1. Introduction

Engineered plasmonic metamaterials have unique optical characteristics depending on the shapes of their sub-wavelength structures; thus, it is possible to control the refractive index of the metamaterial layer [1]. These metamaterial structures can be strongly coupled with an incident electromagnetic wave, yielding unique optical responses such as anomalous transmission/reflection/absorption in an arbitrary wavelength band [2–4]. Recently, flexible metamaterials using a thin polymer film substrate have attracted attention [5–12]. As flexible metamaterials can be bent, wrapped, folded, and laminated, they are suitable for application in flexible devices such as flexible displays [13], smart eyewear devices [14], wearable devices [15,16], and cloaking devices [17,18]. Metamaterials formed on a polymer substrate are also suitable for application in fluid chips, including chemical and biological sensors [6,19]. In addition, lamination of thin polymer films can contribute to the realization of three-dimensional metamaterials and multiband metamaterials. Therefore, as indicated above, flexible metamaterials can be applied to produce attractive devices that are difficult to realize using solid metamaterials.

Conventional flexible optical metamaterials have primarily been formed using electron-beam lithography, vacuum deposition of metals, and transfer processes [6–9]. In particular, the nanoimprint and nano transfer method is a method of depositing a metal film on a stamp mold having a convex structure formed by nanoimprinting. The stamp mold is pressed against the target substrate; hence, a metal nanostructure can be formed on a flexible substrate. However, the metal film must be deposited on the stamp mold using vacuum deposition apparatus in each instance, which generates a corresponding manufacturing cost. Further, the pattern area is increased as a large vacuum deposition apparatus is required; thus, the facility cost is also a concern. On the other hand, flexible metamaterials operating in the microwave or terahertz regions have relatively large pattern sizes, which can be easily formed via a printing technique such as ink-jet printing [20,21], electrohydrodynamic jet printing [22], or laser printing [23]. Printing has advantages of high productivity and large patterning areas. In recent years, printing technique development has been driven by dramatic developments in the field of printed electronics [24], in which electronics are manufactured via a printing process. However, printed optical metamaterials have not yet been realized because the resolutions of conventional printing methods are insufficient, as optical metamaterials require sub-micron structures. Self-organization and array formation of nanoparticles constitutes another simple metamaterial fabrication method, and research and development are ongoing [25–27]. However, the degrees of freedom of the shapes and periods of the metamaterial structures fabricated in this way remain insufficient.

Optical metamaterial formation via a simple printing technique is advantageous in terms of productivity and large-area output. Further, printing has good compatibility with the polymer process. Therefore, printing will be an important component in the future relations of practical flexible metamaterial application. Previously, the present authors developed a high-resolution embedded printing technique using nanoimprint and squeegeeing methods [28–30], achieving a resolution that reached submicron scale. By selecting a suitable nanoink and controlling the surface function of a film substrate, further miniaturization and adaptation to complex shapes are possible.

In this study, we apply our previously developed embedding printing method and demonstrate optical metamaterials comprised of commercially available ink containing silver nanoparticles. We evaluate the optical characteristics of the printed optical metamaterials and confirm that they are derived from the metamaterial structures. In the terahertz region, the metamaterial characteristics are fully demonstrated in a structure made of ink; however, it is unclear whether they are also apparent in the optical region sensitive to the material properties. In order to distinguish between the optical characteristics derived from the nanoparticles and from the metamaterial structures, three types of metamaterial components, i.e., dipole resonators, split-ring resonators (SRRs), and electromagnetically induced transparency (EIT) metamaterials, are fabricated. For the dipole resonators, when the major-axis length is changed, the resonant wavelength changes [31]. For the SRRs, the resonant wavelength can be explained by an inductor-capacitor (LC) resonant circuit, and the resonant wavelength varies in accordance with the structure scale [6,32,33]. For the EIT metamaterials, when the gap between the dipole resonator and quadrupole resonator is varied, both the coupling state and the optical spectrum change [34–36]. Therefore, investigation of the optical spectra of these known structures reveals whether the printed optical metamaterials function effectively.

2. Design and fabrication

Figure 1(a) illustrates a conceptual diagram of the high-resolution embedding printing process for a flexible optical metamaterial. A Si wafer having the convex inverted shape of the metamaterial structure to be fabricated was used as a mold. In order to enhance the releasability in the nanoimprint process, a fluorinated release agent (Optool HD-1100Z, Daikin Industries) of a few nanometers thick was coated on the Si mold. A polymethyl methacrylate (PMMA) film (DF 02 U, Mitsubishi Gas Chemical) with 300-µm thickness was employed. A nanoimprint process was then performed using a thermal nanoimprint method, with a grooved pattern corresponding to the metamaterial structure being formed on the film substrate. Surface treatment of the film substrate was performed at this time with an oil-repellent fluororesin, to improve the printability by enhancing features such as the filling property and removability of ink outside the grooves. The grooves were filled with nanoink (nanoparticles) by squeegeeing the nanoink on the nanogrooved film. Silver nanoink (NPS, Harima Chemicals) having a representative primary particle diameter of 12 nm and metal content of 83% was used. A rubber squeegee was employed to apply the nanoink. Following the first squeegeeing process, some ink remained outside the grooves. Therefore, for the second squeegeeing, the unnecessary ink remaining outside the grooves was completely removed using a wiping cloth. Note that this wiping step may be performed multiple times until the ink is well removed. Due to the surface treatment applied to the film substrate, the ink in the grooves is not removed. Sintering was then performed at 130°C temperature and 0.2-kPa pressure using a vacuum oven. Hence, sintered bodies of nanoparticles were formed, which became silver nanostructures.

 

Fig. 1 (a) Fabrication process of printed optical metamaterials. (b) Printed flexible metamaterials in bent state. (c) Photograph of printed flexible metamaterials before sintering and after sintering at 130°C for 1, 2, and 4 h. (d) SEM images of printed metamaterials (dipole resonators with 250-nm length) for each sintering condition. The black scale bars indicate 200-nm length.

Download Full Size | PDF

Figure 1(b) shows a photograph of the printed flexible metamaterials. In general, the electric resistance of a printed ink pattern varies largely depending on the sintering conditions, and the electric resistance of fully sintered silver ink is close to that of bulk silver [37,38]. For reference, the conductivity of the sintered ink film was measured. After coating of the film substrate with the ink to a thickness of 4.5 μm using the squeegeeing method, the conductivity of the ink film sintered at 130°C and 0.2 kPa for 12 h was 69 μΩ cm. Although the conductivity of the printed metamaterial structure could not be measured, the conductivity is expected to have been higher than that of the 4.5-μm-thick ink film, because the sintering process had a greater effect due to the relatively small ink volume inside the nanogrooves. As the optical response varies depending on the sintering conditions, it is important to investigate the relationship between the sintering conditions and optical characteristics when considering optical applications of ink patterns. Figure 1(c) shows a photograph of the printed metamaterials before and after sintering for periods of 1, 2, and 4 h at 130°C. With increased sintering time, it is apparent that the color of the patterned areas gradually changed from the state before sintering. To clarify the reason for this color change, the printed patterns were observed using a scanning electron microscope (SEM, S-4800, Hitachi). Figure 1(d) shows SEM images of the printed dipole resonators for each sintering condition. Note that osmium was deposited with an osmium plasma coater (OPC60AG, Filgen) to prevent charge-up. For the specimen before sintering, the nanogroove patterns formed by the nanoimprint process and the nanoparticles filling the grooves can be seen. For the component subjected to sintering at 130°C for 1 h, several grown particles in the grooves can be observed, which may indicate porous structures. For sintering at 130°C for 4 h, particle growth to form nanostructures along the groove shapes is apparent. Therefore, although the recommended sintering temperature of the commercial ink used in this study is 230°C, it was found that this ink grows from nanoparticles to nanostructures at a considerably lower temperature. It is believed that sintering proceeds at a lower temperature for this setup because of the relatively small ink volume inside the nanogrooves; therefore, the energy requirement of this fabrication process is reduced.

In order to investigate whether the unique optical responses derived from the metamaterial structures also occurred in the sintered ink patterns, structures with different shapes and sizes were fabricated using high-resolution embedded printing. Figure 2(a) shows schematic diagrams of each unit cell structure and their dimensions: type-A and -B dipole resonators, type-A and -B SRRs, and the EIT metamaterial. The major-axis length dy of the type-A dipole resonator was set to 150–350 nm, the major-axis lengths dwy and py of the type-B dipole resonator were set to 100–400 and 250–1000 nm, respectively, and the gap distance g_e of the EIT metamaterial was set to 0–110 nm. It is known that the resonant wavelengths of dipole resonators made of vacuum-deposited materials vary upon adjustment of the values of dy and dwy [36]. Further, the resonant wavelengths of SRRs are explained by the theory of the LC resonant circuit model [32]; that is, the resonant wavelength changes according to the SRR size. In EIT metamaterials, the coupling strength varies depending on the g_e between the dipole resonator and the quadrupole resonator; therefore, the EIT-like response can be controlled [35,36].

 

Fig. 2 (a) Schematic diagrams of each unit-cell structure with specific dimensions: Type-A and -B dipole resonators, Type-A and -B SRRs, and EIT metamaterial. (b) SEM images of printed metamaterials: Type-A dipole resonator with dy of 350 nm, type-B dipole resonator with dwy of 300 nm and py of 750 nm, type-A and -B SRRs, and EIT metamaterial with g_e of 75 nm. (c) SEM image of A−A cross-section in (b).

Download Full Size | PDF

In this study, by confirming whether these plasmonic responses were generated in the sintered ink pattern, the applicability of the printed metamaterials could be clarified. Figure 2(b) shows SEM images of the printed metamaterials. Ink patterns with complex shapes, line widths of 100 nm or less, and gap distances of several tens of nanometers were well formed, and the structures were not porous after sintering was achieved. Because of the thermal nanoimprint process and the sintering process of the ink, the dimensions of the printed metamaterials were confirmed to exhibit approximately 10% shrinkage compared to the mold shapes. In addition, by cutting the film cross section using a microtome and performing SEM observation of the cross section, the thickness of the type-A dipole resonator was found to be approximately 100 nm, as shown in Fig. 2(c). As molds with the same pattern height were used for the other patterns, almost identical cross-sectional shapes were obtained.

3. Results and discussion

3.1 Type-A dipole resonators

To confirm the change in optical response before and after sintering, along with the polarization dependency, the optical spectra of the printed optical metamaterials were measured. Figure 3 shows experimental transmittance spectra. In detail, Figs. 3(a) and 3(b) show the transmittance spectra obtained when the silver nanoink used in the printing process was directly coated onto bare film. Before sintering, a dip at approximately 490-nm wavelength is apparent, which is caused by absorption of the nanoparticles in the silver nanoink. Polarization dependence was not observed. After sintering at 130°C for 12 h, the dip wavelength was confirmed to shift to 425 nm; however, again, polarization dependence was not observed. In the near-infrared region and the visible region excluding blue light, no difference was observed in the specimens before and after sintering, and no unique optical response was observed.

 

Fig. 3 Experimental transmittance spectra of (a), (b) silver nanoink coated on film and (c),(d) printed type-A dipole resonator. (a), (c) Before sintering. (b), (d) After sintering for 12 h at 130°C. The polarization directions are parallel to the y- (upper graphs) or x-axes (lower graphs), as indicated by the arrows. (d) Dashed curves showing calculated transmittance spectra of dipole resonators by RCWA simulations.

Download Full Size | PDF

Figures 3(c) and 3(d) show the experimental transmittance spectra of the printed dipole resonator before and after sintering. For comparison with optical metamaterials formed via the vacuum deposition process, the transmittance spectra were calculated through a rigorous coupled-wave analysis (RCWA) simulation, which can predict exact solutions for periodic structures based on Maxwell’s equations [39]. The calculation results are plotted as dashed curves in Fig. 3(d). For the dielectric constant of silver, the optical dispersion is shown in Ref [40].

As apparent from Fig. 3(c), the polarization dependence due to the structural shape was clearly confirmed. For polarization along the y-axis, it is apparent that the dip wavelength shifted from 754 to 841 nm as dy was increased from 150 to 350 nm. On the other hand, regarding the polarization along the x-axis, a slight dip can be seen at a wavelength of approximately 500 nm. As dy was increased, the transmittance became low, although there was no shifting of the dip wavelength. This is presumed to be because dy was lengthened while the 500-nm pitch was retained, so that the occupied area of the structure in the unit cell was increased. As described above, it was found that unique optical responses depending on the structural shape were obtained even in the printed ink pattern before sintering, and the optical anisotropy could also be controlled. Further, for the post-sintering case, clear polarization dependence was observed. Dips due to the dipole resonance are apparent in the transmittance spectra. When dy was 150, 250, and 350 nm, the dip wavelengths for polarization along the y-axis were 765, 1003, and 1346 nm, respectively, and the transmittances at that wavelength were 43.4, 30.7, and 15.2%, respectively. On the other hand, in the case of polarization along the x-axis, no optical response was observed in the wavelength region where the resonances were observed with y-polarization. Through comparison of the experimental and calculated values, it was confirmed that the resonant wavelengths and spectrum shapes were generally in good agreement. However, there is a major quantitative discrepancy in dip depth between the experimental and simulated spectra. The possible reasons for this are pattern nonuniformity, the differences between the dielectric constants of the sintered silver ink and bulk silver, and oxidation of the silver ink pattern. Overall, it was demonstrated that, for a component having a simple rod shape such as a dipole resonator, optical properties comparable to those of an optical metamaterial prepared via a vacuum deposition process are obtained in the printed metamaterial.

In order to confirm the flexibility of the printed metamaterials, the change in optical response during bending was evaluated. Figure 4 shows the experimental transmittance spectra of the printed type-A dipole resonator with a dy of 150 nm as a function of the bending radius r. When r was 15 and 7.5 mm, the amplitude of the strain ε_s was 1% and 2%, respectively. In Figs. 4(a) and 4(c), for bending along the y-axis, almost no change in optical response was observed, regardless of the compressive and tensile strains. In Figs. 4(b) and 4(d), for bending along the x-axis, almost no change in optical response was observed up to an r of 15 mm, irrespective of the compressive and tensile strains. However, when r was 7.5 mm, the resonant wavelength and dip depth were changed. As the spectrum changed similarly in both the tensile strain and compressive strain states, this behavior is thought to be due to the influence of the oblique incident light. Although it is necessary to investigate the influence of bending to a greater extent, it was found that defects and peeling do not occur in the printed metamaterial at an r value of approximately 15 mm.

 

Fig. 4 Experimental transmittance spectra of printed type-A dipole resonators as functions of bending radius r for compressive strain along (a) y- and (b) x-axes, and for tensile strain along (c) y- and (d) x-axes. The polarization directions are parallel to the y-axis.

Download Full Size | PDF

3.2 SRRs

To evaluate the optical response in a more complex structure, SRR structures of different sizes were fabricated using the embedded printing technique and their optical spectra were measured. The sintering was performed at 130°C for 12 h. Figure 5 shows SEM images and the experimental transmittance spectra of the printed SRRs. Although somewhat rough, structures close to the designed shape were formed. In the case of polarization along the x-axis, dips in the transmittance spectra occurred for both the type-A and -B SRRs. The resonant wavelengths were 668 and 1005 nm, and the transmittances were 22.5 and 12.4% for the type-A and-B SRRs, respectively. Compared with the simulation results, although there is a small difference in transmittance values, the spectral shapes and features, such as the resonant wavelength positions, are in good agreement. As regards polarization along the y-axis, two dips in each transmittance spectrum for the printed type-A and -B SRRs were observed. From Refs [6]. and [33], these SRR structures have two resonance modes: that at the shorter wavelength is the electric resonance, whereas the other is the magnetic resonance. Thus, electric and magnetic resonances were demonstrated in the printed SRRs. Further, these dip wavelengths shifted from 661 to 962 nm and 1024 to 1921 nm for the electric and magnetic resonances, respectively, with increased SRR size. Compared with the simulation results, the experimental results showed weak and broad optical responses; this was because the conductivity of the sintered silver ink is inferior to that of bulk silver and there were variations in the pattern shapes. However, we confirmed that different optical responses from the electric and magnetic resonances can be obtained in these printed SRRs. Therefore, it was demonstrated that the printed metamaterials can also exhibit the required functionality for shapes such as those of the examined SRRs. Next, EIT metamaterials composed of two kinds of resonators with bright and dark modes were fabricated using the embedded printing technique, and experiments were conducted to determine the occurrence of plasmonic coupling. In EIT metamaterials, when the plasmonic coupling between two resonators is weak, a dip caused by dipole resonance is observed in the transmittance spectrum. On the other hand, when the coupling is strong, an EIT-like effect occurs and a narrow transmission window is observed in the dip wavelength band. The coupling strength can be controlled by adjusting the gap distance g_e between the two resonators.

 

Fig. 5 SEM images and experimental transmittance spectra of printed type-A and -B SRRs. The polarization directions are parallel to the x- (left graph) or y-axes (right graph), as indicated by the arrows. The red and blue curves show the results for the type-A and -B SRRs, respectively. The dashed curves are calculated transmittance spectra of SRRs by RCWA simulations.

Download Full Size | PDF

3.3 EIT metamaterials

Figure 6 shows SEM images and experimental transmittance spectra of printed EIT metamaterials sintered at 130°C for 12 h, having different g_e values of 0, 32, and 110 nm. The transmittance spectra indicated by the solid and dashed curves in Fig. 6 are the experimental and simulation results, respectively. The electric and magnetic field amplitude distributions in the unit structures of the EIT metamaterials were obtained for g_e values of 32 and 110 nm at each resonant wavelength. The electric and magnetic fields (x- and z-components) were calculated at the pattern half heights. When g_e was 32 nm, the coupling between the dipole resonator and quadrupole resonator was strong and, thus, the quadrupole resonator was excited and an EIT-like effect occurred. On the other hand, when g_e was 110 nm, localized electric and magnetic fields appeared at the edges of the dipole resonator. As the coupling between the two resonators was weak, the quadrupole resonator was not excited. As a result, only the dipole resonator was excited and an EIT-like effect did not occur. In the simulation results, when g_e was 110 nm, a dip was observed at wavelengths of approximately 1200 nm, while the transmittance decreased to 25%. For g_e of 32 nm, EIT-like transmission was observed at wavelengths of approximately 1241 nm and the transmittance increased to 61%. On the other hand, according to the experimental results, no EIT-like effect was observed when g_e was 110 nm, dips were observed at wavelengths of approximately 1016 nm, and the transmittance decreased to 48.2%. However, for the 32-nm g_e, generation of the transmission window in the dip was confirmed, and the peak transmittance at that wavelength reached 69.7%. Comparison of the experimental and simulation results reveals that the spectral shapes are similar, although the optical response was weak. Therefore, it was demonstrated that the printed EIT metamaterial also functions in the optical region.

 

Fig. 6 SEM images and experimental transmittance spectra of printed EIT metamaterials with g_e values of 0, 32, and 110 nm. The polarization directions are parallel to the y-axis. The dashed curves in the bottom graph indicate calculated transmittance spectra of EIT metamaterials by RCWA simulations. The electric and magnetic field amplitude distributions in the unit structures of EIT metamaterials with g_e values of 32 and 110 nm at each resonant wavelength are shown. The electric (x-components) and magnetic (z-components) fields were calculated at the pattern half heights.

Download Full Size | PDF

3.4 Sintering conditions of printed metamaterials

Finally, in order to evaluate whether the sintering was sufficient and to examine the relationship between the sintering conditions and optical characteristics, the dependence of the printed metamaterial on the sintering conditions was investigated. We selected the type-B dipole resonator, in which interaction occurs between structures because of the narrow pitch in the x direction, for this investigation. Figure 7(a) shows an SEM image of the printed type-B dipole resonator, along with experimental transmittance spectra as functions of dwy and the sintering time. The bottom graph shows the calculated transmittance spectra by the RCWA method. In the case of polarization along the x-axis, it is apparent that dips resulting from interaction between adjacent structures occurred in accordance with dwy. These dips can be confirmed even before sintering. Further, as dwy entered the region of 100–400 nm, the dip wavelengths changed from 437 to 1427 nm. For a sintering time of 1 h, the dip width differed slightly; however, results almost identical to the spectral shapes before sintering (for example, in terms of the dip depth) were obtained. For a 2-h sintering time, although the dip width became narrower, the response became weaker when dwy was larger than 250 nm. As regards comparison with the simulation results, although the dip depth differed greatly, it can be stated that the dip wavelengths were roughly coincident for each dwy.

 

Fig. 7 (a) SEM image and experimental transmittance spectra of printed type-B dipole resonators as functions of dwy (from 100 to 400 nm) and for different sintering conditions. Bottom graphs: calculated transmittance spectra by RCWA simulations. The polarization directions are parallel to the x- (left graphs) or y-axes (right graphs), as indicated by the arrows. (b) Plots of λ₁ (left graph) and λ₂ (right graph) resonant wavelengths and transmittances at resonant wavelengths as functions of dwy and sintering conditions.

Download Full Size | PDF

Next, in the case of polarization along the y-axis, two resonant dips were observed, unlike the case of the type-A dipole resonator. Let the resonant wavelengths on the short and long wavelength sides be λ₁ and λ₂, respectively. Before sintering, the spectra differed in accordance with dwy, and polarization dependence could also be confirmed; however, sharp resonance was not observed. When dwy was 100 nm, optical anisotropy was not observed, as the structure had a square shape from the top-down perspective. For a sintering time of 1 h, although small dips were obtained for each dwy, the optical responses remained weak. On the other hand, for a 2-h sintering time, transmittance spectra in good agreement with the simulation results were obtained. Further, for sintering times of 4 and 12 h, almost identical spectra were obtained. As the experimental results are close to the calculation results, it can be concluded that the printed ink patterns were sufficiently sintered. In Fig. 7(b), both λ₁ and λ₂ are in good agreement with the simulation results. Further, the transmittances at λ₁ and λ₂ are slightly higher than the simulation results. However, for the sintering times of 4 and 12 h, the optical characteristics are almost unchanged and saturated. Therefore, for this printing process, a sintering time of approximately 4 h is sufficient.

Here, we would like to emphasize that the sintering was performed at 130°C, although the recommended sintering temperature of the ink employed in this study is 230°C. When application to flexible devices is targeted, as in this study, a low-temperature process is an indispensable condition, because the thermal resistance of the polymer film substrate must be carefully considered. It is thought that sintering can be completed even at this low temperature for this process because the thermal energy is applied to a very small volume corresponding to the amount of ink filling the nanogrooves; this is an important finding as regards realization of high-performance optical devices using printing technology. In addition, another new finding was obtained, i.e., optical spectra close to simulated results can be obtained for optical metamaterials, which are strongly affected by the dielectric constants of the materials, even in sintered ink structures. As regards the sintering process, the experiments were performed using a 130°C oven in this study; however, we would like to experiment with other sintering temperatures and other methods such as photonic and/or microwave sintering [37,38] in the future.

4. Conclusion

In conclusion, printed optical metamaterials made of commercially available ink consisting of silver nanoparticles were experimentally demonstrated. The printing method was a combination of the thermal nanoimprint method and the squeegeeing method. We fabricated three types of metamaterials: dipole resonators, SRRs, and EIT metamaterials. In optical evaluations, dipole resonances corresponding to the lengths of the printed dipole resonators were observed at wavelengths from 765 to 1346 nm. Evaluation of the flexibility revealed that defects and peeling do not occur in the printed metamaterial at a bending radius of approximately 15 mm. For the printed SRRs, electric and magnetic resonances were observed and the resonant wavelengths were tuned from 661 to 962 nm and 1024 to 1921 nm, respectively, as the SRR size increased. For the EIT metamaterials, when the gap distance between the two resonators was 32 nm, EIT-like transmission was demonstrated, and the peak transmittance at that wavelength increased to 69.7%. The sintering conditions of the printed metamaterials were also investigated; hence, we confirmed that the optical characteristics of metamaterials sintered at 130°C for 4 h or more qualitatively agreed with simulation results. In addition, as optical anisotropy was also observed in each metamaterial structure, the obtained optical spectra were derived from the individual printed structures. An important finding was made, i.e., for metamaterials strongly affected by their constituent materials, a metamaterial structure made of ink exhibits optical properties comparable to those produced by the vacuum deposition process. Moreover, it was found that unique optical responses also occur in the ink structure before the sintering process, and the optical responses vary greatly depending on the sintering conditions. We hope that this printing technique, which can form large-area nanostructures directly on thin polymer film, will contribute to the development of flexible applications.