectric layers can span the ultraviolet, visible and infrared spectra while preserving the sub-micron length scale of the nanodisks. Specifically, for a ~7 nm thick Al2O3 layer, we show that the fundamental gap plasmon resonance occurs at a wavelength larger than six times the diameter of the nanostructures in the top layer. Thus, the gap plasmons can be used to generate infrared images, in addition to the visible color images demonstrated in earlier work [21,22].
Aluminum (Al) is a suitable material for this study as it supports surface plasmons [23–25] and forms a self-limiting 3–4 nm thick layer of native Al2O3 on its surface . Native Al2O3 had been used as the insulator layer for metal-insulator-metal (MIM) structures ; Al2O3 had also been deposited using atomic layer deposition  or physical vapor deposition . However, the absorptances of MIM structures with a 3 nm thick Al2O3 layer are low, and strong absorptance at resonance requires at least twice the native oxide thickness. Nanostructures with diameters larger than 120 nm support higher-order modes at visible wavelengths, in addition to the fundamental resonance at infrared wavelengths, whereas smaller nanostructures with diameters less than 100 nm only support the fundamental resonance mode. Thus, images can be created at both visible and infrared wavelengths and recorded using appropriate optics. In addition to micro-tags, the concepts discussed here could find applications for security printing, anti-counterfeiting, and covert information storage.
2.1 Structure of the gap plasmon resonator
Each color-IR pixel consists of a square (arbitrarily chosen) array of Al disks placed on an Al2O3-Al-bulk silicon (Si) substrate. The schematic is shown in Fig. 1(a), while the top-view scanning electron micrograph of the physical sample is shown in Fig. 1(b). The disks were fabricated via electron beam lithography using poly(methylmethacrylate) (PMMA) resist, followed by metal deposition by electron-beam evaporation and lift-off.
The fabrication procedure involved an initial Al deposition step, electron beam lithography, a second Al deposition step, and lift-off to form Al disks (Fig. 2). This process was adapted from previous works by Dong et al.  and Ho et al. .
First, a 100 nm thick film of Al was evaporated onto a bare Si substrate using an electron beam evaporator (Labline, Kurt J. Lesker Co.). The working pressure and deposition rate were 1 × 10−6 Torr and 2 Å/s respectively. Upon exposure to air for one day, a self-limiting ~4 nm thick layer of native Al2O3 was formed.
Next, the positive-tone electron-beam resist PMMA (950K A4, MicroChem Corp.), was spin-coated at 6000 rpm for 60 s and then baked at 180 °C for 60 s to a thickness of 170 nm. Electron beam lithography (EBL) was performed using the eLine Plus, Raith GmbH at 30 kV acceleration voltage and 400 pA beam current. To produce the color palette (Fig. 5), a write field of 100 µm × 100 µm was used, whereas for the security pattern (Fig. 6), a write field of 50 µm × 50 µm was used. Proximity effect correction was performed for the palette due to the larger write field used. The resist was developed in 1:3 methyl isobutyl ketone/isopropanol (MIBK/IPA) at −15 °C for 30 s, rinsed in IPA for 5 s and blow-dried using a nitrogen gun.
Thereafter, a second evaporation step of 44 nm of Al using the same electron beam evaporator was performed as before. Finally, a lift-off process was done to remove the unexposed resist by soaking the samples in acetone at 60 °C, and obtain the desired metal-disk-on-oxide-on-metal structures. The disk diameter ranged from 60 nm to 280 nm, while the inter-disk gap was between 30 nm and 140 nm.
2.3 Optical and infrared microscopy
Brightfield and darkfield optical microscopy, infrared microscopy and reflection spectra measurements were performed on the samples to study their responses to visible and infrared light. The brightfield and darkfield images were taken in reflection mode using an upright compound microscope (Nikon Eclipse LV100ND, Nikon Instruments Inc.) and a digital complementary metal-oxide-semiconductor (CMOS) camera (Nikon DS-Ri2, Nikon Instruments Inc.). The optical images were magnified using a brightfield/darkfield (BD) objective lens (TU Plan Fluor BD 50 × /0.80 NA). Infrared images were taken using a cryogenically cooled infrared analysis system (FEI Meridian IV, FEI Inc.) attached with a 50 × /0.45 NA objective and an indium gallium arsenide (InGaAs) camera (DiamondBack eXtended, DCG Systems Inc.). The light source was a gallium arsenide (GaAs) light emitting diode (LED) with a central wavelength of 1.2 µm and full-width at half-maximum of 0.05 µm.
2.4 Reflectance spectroscopy
Reflectance spectra were measured using a UV-visible-NIR microphotospectrometer (CRAIC QDI 2010, CRAIC Technologies Inc.) with a 75 W xenon lamp at normal incidence. The light was passed through a 7.1 µm × 7.1 µm aperture and an objective lens (36 × /0.5 NA) and the reflected light (0.3–1.7 µm) was collected by charge-coupled device (CCD) detectors. The two detectors are a silicon detector (working range of 200–950 nm) and an InGaAs detector (850–2300 nm). Linear interpolation was done for data points between 1.37 to 1.42 µm to remove the noise in the spectra occurring at ~1.4 µm due to absorption in the optical fiber connecting the lamp to the spectrometer.
2.5 Scanning and transmission electron microscopy
Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were performed on the palette to determine the physical dimensions of the fabricated structures. The SEM images were taken with a 10 keV electron beam using an EBL-SEM system (eLine Plus, Raith GmbH). TEM images were taken using a 200 kV system (Tecnai F20, FEI Inc.) of a cross-sectional piece of the palette, obtained by depositing a protective platinum layer and milling the palette using a focused ion beam system (Helios NanoLab 600i, FEI Inc.) with a gallium ion source.
2.6 Finite-difference time-domain simulations
A commercial finite-difference time-domain (FDTD) package (Lumerical FDTD Solutions, Lumerical Inc.) was used to calculate the electromagnetic fields and reflectance of the structures. A bulk layer of Al was used as the substrate, while a 7 nm thick layer of Al2O3 was placed atop the Al substrate. The disk was modelled as a 44 nm tall tapered cylindrical core of Al with a thin 4 nm coating of Al2O3 on the top and sides. The diameter of the disk was set to a value between 60 nm and 280 nm with 5 nm increments. The side walls of the disk were tapered with a gradient of 2.2, while the edge connecting the top face of the disk and the side wall was curved with a 10 nm radius of curvature. The permittivity data for Al and Al2O3 were obtained from Palik , while the background of the simulation region was taken to be vacuum (refractive index of 1).
A rectangular simulation region was used with the z-axis perpendicular to the surface of the substrate and the simulation region extents in the x- and y-directions both equal to the period of the disk array. Periodic boundary conditions were used for the x-min, x-max, y-min and y-max faces of the boundary, whereas perfectly matched layer boundary conditions were used for the z-min and z-max planes. A uniform mesh of 1 nm step size was used for a rectangular region around the disk, while a non-uniform conformal mesh was used for the other regions. A plane wave source parallel to the xy-plane was placed 700 nm above the surface of the substrate. It was linearly polarized in the x-direction and had a wavelength range of 300 nm to 1700 nm. Field monitors were respectively placed in the xy-plane 100 nm above the source and in the xz-plane through the center of the disk to measure the reflected power and the electric and magnetic fields.
3. Characterization of resonators
The physical dimensions of the disks and substrate were measured by taking a cross-sectional TEM image of the sample (Fig. 1(c)). The thickness of the Al2O3 in the regions of the substrate that are free of disks was measured as 4 nm, while the thickness of the Al2O3 below the disk was 7 nm. The TEM image also shows a 3–4 nm thick Al2O3 layer coating the upper surface, side walls and bottom surface of the disk. Plausibly, a second layer of native oxide formed around the disk after the second deposition leading to a total thickness of Al2O3 below each Al disk to be 7 nm. The Al2O3 might have been formed in two steps: (1) after the base Al layer was deposited onto the Si substrate using an electron beam evaporator, a ~4 nm thick layer of native Al2O3 was formed on the surface; (2) during the lift-off process, an additional ~3 nm layer of Al2O3 was formed all around the disks.
3.1 FDTD simulations and experimental reflectance spectra
Small Al disks with diameters < 100 nm support fundamental resonance modes in the visible range, between 400 nm and 800 nm. Based on the FDTD simulation results in Fig. 1(d), the fundamental resonance wavelength for a disk array with a diameter of 80 nm and a period of 150 nm is ~580 nm. There is a strong absorption peak at resonance, corresponding to a low reflectance of ~0.25. Inter-band transitions between the W2’ and W1 symmetry points in the electronic bandstructure of Al manifest as a dip in the reflectance spectra at ~800 nm wavelength . As the diameters of the disks increase, the fundamental resonance redshifts into the infrared region, and higher order resonances appear at visible and ultraviolet wavelengths. Thus, for a disk array with a diameter of 240 nm and a period of 310 nm, the fundamental resonance occurs at 1700 nm, while the third and fifth order resonances occur at 600 nm and 360 nm respectively. The tunability of strong resonances across both infrared and visible wavelengths suggests that it is possible to control the infrared and visible appearances of the disks by changing the disk sizes and array periods, and to design specific patterns observable at either visible or infrared wavelengths.
To understand the gap plasmon modes, the electric and magnetic fields of the fundamental and odd order modes for a disk array with a disk diameter of 240 nm and a period of 310 nm are plotted in Fig. 1(e). A gap plasmon is excited in the Al2O3 layer when electromagnetic radiation, polarized along the x-axis, is normally incident. For normal incidence, only the odd anti-symmetric modes are excited. Opposite charges accumulate at opposite edges of the base of the Al disk, and are balanced by charges at points on the surface of the Al substrate directly below the edges of the base of the disk. Thus, the electric field is strongest in the Al2O3 gap layer and points in the vertical direction between the base of the Al disk and the substrate. The charges produce a displacement current loop, and a strong magnetic field is formed in the central region of the Al2O3 layer, parallel to the plane of the substrate and perpendicular to the polarization of the plane wave light source. Electromagnetic energy is absorbed efficiently and dissipated in the Al disks and substrate, producing the reflectance minima shown in Fig. 1(d). For the third and fifth order modes, the number of displacement current loops increases to three and five respectively, and the electromagnetic energy that was formerly concentrated in the Al2O3 gap layer extends into the air surrounding the Al disk, i.e. the mode has become less localized.
The oxide thickness affects both the tunable range of wavelength shifts, and the strength of the absorptance () at resonance. Tunability is the highest for the thinnest oxides, while the strength of the absorptance follows a broad envelope with the thinnest oxides having the strongest absorptances at longer wavelengths. The reflectances obtained from FDTD simulations of disk arrays on oxide layers of different thicknesses (3 nm, 7 nm and 20 nm) are presented in Fig. 3(a). We consider disk diameters of 60–240 nm for all thicknesses. In this range of diameters, the thinnest oxide of 3 nm shows the largest resonance shifts, Δλ = 1740 nm (620 nm to 2360 nm). Increasing the oxide thickness blueshifts the resonances and decreases the tunability. For instance, Δλ is halved to ~880 nm for an oxide thickness of 20 nm. Figure 3(b) is a plot of the maxima of the absorptances for disk arrays with the three oxide thicknesses, with each data point taken from one disk diameter. The smallest and largest diameters for the data set of each oxide thickness are indicated. For the disk diameters considered, the maximum absorptance occurs at the fundamental resonance. We observe a clear trend in the maxima of the absorptances, i.e. that the disks on 3 nm thick oxide have weak absorptances in the visible but good absorptances in the IR. On the other hand, the disks on 20 nm thick oxide have stronger absorptances in the visible, and are hence useful in producing plasmonic colors [22,33]. The 7 nm thick oxide thus provides a desirable trade-off between tunability and strong absorptances spanning both visible and IR, since the maximum absorptance is at least 0.55 and Δλ is 1220nm for disk diameters of 60–240 nm.
The experimental reflectance data from square arrays of disks with diameters between 60 nm and 280 nm and a fixed gap of 30 nm are presented in Fig. 4(a). Thus, the pitch size ranges from 90 nm to 310 nm. There are three minima lines observed, i.e. the fundamental, third and fifth order modes as labelled. The fundamental resonance is present for all the disk diameters studied, while the third and fifth order modes are observed for disk diameters larger than 120 nm. The largest shift is for the fundamental resonance, as the resonance wavelength starts at 350 nm for 60-nm-wide disks and redshifts to 1650 nm for 280 nm wide disks. In contrast, the third-order and fifth-order resonances only redshift by 250 nm and 150 nm respectively. Inter-band transitions in Al lead to a reduction of 0.1 in the reflectance around 800 nm, and cause a broadening of the fundamental resonance for disk diameters between 120 nm and 150 nm.
The reflectance data in Fig. 4(b) were obtained from FDTD simulations, for comparison with the experimental data and to verify the accuracy of the structural and material models used. There is good agreement between the two sets of data, as the trend lines for the fundamental, third and fifth order resonances and inter-band transitions are similar for the simulated and experimental reflectance data. However, the simulated fundamental resonances are blue-shifted by about 100 nm compared to the experimental results. The reflectance minima for the fundamental and fifth order resonances measured experimentally are larger than those of the simulated data, as the experimental minima were between 0.2 and 0.3, while the simulated minima were between 0 and 0.2. This discrepancy can be explained by disk diameter variations due to fabrication errors and proximity effects in the EBL process that cause dose reduction at the boundaries of the pixels relative to their centers. The actual oxide thickness may also be slightly larger than the thickness determined from the TEM image. Nonetheless, the linear dependence of resonance wavelength on disk diameter is evidence of the lack of coupling between disks due to the tightly confined modes in the gap, despite the close spacing of the disks fixed at 30 nm. By fitting the data to the gap plasmon resonator formula, , we obtained the effective refractive index of the resonator to be neff ~3, a large value that enables the large wavelength scaling of ~6 × . In the equation, w is the width of the resonator, λ is the resonance wavelength, m is the order of the resonance mode, and ϕ is the phase change due to reflection at the edges of the resonator.
3.2 Optical and infrared micrographs
The brightfield optical image of the fabricated nanodisk arrays is included in Fig. 5(a) to show their responses to visible light. The patches are 10 µm × 10 µm large square arrays of nanodisks. The disk diameter and inter-disk gap in each array were varied from 60 nm to 280 nm and from 30 nm to 140 nm respectively. The increment for each parameter was 10 nm, except for disk diameters between 60 nm and 100 nm, for which the increment was 5 nm. Under visible light, the pixels exhibit a broad range of colors. The colors are largely determined by the disk diameter, as the hues are similar for disks of the same diameter, but the intensity decreases when the gap increases . The discontinuity in color variation for the columns with disk diameters between 60 nm and 90 nm was due to intra-field dose variation during the lithography process.
The colors for disk diameters between 60 nm and 150 nm are caused by the reflectance minimum associated with the fundamental resonance mode described in Fig. 4. When the diameter increases, the colors turn from yellow to blue. This apparent blueshift in color arises due to the subtractive nature of plasmonic colors. The redshift of the resonances with increasing diameter causes the remaining wavelengths that are reflected to consist of shorter wavelength components. The colors are the most saturated for disk diameters between 65 nm and 100 nm, because the fundamental resonances are strong and centered at a wavelength between 400 nm and 800 nm. For even larger disks, the fundamental resonance shifts into the IR regime and no longer affects the optical response at visible wavelengths. The colors are a result of the third and fifth order resonances instead.
An infrared optical micrograph of the nanodisk arrays is shown in Fig. 5(b). An illumination source of λ = 1.2 µm was chosen as it is sufficiently far from the inter-band transition wavelength. The infrared image of the arrays is markedly different from the visible light image. Only disk diameters between 180 nm and 240 nm appear dark (black), whereas the smaller disks appear bright (grey). The appearance of the dark arrays is due to absorption by the disks at their fundamental resonance, as can be observed from Fig. 5(a), which reduces the reflectance at the source wavelength. Similarly, the bright pixels are due to strong reflectance at the source wavelength. The pixels with disk diameters between 110 nm and 170 nm and between 250 nm and 280 nm appear dark grey, as they partially absorb light at the source wavelength. Some inter-disk coupling of plasmon modes is also evident as a general darkening of the arrays in disks patterned closer than a separation of ~50 nm and with diameters of 140 nm to 190 nm.
The CIE 1931 xy chromaticity coordinates are plotted in Fig. 6 to explicitly show the range of colors obtained from the fabricated Al nanodisk arrays. The coordinates were calculated from the reflectance data according to standard equations . The colors encompass most hues, including orange, yellow, cyan, blue, purple and magenta; unsaturated green was also obtained. While the color gamut achieved only encompasses ~25% of the area of the sRGB triangle, it is comparable to that previously described for Al-based gap plasmonic nanostructures  and significantly broader than that for gold nanostructures .
4. Micro-tag for authentication
The disparity between the visible and infrared responses of the nanodisks allows for the design of a security tag that hides a covert infrared image within a colorful visible image in the following manner. Firstly, two pairs of disk diameters and gaps are selected from the color and IR palettes of Fig. 5. One pair needs to exhibit the same color but different IR reflectance, while another needs to have different colors but the same IR reflectance. As the visible colors of the disk arrays appear to repeat when the disk diameter is increased, while the infrared responses do not, pairs of arrays that have similar colors but different IR reflectance intensities should be carefully selected. As an additional requirement, the arrays should have similar color lightness (or brightness), quantified using the CIELAB model, so that the security tag will be difficult to spot. For instance, two yellow arrays can be chosen – one with small disk diameters around 60 nm and a second with large disk diameters around 200 nm (labelled 1 and 2 in Fig. 5 respectively). A magenta color of similar lightness to yellow but different hue is chosen for the second pair of arrays. The magenta arrays, with disk diameters of 80 nm and 230 nm respectively, are also indicated in Fig. 5. Under illumination of infrared light of 1200 nm wavelength, the arrays with large disk diameters will appear black while the arrays with small disk diameters will appear grey, as expected from the reflectance data in Fig. 5(c). Thus, there are four types of disks used, each with one unique combination of IR appearance (bright/dark) and visible color appearance (yellow/magenta). The exact disk diameters, gaps and periods are listed in Table 1.
Secondly, the data for two binary images are input into a computer algorithm using MATLAB (MathWorks Inc.) to generate the layout file for input to the EBL tool. The algorithm allocates the correct disk diameter and gap for each pixel in the image field based on the pair of binary values for the two images, and specifies an integral number of disks to fill up a square pixel. The size of the pixel is a common multiple of the periods of the disks to avoid gaps between pixels. For potential security applications, a bar code was used as the infrared image, while a quick response (QR) code was used as the visible color image (in yellow/magenta).
The brightfield visible color and infrared micrographs of the physical sample made are shown in Fig. 7(a) and Fig. 7(b) respectively, and show good reproductions of the original QR code and barcode. The QR code appears as a magenta pattern on a yellow background, whereas the barcode appears as black lines on a grey background. For better legibility of the codes, the image files (JPEG) of the visible color micrographs taken of the fabricated sample were enhanced in PowerPoint (Microsoft Corp.) by increasing the contrast and thresholding to produce a black-and-white image. The following steps are for the example in Fig. 5(a). First, the contrast of the image is increased by 80% and its saturation increased to 200%. Next, the image was recolored to obtain a grayscale image and saved as a new image. Finally, the new image was recolored using the “Black and White: 50%” setting to obtain a black-and-white image. This set the threshold for designating white pixels to the 50th percentile of the luminance of all pixels in the image.
The low crosstalk between the visible color and infrared images prevented the infrared image from being easily deciphered from the visible color micrograph. A darkfield visible light micrograph of the sample is also shown in Fig. 7(c), where the barcode appears as blue lines on a black background. The large disks that form the lines diffract blue light due to their correspondingly large periodicity of 300 nm. As their periodicities are identical, they generate the same hue, saturation and luminance. On the other hand, the small disks do not diffract light due to their correspondingly small periodicities of 100 nm and 200 nm. The top view SEM image of a section of the sample is shown in Fig. 7(d), and the four types of disk diameters used can be seen.
The observability of an infrared image, encoded in an otherwise ordinary-looking colorful pattern, has several potential applications for security printing, law enforcement, military and industry. For example, valuable artwork can be tagged with the pattern for authentication before any transactions or loans to other institutions. A portable universal serial bus (USB) infrared microscope with a filter can then be used to detect the covert infrared image. Current security prints and labels are either easily counterfeited, e.g. holograms can be duplicated optically or mechanically , or difficult to embed into existing art pieces, e.g. security threads have to be inserted during the production of the canvas or paper . Radio-frequency identification (RFID) tags can also be easily hacked . However, our micro-tag has the advantages of containing two sets of visual information – (1) a visible color image, (2) a covert infrared image – as well as being small enough to be easily embedded into an existing art piece, and requiring sophisticated nanofabrication tools to replicate.
We demonstrate a new type of micro-tag using metal-insulator-metal nanostructures, so that a visible color image and an infrared image are patterned onto the same area. With the native oxide of Al as the insulator, an insulator thickness of ~7 nm was obtained as measured from TEM cross-sectional images. This oxide thickness suitably allows for strong absorptances spanning the visible to infrared regimes. Due to the tight confinement of modes in this ultra-thin oxide, the disk resonators exhibited little coupling between disks and retained a small footprint < 300 nm diameter while supporting resonances up to 1.8 µm wavelength. A design process is developed for encoding two images into a tag with an array of disks using four types of pixels, each with a different disk diameter and separation. The two images are observable using visible light and a narrow-wavelength infrared source respectively, with low crosstalk between the two. Such a tag could be used for authentication, anti-counterfeiting and cryptography. Though UV inspection was not explored here, it remains an additional channel for encryption for artwork not susceptible to damage by UV-irradiation.
National Research Foundation (NRF) Competitive Research Programme (CRP) (15-2015-03); Agency for Science, Technology and Research (A*STAR) Science and Engineering Research Council (SERC) Pharos Programme (1527300025); A*STAR-Joint Council Office (A*STAR-JCO) (1437C00135).
We thank Tao Wang, Eleen Koay, Nan Zhang, Xiao Ming Goh, Hao Wang and Yejing Liu for fruitful discussions.
The authors declare that there are no conflicts of interest related to this article.
1. A. Kristensen, J. K. W. Yang, S. I. Bozhevolnyi, S. Link, P. Nordlander, N. J. Halas, and N. A. Mortensen, “Plasmonic colour generation,” Nat. Rev. Mater. 2(1), 16088 (2017). [CrossRef]
4. S. J. Tan, L. Zhang, D. Zhu, X. M. Goh, Y. M. Wang, K. Kumar, C.-W. Qiu, and J. K. W. Yang, “Plasmonic color palettes for photorealistic printing with aluminum nanostructures,” Nano Lett. 14(7), 4023–4029 (2014). [CrossRef] [PubMed]
7. Z. Dong, J. Ho, Y. F. Yu, Y. H. Fu, R. Paniagua-Dominguez, S. Wang, A. I. Kuznetsov, and J. K. W. Yang, “Printing beyond sRGB color gamut by mimicking silicon nanostructures in free-space,” Nano Lett. 17(12), 7620–7628 (2017). [CrossRef] [PubMed]
8. X. M. Goh, R. J. H. Ng, S. Wang, S. J. Tan, and J. K. W. Yang, “Comparative Study of Plasmonic colors from all-metal structures of posts and pits,” ACS Photonics 3(6), 1000–1009 (2016). [CrossRef]
9. R. J. H. Ng, X. M. Goh, and J. K. W. Yang, “All-metal nanostructured substrates as subtractive color reflectors with near-perfect absorptance,” Opt. Express 23(25), 32597–32605 (2015). [CrossRef] [PubMed]
10. X. M. Goh, Y. Zheng, S. J. Tan, L. Zhang, K. Kumar, C.-W. Qiu, and J. K. W. Yang, “Three-dimensional plasmonic stereoscopic prints in full colour,” Nat. Commun. 5(1), 5361 (2014). [CrossRef] [PubMed]
13. M. Yousaf and M. Lazzouni, “Formulation of an invisible infrared printing ink,” Dyes Pigm. 27(4), 297–303 (1995). [CrossRef]
14. F. Ding, R. Deshpande, and S. I. Bozhevolnyi, “Bifunctional gap-plasmon metasurfaces for visible light: polarization-controlled unidirectional surface plasmon excitation and beam steering at normal incidence,” Light Sci. Appl. 7(4), 17178 (2018). [CrossRef]
17. T. Maurer, P.-M. Adam, and G. Lévêque, “Coupling between plasmonic films and nanostructures: from basics to applications,” Nanophotonics 4(3), 363–382 (2015). [CrossRef]
20. F. Benz, B. de Nijs, C. Tserkezis, R. Chikkaraddy, D. O. Sigle, L. Pukenas, S. D. Evans, J. Aizpurua, and J. J. Baumberg, “Generalized circuit model for coupled plasmonic systems,” Opt. Express 23(26), 33255–33269 (2015). [CrossRef] [PubMed]
24. J. Olson, A. Manjavacas, L. Liu, W. S. Chang, B. Foerster, N. S. King, M. W. Knight, P. Nordlander, N. J. Halas, and S. Link, “Vivid, full-color aluminum plasmonic pixels,” Proc. Natl. Acad. Sci. U.S.A. 111(40), 14348–14353 (2014). [CrossRef] [PubMed]
25. S. Tian, O. Neumann, M. J. McClain, X. Yang, L. Zhou, C. Zhang, P. Nordlander, and N. J. Halas, “Aluminum nanocrystals: a sustainable substrate for quantitative SERS-based DNA detection,” Nano Lett. 17(8), 5071–5077 (2017). [CrossRef] [PubMed]
27. S. Ayas, A. E. Topal, A. Cupallari, H. Güner, G. Bakan, and A. Dana, “Exploiting native Al2O3 for multispectral aluminum plasmonics,” ACS Photonics 1(12), 1313–1321 (2014). [CrossRef]
28. H. Im, N. J. Wittenberg, N. C. Lindquist, and S.-H. Oh, “Atomic layer deposition (ALD): A versatile technique for plasmonics and nanobiotechnology,” J. Mater. Res. 27(4), 663–671 (2012). [CrossRef] [PubMed]
30. J. Ho, Y. H. Fu, Z. Dong, R. Paniagua-Dominguez, E. H. H. Koay, Y. F. Yu, V. Valuckas, A. I. Kuznetsov, and J. K. W. Yang, “Highly directive hybrid metal–dielectric Yagi-Uda nanoantennas,” ACS Nano 12(8), 8616–8624 (2018). [CrossRef] [PubMed]
31. E. D. Palik, Handbook of Optical Constants of Solids (Academic, 1998).
32. H. Ehrenreich, H. R. Philipp, and B. Segall, “Optical Properties of Aluminum,” Phys. Rev. 132(5), 1918–1928 (1963). [CrossRef]
33. S. D. Rezaei, J. Ho, R. J. H. Ng, S. Ramakrishna, and J. K. W. Yang, “On the correlation of absorption cross-section with plasmonic color generation,” Opt. Express 25(22), 27652–27664 (2017). [CrossRef] [PubMed]
34. T. Søndergaard, J. Jung, S. I. Bozhevolnyi, and G. D. Valle, “Theoretical analysis of gold nano-strip gap plasmon resonators,” New J. Phys. 10(10), 105008 (2008). [CrossRef]
35. S. P. McGrew, “Hologram counterfeiting: problems and solutions,” Proc. SPIE 1210, 66–76 (1990). [CrossRef]
36. T. T. Crane, “Security paper for currency and banknotes,” U.S. patent 4,652,015 (March 24, 1987).
37. V. Chawla and D. S. Ha, “An overview of passive RFID,” IEEE Commun. Mag. 45(9), 11–17 (2007). [CrossRef]