1. Introduction

Studies into controllable light-matter interactions at visible wavelengths using plasmonic nanostructures to produce structural coloration has attracted significant attention due to their outstanding ability to selectively absorb, reflect and transmit light [1,2]. This effect is enabled by the excitation of localized surface plasmons resonances exhibited by metallic nanostructures and their associated optical response that can be tuned by changing the geometry or material composition of the structure, allowing efficient color tunability [3]. Plasmonic coloration offers the advantages of good reliability, environmental sustainability, and durability, compared to use of conventional dyes and pigments [2–4]. Depending on the geometry of the nanostructures, however, plasmonic color can be either dependent or independent of the polarization state of the incident light.

Nanostructures with broken symmetry geometries such as rectangular rods [5,6], ellipses [7], shallow nanogratings [8] and three-dimensional chiral structure [9,10] have demonstrated excellent color switching capacity through the change in the direction of linearly polarized light or the handedness of circularly polarized light. Although polarization dependent plasmonic colors are sought-after in various applications such as high-density data storage [10–12], optical security tags [8,9,12], sensors [8,13,14] and color filters [15–17], they however, suffers from ‘cross-talk’ due to polarization multiplexing [7]. This can lead to undesired color mixing or color desaturation, causing the vividness (i.e the saturation and brightness of the color) to deteriorate and the color less apparent particularly when illuminated under unpolarized ambient light. Plasmonic color devices comprising symmetric geometries, on the other hand, exhibit minimal color change with polarization states of light and are, therefore, advantageous for preserving the hue and vividness of the generated color. This is particularly important in architecture [18], product packaging [19], automotive paint [20,21] and other applications where the surface will be viewed in sunlight or artificial lighting. Previous studies have resorted to simple disk, square, pentagonal and hexagonal shaped nanostructures to produce polarization independent colors [22–27]. These designs incorporating disk and polygon-shaped structures, however, generally rely on the changing the size of the periodic lattice for color tuning although sub-pixel mixing involving multiple disks in the unit cell has been demonstrated to produce a broad color gamut [27]. This means that simple designs have tended to produce only a limited range of hue when the periodicity is fixed. Increasing the periodicity of the structure expands the footprint of the device and can introduce an angle-dependence of the resulting colors via diffraction.

In this work, we demonstrate the use of a symmetric cross-shaped nanoantenna-hole hybrid structure (Fig. 1(a)), to produce a vivid, polarization insensitive surface coloration both in reflection and transmission. The devices also show sufficing tolerance to the change in angle of incidence (in air) up to 30$^{\circ }$. The spectral positions of reflectance and transmission minima or maxima are tuned by adjusting only the armlength of the cross, while keeping the other geometric parameters fixed, thus allowing the production of various colors. In this way, the print resolution can be kept constant. Compared to other designs, however, the cross-shaped structure, in common with the rod-like plasmonic pixel, exhibits a stronger resonance and, hence, more vivid coloration [5]. Furthermore, the structure has narrower resonances compared to other designs which allow fine tuning of the color. Additionally, we also compare the reponse of devices incorporating different metals, namely silver (Ag), aluminium (Al) and gold (Au) integrated into identical nanoimprinted templates. Colors produced by all devices are found to have a high tolerance to the polarization state of the incident light. With the capacity for large area fabrication using high-throughput NIL [28], the plasmonic color devices are potentially highly attractive for diverse applications, ranging from high-definition displays, high-resolution color printing, architecture, as well as product-branding applications.

 

Fig. 1. (a) Schematic of the symmetric cross-shaped metallic nanoantenna vertically displaced above of its complementary perforated film. The incident light (polarized or unpolarized) is illuminated from the nanoantenna side of the sample. (b-e) Schematic shows the fabrication process of the polarization independent plasmonic color utilizing the cross-shaped structures. (f) The representative SEM images shows the morphology of 200 nm long cross-shaped nanoprotrusions on the mold and nanocavities on the sample after the imprinting and metallisation process. The cross-sectional image shown in (g) reveal the vertical gap size of the structure i.e $g=65\pm 1$ nm. All scale bars refer to 500 nm.

Download Full Size | PDF

2. Experiment details

2.1 Design and fabrication

A schematic showing the device under investigation is shown in Fig. 1(a). It consists of periodic arrangements of a unit cell composed of a cross-shaped metallic (Ag, Al or Au) nanoantenna and the complementary perforated film of the same material. With the exception of the shape of the unit cell, the geometry is similar to the plasmonic pixel [5,6]. The cross has a symmetric geometry with both arms having the same length, $L$ and width, $W$, to minimize any dependence of the optical response of the structure on the direction of polarization. Various colors are obtained by tuning the arm lengths, $L$ while their width is kept fixed at $W=40$ nm. The nanoantenna is vertically displaced from the complementary aperture on the film assuming a nominal fixed distance of $g=65$ nm to permit strong coupling between the plasmonic modes exhibited by both the nanoantenna and the perforated film allowing the metasurfaces to behave as a wavelength-selective plasmonic absorber [5,19]. The thickness of the metal film is fixed at $t=50$ nm for all samples while the unit cell is arranged in a square array with a periodicity of $P=300$ nm.

To demonstrate that the fabrication of the device has the capacity to be scaled for mass production, we utilize techniques compatible with large-scale fabrication, including UV-assisted nanoimprint lithography (UV-NIL) to first create the pattern template, followed by electron beam evaporation to deposit a layer of metal film on the patterns as shown in Fig. 1(b-e). For NIL, a Si mold comprising arrays of cross-shaped nanoprotrusions was used in the nanoimprinting process to replicate the pattern on a UV-curable resist film (NXR-2030, Nanonex) spin coated onto a borosilicate glass substrate (Schott) (Fig. 1(b)). The process used to fabricate the Si mold is described in Supplement 1. During NIL the mold is pressed into the resist at a pressure of 50 psi for 3 minutes before being exposed to UV light for 45 seconds to ensure cross-linking of the resist (Fig. 1(c)). The mold-substrate stack was then manually demolded using a razor blade thus leaving a template consisting of an array of cross-shaped nanocavities as shown in Fig. 1(d). A 50 nm layer of metal was subsequently evaporated onto the replicated pattern using electron beam evaporation (Fig. 1(e)). The final sample, therefore, has the structure partially embedded in the polymer resist film (on the nanoantenna side) that has similar refractive index to the glass substrate ($n$=1.45). The SEM image in Fig. 1(f) shows the morphology of a representative sample (Ag) depicting the nanoimprinted cross-shaped nanocavity array after the metal evaporation process which are in very good agreement with their transverse counterparts on the Si mold, depicting excellent fidelity of the pattern replication. Each of the nanoimprinted pixels covers an area of $0.5\times 0.5$ mm$^2$. An image of the cross-section in Fig. 1(g) permits determination of the vertical distance between the top and bottom layer metal (nanoantenna and the perforated film) which is measured to be $g=65\pm 1$ nm.

2.2 Optical characterization

The reflectance and transmission spectra of the fabricated devices were measured using a custom optical setup. Light from a broadband halogen lamp (HL-2000-FHSA, Ocean Optics) illuminated the sample at normal incidence. Either linearly polarized or unpolarized light passes through an objective lens (NA$=0.4$, 20$\times$ UPlan, Nikon) and is focused onto the sample. The reflected light from the surface was collected with the same objective lens while reflected light from an unpatterned region of the sample (metal thin film) was acquired as a reference. Another objective lens (NA$=0.55$, 50$\times$ CFPlan, Nikon) was placed behind the sample to collect the transmitted light while the light transmitted through a blank glass slide was acquired as reference. Bright-field optical images of the patterned surfaces were obtained using a Nikon N1 camera attached to an optical microscope (Olympus BX60) with a halogen lamp as a light source. White balance of obtained image is achieved by using a white balance card as a reference. A linear polarizer (GTH5M, Thorlabs) located in the path of the light source was used to determine any polarization sensitivity exhibited by the device. Three-dimensional simulations of these structures were performed using the Finite Element Method (FEM) where the details is described in Supplement 1.

3. Results and discussion

3.1 Preserving hue and vividness of plasmonic color under ambient light

The geometry of the plasmonic structure determines the plasmonic modes and the resulting spectrum. To study the influence of the geometry on the sensitivity of the resulting colors to the state of polarization, the measured reflectance spectra from a device consisting of arrays incorporating 120 nm long rectangular-shaped Ag nanoantennas or a symmetric cross Ag nanoantennas of the same length and width are shown in Fig. 2(a) and (b) respectively. The structures were illuminated with light linearly polarized in either the $x-$ or $y-$direction at normal incidence using the optical setup described above, while the average of the reflectance obtained from both polarizations gives the equivalent spectra for unpolarized illumination. The discussion here is restricted to only the reflectance exhibited by the devices and the transmission will be discussed later.

 

Fig. 2. The measured reflectance plot of the plasmonic color comprising (a) the asymmetric rectangular-shape nanostructure and (b) the symmetric cross-shaped nanostructure under linearly polarized and unpolarized light. Inset shows the bright-field optical images of the reflected colors corresponds to the spectral response.

Download Full Size | PDF

As expected, the device containing rectangular nanostructures exhibits two resonance wavelengths i.e at $\lambda =610$ nm and $\lambda =480$ nm corresponding to directions of polarization parallel to either the long or short axes of the rectangular elements as manifested by the minima in the reflectance spectra (solid and dashed red lines) shown in Fig. 2(a). The optical images in insets of Fig. 2(a) exemplify a switchable output color from purple to orange, produced by single-length rectangular structures by means of rotating the polarizer positioned in the path of the incident light. Under unpolarized light, therefore, the structure exhibits broadband reflectance with a flatter spectral profile (solid blue line) resembling a desaturated color as depicted in the inset of Fig. 2(a). This shows that polarization dependent plasmonic structures exhibit ‘cross-talk’ [7] when viewed under unpolarized ambient light where there is partial suppression of coloration specified for one polarization when viewed in the orthogonal polarization or unpolarized light leading to potentially undesirable color mixing. Consequently, the resulting color become less saturated or the surface could even change hue when viewed under ambient lighting conditions.

In contrast to the device consisting of rectangular elements, the cross-shaped nanostructure device yields almost identical spectral characteristics (Fig. 2(b)) for both polarization states where the reflectance minima remain at 550 nm while the reflectance intensity is almost unchanged. This results in the production of a purple color that is almost unperturbed despite the modification in polarization state of light (insets of Fig. 2(b)), suggesting the robustness of the output color under ambient light. The strong tolerance to the direction of polarization of light exhibited by this structure can be attributed to the symmetric geometry of the cross where the same dipole modes are generated by orthogonal polarization directions. Furthermore, the simulated reflectance spectra presented in Fig. S1 in Supplement 1 show no significant shift in resonance wavelengths with an increase in the angle of incidence up to 30$^{\circ }$, suggesting that the devices also exhibit only a weak color sensitivity to viewing angle.

3.2 Cross-shaped polarization-independent plasmonic color based on various metals

To demonstrate production of a wide range of colors with fixed saturation and brightness, the armlength, $L$ of the cross structure was systematically varied from 80 to 200 nm, while keeping all other spatial parameters fixed. Measurements of the reflection spectra and bright-field optical images were obtained to investigate the optical responses of the fabricated plasmonic pixels. Figure 3 shows the measured reflectance spectra for the Ag, Al and Au-based cross-shaped pixel arrays respectively when illuminated with linearly polarized (dashed-line) and unpolarized (solid line) light at normal incidence. Figure 4(a-c) shows the bright field optical images of the output colors corresponding to the spectra shown in Fig. 3. To investigate the effect of polarization state of light on the quality of the resulting color and to validate the optimum dimensions to produce distinct coloration under ambient light, the reflectance spectra corresponding to each armlength are converted to the hue, saturation and brightness values ($h$, $s$ and $v$) to quantify the colors. The hsv values are then plotted in the 2-dimensional hue-saturation (HS) polar plot (Fig. 4(d-f)), which represents the position of the color according to its hue (azimuthal coordinate) and saturation (radial coordinate), and also the brightness ($v$) plot (Fig. 4(g-i)).

 

Fig. 3. Measured reflectance spectra correspond to the (a) Ag, (b) Al and (c) Au devices comprising the cross-shaped structures of increasing armlength, $L$ from 80 nm to 200 nm, with fixed periodicity, $P=300$ nm, under $x-$ and $y-$polarized light (black and blue dashed lines, respectively). Average of both reflectance represents unpolarized light (solid lines).

Download Full Size | PDF

 

Fig. 4. Properties of reflected colors produced by the Ag, Al and Au devices consisting the symmetric cross-shaped nanoantenna-hole structure. (a-c) Optical images show color swatches produced by the Ag, Al and Au devices respectively as results of the increased armlengths under different light polarization state. Each color swatch has the total area of $0.5 \times 0.5$ mm$^2$. (d-f) The hue-saturation (HS) polar plots and (g-i) brightness ($v$) plots associated with the output colors in (a-c) exhibited by the Ag, Al and Au devices respectively, under polarized and unpolarized light. The arrow shows direction of increasing armlength of the cross.

Download Full Size | PDF

In general, Fig. 3 indicates that the reflectance spectra generated by all devices show little sensitivity to the state of polarization of the incident light. Similarly, the hue, saturation and brightness values remain almost the same under both polarized and unpolarized light as shown in Fig. 4(d-i), producing colors with similar shades and vividness (Fig. 4(a-c)). The simulated reflectance spectra presented in Fig. S1(c) and (d) in Supplement 1 also shows that the devices have only a weak response to changes in the angle of incidence up to 30$^{\circ }$. In the case where the armlength of the cross structure is increased, the reflectance spectra in Fig. 3 show that the resonance wavelengths for all devices are red-shifted due to the elongation of the armlength consistent with that seen in the case of similar, albeit polarization sensitive, nanorod devices [5]. This is manifested by the translation of the reflectance minima to longer wavelengths. The simulated spectra shown in Fig. S2 in Supplement 1 are in a good agreement with the spectra obtained experimentally. The increased in the armlength of the cross, therefore, results in the production of various colors as depicted in the color swatch shown in Fig. 4(a-c). On the other hand, the increase in the size of periodic lattice lead to the appearance of an additional reflectance minima at longer wavelength due to diffraction in addition to the red-shifting of the resonance wavelengths, as shown by the simulated reflectance spectra in Fig. S1 in Supplement 1. Therefore, for this study, the size of the periodic lattice, $P$ is kept fixed at 300 nm.

Figure 3(a), showing the measured reflectance spectra for the Ag ‘cross’ pixel array, indicates the red-shifting of the resonance wavelength from $\lambda =$ 510 nm to beyond 800 nm in the near infrared (NIR) due to the elongation of the armlength of the cross structures. As the length reaches 200 nm, the spectral profile becomes flatter and results in less saturated color. Additionally, there is also an additional minimum in reflectance at shorter wavelengths ($\lambda \sim 450$ mn) corresponding to the excitation of a high-order quadrupole mode (Supplement 1, Fig. S3). The optical images shown in Fig. 4(a) shows that the Ag device produces colors that varies from bright pink to blue to bright green, associated with an increase in armlength of the cross independent of the polarization state of the light. The HS plot in Fig. 4(d) shows that the Ag device has output hue values that revolve in a clockwise direction from $h=284^{\circ }$ to 111$^{\circ }$, indicating a wide coverage of colors spanning near magenta to green. The saturation ($s$) of the colors also varies with the lengths of the arm consistent with the variation in linewidth of the spectra. The reflected colors have a brightness value ($v$) of greater than 0.6 which can be attributed to the efficiency of the structures with reflectance up to 80$\%$ (Supplement 1, Fig. S4(a)).

For the Al device, the elongation of the armlength leads to a change in color due to a shift in the resonance wavelength from approximately 460 nm to around 800 nm in the reflectance spectra as shown in Fig. 3(b). The HS plot in Fig. 4(e) shows that under unpolarized light, the hue of the Al device revolves in a clockwise direction from $h=26^{\circ }$ to $122^{\circ }$ (black line), thus indicating that the Al device is able to produce a wider range of colors covering yellow, magenta and cyan, primary colors that cannot be fully achieved using Ag (see Fig. 4(a)). Compared to Ag devices the Al devices, however, exhibit shallower reflectance minima (Supplement 1, Fig. S4(a)) leading to less saturated colors compared to those generated by Ag devices as seen in Fig. 4(e).

The Au devices exhibit the fundamental resonance at longer wavelengths ($\lambda >550$ nm) compared to Al and Ag (Fig. 3(c)). The resonance wavelength red-shifts towards the NIR region as the length of the arm increases. Additionally, a broad reflectance minima appears at shorter wavelengths ($\lambda <550$ nm) due to the interband transition in Au ($\sim 2.4$ eV) [29]. The HS plot in Fig. 4(g) shows that the hue produced by the Au device under unpolarized light changes significantly from $h=248^{\circ }$ (bright pink) to $197^{\circ }$ (orange), as the armlength increases from 80 nm to 140 nm. As the armlength further increases to 200 nm the hue, however, shows an almost constant hue value at $h=357^{\circ }$, covering only the red region of the plot, exemplifying the limited color coverage obtainable with the Au devices. The increase in the size of the armlength is accompanied by a broadening of the reflectance spectra resulting in less saturated colors. Overall, the results shown in Fig. 3 and 4 suggest that although the Al devices have the widest color coverage of the three metals, they produces less saturated colors compared to an otherwise identical Ag device.

Transmission through the devices was also studied. Figure 5(a-c) presents the transmission spectra exhibited by the Ag, Al and Au devices respectively. Figure 5 shows that all devices yield almost identical spectral profiles for both unpolarized (solid lines) and linearly polarized light (dashed line) depicting high tolerance of the devices to the polarization state of the light and unperturbed color properties. Figure 6(d-f) further exemplifies this behavior whereby the position of the hue and saturation of each of the transmitted colors in the HS polar plot are almost unchanged despite the polarization state of light. Similarly, the brightness remains constant regardless of the polarization state of the incident light and changes in the size of the armlength.

 

Fig. 5. Measured transmission spectra correspond to the (a) Ag, (b) Al and (c) Au devices comprising the cross-shaped structures of increasing armlength, $L$ from 80 nm to 200 nm, with fixed periodicity, $P=300$ nm, under $x-$ and $y-$polarized light (black and blue dashed lines, respectively). Average of both reflectance represents unpolarized light (solid lines).

Download Full Size | PDF

 

Fig. 6. Properties of transmitted colors produced by the Ag, Al and Au devices consisting the symmetric cross-shaped nanoantenna-hole structure. (a-c) Color swatches produced by the Ag, Al and Au devices respectively as results of the increased armlengths under different light polarization state. Each color swatch has the total area of $0.5 \times 0.5$ mm$^2$. (d-f) The hue-saturation (HS) polar plot and (g-i) brightness ($v$) plots associated with the output colors in (a-c) exhibited by the Ag, Al and Au devices respectively, under polarized and unpolarized light.

Download Full Size | PDF

The plots shown in Fig. 5(a) indicate that the minima in the transmission spectra through the Ag device are red-shifted from $\lambda =590$ nm to 650 nm as the armlength of the cross structure increases from $L=80$ nm to 140 nm. This results in the hue value changing from $h=225^{\circ }$ to $193^{\circ }$ (clockwise) (Fig. 6(d)) thus producing various colors varying from dark purple to blue as shown in the color swatch in Fig. 6(a). Further increasing the size of the cross structure to $L=200$ nm pushes the position of the minima closer to the NIR region which results in minimal change in hue accompanied with a gradual desaturation of the color. The plot in Fig. 5(a) also shows the appearance of an additional minimum in transmission at $\lambda =450$ nm for cross structures with armlengths of $L>120$ nm, similar to those observed in reflection which can be attributed to the excitation of quadrupole mode exhibited by the structure. In comparison, the simulated transmission spectra (Supplement 1, Fig. S5(a)) shows reasonable agreement with the experiment.

On the other hand, Fig. 6(b) demonstrate that the Al device produces assorted colors covering yellow, magenta and near-cyan due to the red-shift in transmission minima from $\lambda =466$ nm to 612 nm as the armlength increases (Fig. 5(b)). The simulated spectra show considerable consistency and are in a good agreement with the measured transmission (Supplement 1, Fig. S5(b)). Additionally, elongation of the armlength of the crosses also leads to an increase in the transmission efficiency of the device. These conditions result in a change in hue value from $h=31^{\circ }$ to $206^{\circ }$ (anticlockwise) and an increase in saturation of the colors as presented in the HS plot in Fig. 6(e).

In contrast to the Ag and Al devices, the Au device exhibits a maximum in transmission at $\lambda =500$ nm (see Fig. 5(c)). In contrast to devices fabricated with Ag and Al, the position of the peak shows no shift despite the increase in the size of the armlength of the cross structure. This broadband peak spanning from $\lambda =450$ to 550 nm arises from the interband transition in Au. In addition, there is a minimum in transmission corresponding to the excitation of fundamental dipole mode adjacent to the maximum. This minimum red-shifts from $\lambda =600$ nm to 700 nm with the increase in the size of the cross structure. Despite this, the size-independent peak in the transmission at $\lambda =500$ nm dominates the color and leads to minimal change in hue value where the variation of hue form small revolution from $h=172^{\circ }$ to $h=160^{\circ }$ (clockwise) in the HS plot in Fig. 6(h). In terms of color production in transmission, the use of Al offers more hue coverage with high and stable saturation compared to Ag meanwhile Au produces a limited variation in hue.

Compared to the reflected colors, however, the transmitted colors produced by all devices have relatively low brightness values $v<0.4$ (see Fig. 6(g-i)) which can attributed to their low transmission efficiency. Among all devices, the Ag device shows the highest transmission efficiencies up to 12$\%$ (Supplement 1, Fig. S4(b)) and hence has higher brightness values while others have transmissions below 10$\%$. This deficiency, however, could be compensated by increasing the size of the gap between the nanoantenna and its complementary hole. The simulated spectra shown in Fig. S6 in Supplement 1 suggests that increasing the gap size from $g=65$ nm to $g=145$ nm could increase the transmission efficiency of the Ag device by a factor of two. Overall, Ag devices demonstrate good performance with 80$\%$ reflection and 12$\%$ transmission efficiencies, producing saturated and brighter colors. The high reflection efficiency shown by the Ag devices is consistent with other studies [25,30]. In the case where a higher transmission is required, however, the vertical gap, $g$, can be increased as shown by the simulated reflectance spectra (Supplement 1, Fig. S6). Compared to Ag and Au, on the other hand, the higher plasma frequency of Al permits access to the entire visible spectrum enabling production of a wider color gamut [2,5]. Both Ag and Au, however, are less suitable for mass production due to their relatively high price. Aluminium, on the other hand, is a cost-effective alternative due to its abundance in nature and is also environmentally stable especially for ambient use due to the formation of the native oxide layer. Combined with inexpensive, high-throughput, nanoimprinting fabrication approaches, the use of aluminium is, therefore, highly attractive for large scale production of the plasmonic color devices. To improve the environmental stability and usability of the device particularly in ambient condition, a thick layer of silica can be coated on the metal surface as a protection layer from oxidation process and fingerprints. This, however, will affects the resulting color and would need to be accommodated in the design.

4. Conclusion

In summary, we have demonstrated the production of vivid, polarization-robust plasmonic coloration using different metals in otherwise identical ‘plasmonic pixel’ devices when viewed under ambient light. This was achieved by utilizing symmetric cross-shaped nanoantenna-hole structures fabricated using NIL. All devices show little change in the spectral and color response in both reflection and transmission despite the change in polarization state of incident light and angle of incidence, suggesting an almost constant vividness of the generated color particularly in ambient lighting condition. Wide color gamuts were generated by simply changing the size of the armlength of the cross (while keeping the size of lattice fixed) and through the use of various metals. The Ag devices demonstrate the highest reflection and transmission efficiencies of the three metals studied, although the Al devices generate a wider color gamut. With the capability of large area fabrication, the proposed plasmonic coloration devices can be potentially applied in sensing technology, high-resolution color printing and product-branding applications.