1. Introduction

Chiral optical metamaterials attract strong interest due to their giant circular dichroism controllable via artificial structuring [1]. Most straightforward way to realize them is by assembling periodic 2D or 3D arrays of chiral particles, such as helices, but in practice this is difficult, especially at optical frequencies, where 3D fabrication with high spatial resolution is required in order to properly downscale unit cell of the structure. While some of the available fabrication techniques, such as Glancing Angle Deposition (GLAD) [2] or focused ion-beam induced deposition [3, 4] provide high spatial resolution and seem promising for fast mass-production, they still lack flexibility in terms of control over geometry (i.e., shape, size, orientation, handedness) of the inclusions. The most versatile approach used so far uses Direct Laser Write (DLW) technique [5, 6] to fabricate dielectric templates of structures which can be subsequently metalized. DLW is non-linear, maskless 3D laser lithography capable of nearly arbitrary 3D structuring of photoresists and transparent dielectrics with sub-micron spatial resolution, and has been fruitfully applied for fabrication of dielectric optical micro-structures, such as 3D photonic crystals [7], and diffractive elements [8]. DLW and metalization have been fruitfully exploited to demonstrate metallic photonic crystals [9] and helix-based metamaterials, applicable as broadband circular polarizers at infra-red (IR) frequencies [10,11].

Although helical metamaterials also exhibit other exotic optical properties, such as optical activity, polarization transformations, and perfect absorption phenomenon, their deeper experimental exploration at optical frequencies is impeded by difficult fabrication. Perfect absorbers (PA) are materials whose reflectance and transmittance vanish simultaneously, leading to unity absorbance. Thin resonant PAs with structurally-tunable optical properties can be realized using electromagnetic metamaterials [12]. Helix-based metasurface PAs operating at infra-red (IR) or near-infrared (NIR) frequencies would be attractive candidates for use in narrow-band thermal absorbers, emitters, energy converters, and applications relying on enhanced IR absorption [13, 14]. Unlike planar metamaterial PAs consisting of densely packed patterned metallic and dielectric films [15–18], helix-based PAs have sparse all-metallic 3D architecture, which is permeable to liquid or gaseous substances, and is more stable against thermal damage due to better heat dissipation [19]. Moreover, helix-based PAs do not require metallic substrate for their operation and thus in principle may enable PAs which are transparent at off-resonant frequencies.

Recently, helical PA structure consisting of macroscopic periodically arranged single-turn helices was proposed and successfully realized for radio-frequency (RF) spectral range [20] by mechanical machining. Tuning the PA resonance towards IR or NIR frequencies would require scaling the unit cell down to micrometers, or by ~ 10⁴ times, which is a challenging task. Therefore helix-based PAs for optical frequencies have been mainly explored via numerical simulations [21, 22]. In this study we address tunability of the helix-based PA architecture to IR frequencies experimentally, via use of DLW technique and simple metalization process. We demonstrate that IR spectral range is indeed accessible using this relatively simple and accessible route. In accordance with theoretical expectations, the fabricated samples exhibit resonant absorption bands with peak absorbance A > 0.8 and central wavelength λ_c tunable within the 6 – 11 μm range by modifying geometrical parameters of the unit cell. Although simple metalization process used in this study has inadvertently created reflecting metallic plane behind the helices, thus preventing demonstration of genuinely transparent PA, our data suggest that dominant contribution to absorption in our structures comes from helical resonators, with substrate playing a minor role. It also demonstrates that transparent PAs can be realized in the future via use of more advanced, spatially selective metalization methods.

2. Design of helix-based PA architectures

A generic metasurface, or periodic array consisting of electric and magnetic resonators/dipoles will exhibit perfect absorption when amplitudes of reflected and transmitted waves vanish simultaneously. This condition imposes specifically that i) electric and magnetic resonances must have equal strengths and occur at the same frequency, ii) individual polarizabilities of resonators must be purely imaginary, iii) electromagnetic and magnetoelectric coupling in the whole structure must be compensated, and iv) the constituent materials must possess finite dissipative loss. Using these criteria, tailoring of individual resonators and their periodic arrays can be performed to realize PA. Physical principles underlying metamaterial PA are described in the existing literature [20, 23, 24]. Various resonator topologies, such as omega inclusions [25], canonical helices [26], and single-turn helices [27] were designed following these criteria. Metamaterials implementing other functionalities, for example circular or twist polarizers [27,28] can be also designed using similar principles.

Fig. 1(a) shows schematic architecture of helix-based PA. Its primary building block is a group of four metallic single-turn helices centered on the corners of a square, and oriented anti-parallel along its diagonals. Such symmetry creates balance between optically induced electrical currents in the opposite helices for normally-incident electromagnetic waves of any polarization. By selecting appropriate parameters (helix diameter d and pitch h, helix arm diameter w, spacing between helices within the group s), one can achieve compliance with conditions (i) and (ii). Chirality of the helices and condition (iii) require inclusion of equal number of left-handed (LH) and right-handed (RH) helices, which is done by alternating four-helix groups as shown in Fig. 1(a). Hence, unit cell of low-reflection helix-based PA structure in general comprises 4 groups separated by distance p. To satisfy the condition (iv), resonators comprising the unit cell must exhibit a moderate amount of dissipative loss without drastic reduction of their quality factor.

 

Fig. 1 (a) Geometry and parameters of low-reflectance helix-based PA structure for IR spectral range, with right-handed and left-handed helices shown in yellow and blue, respectively, (b) simulated optical spectra of PA with parameters indicated in the Fig., (c) geometry and parameters of a realistic PA structure obtainable via DLW and metalization, the inset illustrates shape of a single helix fabricated by DLW, (d) simulated absorbance spectra at normal incidence for chiral-plate and compensated PA structures with optimized parameters.

Download Full Size | PDF

Helix-based PA architecture can operate without a metallic substrate, thus enabling high off-resonance optical transparency and low reflectivity. Low-reflectance helix-based PA was successfully designed using the principles described above, and realized in practice for RF spectral range [20,24]. At these frequencies the required unit cell size is on the order of centimeters, and PA was assembled from macroscopic helices made of Ni-Cr wire. However, practical realization of even simplest helix-based architectures for NIR and IR spectral ranges is difficult, since drastic downscaling of the unit cell is required. Here, this challenge is addressed by employing high-resolution DLW fabrication to prepare dielectric templates of PA structures, followed by their metalization using plasma sputtering. In order to adapt helix PA architecture for IR spectral range, we have downscaled the unit cell, and took into account peculiarities of the fabrication techniques and materials used. Geometry of PA was optimized to achieve best compliance with the requirements (i–iii), and metal for the helices was selected in order to satisfy the requirement (iv). We note beforehand, that among several noble metals (Au, Ag, Al, Cu) examined the best results were obtained for gold, most likely due to its moderate optical losses at IR frequencies [29]. Therefore, gold was used in this study.

Theoretical design and optimization was performed using numerical simulations based on electromagnetic Finite-Element Method (FEM). First, we examine the effect of proportional downscaling of the PA unit cell on the optical properties. Fig. 1(b) shows simulated reflectance, transmittance, and absorbance spectra of the structure (geometrical parameters are indicated in the Fig.) for linearly-polarized, normally-incident waves. Here, helices are constructed using circular gold wire. As expected, the structure exhibits low reflectivity (R < 0.05) and high transmission across a broad spectral range, except in vicinity of the resonance wavelength of 8.8 μm, where low transmission and strong PA resonance with absorbance reaching A = 0.98 are obtained.

In addition to downscaling, further modifications of the idealized low-reflectance PA design are necessary. In order to mechanically support helices above glass substrate on which photoresist is deposited prior to DLW, vertical rods of length l were added between the helices and the substrate as shown in Fig. 1(c). Furthermore, non-uniformly elongated elliptical cross-sectional shape of helix arms fabricated by DLW has to be taken into account during design and numerical simulations. Since focal region of the writing laser beam is elongated along the z-axis direction, arms of helices fabricated by translating the focus also have elongated elliptical cross-section. The elongation becomes most pronounced in horizontal segments of helices (see inset in Fig. 1(c)), where it reaches maximum aspect ratio of w_z/w_xy ≈ 2.8 determined by numerical aperture NA=1.35 of the focusing lens and dominant two-photon absorption at the focus.

Also, it is necessary to consider the role of metallic substrate resulting from spatially non-selective metalization process used. Plasma sputtering allows fast, conformal deposition of metal films on curved surfaces. However, entire sample, including helices and the underlying glass substrate becomes metalized as illustrated in Fig. 1(c), making PA structures optically opaque. Strictly speaking, this circumstance prevents realization of transparent PA metasurface in this study. Nevertheless, other attractive characteristics related to the sparse all-metallic architecture of the helix layer (e.g., environmental permeability, heat exchange capability) are retained. According to our theoretical simulations, resonant oscillations and extinction of the optical field occur mainly in the helices, whereas losses in the underlying metallic substrate are much lower.

Metallic substrate also acts as inverter of circular polarization, which allows one simplify the unit cell of PA by including helices of single handedness only. In such a case, circularly dichroic helix layer [30] transmits half of the incident radiation, whose circular polarization becomes inverted by the reflecting substrate, leading to complete extinction in the helix layer. In our stuctures, where metallic substrate is always present, “compensated PA” architecture comprised of both LH and RH helices can be simplified to a single 4-helix group, and the architecture can be referred to as a “chiral-plate” PA [31] due to dominant chirality of the helix layer. Both architectures are schematically shown in Fig. 1(c). Fig. 1(d) compares simulated absorbance spectra of compensated and chiral-plate PA structures with optimized parameters. Here, realistic shapes of helices and vertical supporting rods were included, and coating by Au film with thickness t = 80 nm exceeding the skin-depth at IR wavelengths was assumed. As can be seen, both architectures exhibit nearly identical performance with absorption A > 0.99 at the resonance wavelength of 7 μm. This theoretical result is also supported by experimental data (see Fig. 3(f) and discussion below).

Fig. 2(a) shows simulated absorbance of the PA structure as a function of incidence angle for transverse electric (TE) and transverse magnetic (TM) linearly polarised waves. One can see that resonance with nearly-invariant central wavelength and absorbance A > 80 is retained for up to 30° for TE polarization. For TM polarization, A > 0.95 is retained for up to 45°, except for ≈ 10°, where sudden drop is observed. Anisotropic and non-monotonous angular dependence of absorbance can be tentatively ascribed to parasitic dipole moments which evolve at slanted incidence, and violate the conditions of perfect absorption. Nevertheless, one can estimate polarization and incidence angle-averaged absorbance A ≥ 0.7 for incidence angles up to 30°. Notice, that high-absorbance can be still also obtained at certain wavelengths for incidence angles as high as 80° for both polarizations.

 

Fig. 2 (a) Simulated absorbance spectra as a function of incidence angle of a chiral-plate PA structure for TE and TM polarizations, respectively, (b) electric field distribution in the unit cell (side view), (c) distribution of surface current density at TE and TM polarization of incident light, respectively (top view).

Download Full Size | PDF

 

Fig. 3 (a) Fabrication of dielectric templates using DLW lithography, (b) metalization of samples by gold sputtering using a stage tilted at 45° angle. The sputtering was done four times for different sample orientations. SEM images of (c) compensated and (d) chiral-plate PA samples, with insets showing unit cell, (e) detailed view of a single-turn metallic helix, (f) experimental absorbance spectra of compensated and chiral-plate PA structures.

Download Full Size | PDF

Fig. 2(b) shows distribution of electric field in the unit cell of a helix-based chiral-plate PA architecture. The fields are shown on a cross-sectional plane parallel to the z–axis direction, at the resonant wavelength of 7 μm. As can be seen, electric field is predominantly localized in the helix layer. Fig. 2(c) shows current density distribution at the resonance for excitation by normally incident TE and TM polarized waves. Strictly speaking, distinction between TE and TM polarizations is lost at normal incidence, but here we use these definitions to define two mutually orthogonal linear polarization states. Selective excitation of different helix pairs by TE and TM excitation can be seen. For intermediate orientation, contribution from both pairs leads to polarization-invariant response. We emphasize, that current density is highest in the inner circumference of the helices, which therefore defines length and resonance frequency of individual resonators in the structure. This circumstance may be helpful for realization of helix-based PA for higher frequencies, because helices with thicker arms (which are easier to fabricate) can be used, provided that their inner circumference radius is small enough. Current density in the vertical supporting rods (not seen in top-view images) is much lower, suggesting that at these frequencies helix layer and substrate essentially behave as electrically isolated entities, despite the metallic rods.

3. Structural and optical properties

Experimental details regarding sample fabrication, processing and characterization are described in Sect. 5. Figs. 3(a,b) schematically show implementation of DLW lithography technique used to fabricate dielectric templates of PA structures, and their metalization by gold sputtering. Figs. 3(c) and (d) show Scanning Electron Microscopy (SEM) images of the compensated and chiral-plate PA architectures, respectively. Design parameters of these structures were were obtained and optimized using numerical simulations. Their actual parameters estimated from SEM are slightly different due to shrinkage and deformation of photoresist templates. It should be noted that lateral width of the helix arms is about 220 nm, while thickness of metallic film is about 70–90 nm, which shows that photoresist templates were fabricated with sub-150 nm spatial resolution. Fig. 3(e) shows a detailed SEM image of a single-turn metallic helix. Obviously, the samples exhibit high uniformity, although some disorder and nanoscale surface roughness can be seen.

Optical properties of PA samples were characterised using Fourier-Transform Infra-Red (FTIR) spectroscopy. Experimental IR absorbance spectra of compensated and chiral-plate PA architectures are shown in Fig. 3(f). A broad absorption band centred near the 7.6 μm wavelength with a peak absorbance A = 0.8 can be seen. Although simulations predict a stronger, narrower and blue-shifted absorption band, qualitative agreement with experimental data is satisfactory, since simulations neglect random variation of the helix parameters and nanoscale roughness of the metal, which may lead to spectral shifts and inhomogeneous broadening. Also, homogeneous broadening due higher optical loss in polycrystalline sputtered gold may contribute to the difference.

As mentioned previously, chiral PA structures are dichroic and absorb only one circular component of the incident wave. Thus, an ideal chiral PA without a metallic substrate would exhibit R = 0, T = 0.5 and A = 0.5 for linearly-polarized or unpolarized incident waves at the resonance. Adding a reflecting metallic substrate behind the chiral layer would result in T = 0, but R = 0 (leading to A = 1) would be retained despite high reflectivity of the substrate, since circular polarization of unabsorbed components becomes inverted upon reflection, leading to their absorption in the helix layer as discussed above. Indeed, experimentally observed reflectance in our samples drops down to R < 0.1 at the resonance wavelength of 7.7 μm, leading to A > 0.9. As can be seen from Fig. 3(f), absorbance spectra of chiral-plate and compensated PA structures are nearly identical, as was suggested previously by simulations (Fig. 2(d)). This data also allows to expect that performance of a compensated PA structure would be nearly the same (except for off-resonance transparency) once the metallic substrate is removed.

As can be seen from Fig. 3(f), absorbance of the compensated PA exhibits an additional weaker peak near λ = 9.2 μm. This peak was also observed in other compensated PA samples, and is qualitatively reproduced by simulations (a narrow spike near λ = 7.8 μm in Fig. 1(d)). Detailed analysis of the simulated field patterns associated with this peak has revealed that its physical origin can be associated with electromagnetic coupling between groups of four helices of different handedness, in contrast to the main resonance which can be associated with coupling between helices within groups of the same handedness. Therefore, the secondary peak can be also associated with compensated chirality of the PA structure.

Spectral tunability of PA structures is illustrated by the data shown in Fig. 4. For these studies chiral-plate PA structures with unit cell parameters proportionally scaled down by the factor of 1.2 were fabricated on the same substrate. The substrate was subsequently investigated by an imaging FTIR spectrometer. Experimental absorbance spectra of the structures shown in Fig. 4(a) illustrate tunability of PA resonance in the 6 – 11 μm wavelength range. 2D reflectivity maps taken at the resonant wavelength of each structure are shown in Fig. 4(b). In the images, blue, low-brightness areas correspond to low reflectivity and high absorbance at the resonance. On the other hand, yellow, high brightness areas indicate high off-resonance reflectance and low absorbance, since helix layer is nearly transparent to radiation, and does not block the radiation incident on the sample or reflected from the metallic substrate. Reflectance in the gaps separating different PA structures does not increase to the reflectivity of the gold film because size of the imaged area during the scanning, (5×5) μm², is comparable to the gap width, and scanning in the gaps areas always integrates some loss due to residual scattering from the neighboring PA areas. Scanning further away from the PA (not shown) results in high reflectivity of bare gold film. Overall correlation between spectra and images shown in Fig. 4(a,b) is good. PA structures may be regarded as spectrally-sensitive pixels, from which larger arrays can be composed for imaging-related applications. Such arrays can be relatively easily prepared using DLW-assisted fabrication.

 

Fig. 4 Spectral tunability of chiral-plate PA structures, (a) reflectance spectra of several samples with unit cell size proportionally downscaled by the factor of 1.2 (from top to bottom, (b) 2D reflectivity maps of the samples measured at their respective resonance wavelengths of λ = 11.1, 9.4, 7.7, and 6.1 μm, (c) simulated PA absorbance spectra versus the length of supporting vertical rods l, (d) experimental absorbance spectra of samples with different l.

Download Full Size | PDF

Interestingly, spectral tuning of PA resonance can be also achieved without scaling the helices, but by changing length l of the vertical rods instead. This possibility is illustrated in Fig. 4(c,d). According to simulations, resonance wavelength of chiral-plate PA structures increases linearly with l as is demonstrated by Fig. 4(c). Notice that available tuning range in this case (λ ≈ 6 – 9 μm) is nearly as broad as that achievable by scaling the unit cell. At shortest wavelengths examined (λ ≈ 6 μm), a second line of peaks can be seen emerging in the top-left corner of the plot when l is increased by an amount of Δl ≈ λ/2. This behaviour suggests formation of a planar waveguide modes confined between the helix layer and the substrate. By changing the rod length l, thickness of the waveguide (or size of an extended 3D unit cell) is scaled, leading to proportional spectral shifts. Hence, in this case interactions between the metallic substrate and the helix layer are important. Numerical simulations represent qualitatively the experimental data shown in Fig. 4(d), where absorbance spectra of several PA samples having different rod lengths are shown. As one can see, practical tuning of PA resonance within the 6 – 8 μm wavelength range is possible. Since fabrication of vertical rods involves simple single-axis translation, adjusting the rod length provides a highly practical alternative way to fine-tune the PA structures.

4. Conclusion

Although helix-based PA structures are 2D single-layer metasurfaces, they are composed of complex 3D metallic inclusions which makes their fabrication and tuning to optical frequencies difficult. This work demonstrates that DLW-based approach can make helix-based PAs available for practical investigations and applications. One may wonder whether it makes sense to pursue complex chiral geometries, since very efficient PAs for spectral range extending into visible frequencies can be facilitated using traditional planar lithography. In this regard it is helpful to emphasize here, that DLW-based rapid prototyping can naturally deliver complex 3D sub-micrometer scale structures without limitations of the planar approach. Throughput of DLW technique is being constantly improved, making mass-production of large-size samples possible in the future. In our samples, polarization-invariant absorption A ≥ 0.8 within IR wavelength range of 6–11 μm at normal incidence, and A ≥ 0.7 at incidence angles up to 30° was estimated. These results clearly illustrate the possibility to downscale unit cell of helix-based absorbers thus tuning the PA resonance to IR spectral range.

One unique property of helix PA architecture is its broadband electromagnetic transparency. It was not possible to demonstrate genuinely transparent helix-based PAs in this study due to non-selective nature of the simple metalization process used. However, theoretical analysis and numerical simulations indicate that our PA structures would become transparent if metalization was selectively applied to helices only. Practical fabrication of transparent, optically thin metasurface with resonant perfect absorption at IR frequencies using fast prototyping technique such as DLW would be an important achievement, and although our work has not yet reached this goal, it can be regarded as the first step in this direction. Selective metalization may allow fabrication of thin single-layer PAs or even multi-layer structures capable of multi-frequency operation [32]. Helix-based metamaterial architecture may be also exploited for other optical functions, for instance transmit-arrays, metalenses, and polarizers.

5. Experimental details

Fabrication of dielectric templates by DLW technique

The laser source was a femtosecond oscillator (Mai Tai, Spectra Physics) with a temporal pulse length of 120 fs, a central wavelength of λ = 800 nm, and a repetition rate of 80 MHz. The beam was focused into the sample using oil-immersion microscope objective lens with a numerical aperture of NA=1.35. During the DLW, the sample was translated using high-precision 3D piezo-stage (Physik Instrumente, P-563.3CD). Spatial resolution approaching 100 nm was obtained by carefully adjusting average laser power slightly above the photopolymerization threshold. Typically, 3 mW average power at translation speed of 10 – 20 μm/s was used.

Samples for DLW

The initial material was negative-tone Zr-containing hybrid organic-inorganic photoresist SZ2080 with 0.4% 4,4′-Bis(diethylamino)benzophenone added as photo initiator [33]. The samples were prepared by drop-casting photoresist on a microscope cover glass substrates (Matsunami) and subsequently drying them on a hot plate using temperature ramp (for 5 min) between 40, 60, and 80°C (for 20 min). After the DLW the samples were developed in 1-propanol:isopropanol (50:50) solution for 5 min, rinsed in ethanol, and dried in a super-critical point drying apparatus (JCPD-5, Jeol).

Metalization

Deposition of gold film was performed in a plasma sputterer (Quick Coater SC-701 MkII). To produce coating as uniform as possible, the samples were mounted tilted at a 45° angle, and sputtered at four different sample orientations (0, 90, 180, and 270°). See Fig. 3(b) for more details. The resulting Au film had maximum thickness of approximately 90 nm.

Characterization

Visual observation was performed using 3D laser microscope (Keyence, VK-X200), and SEM (JEOL-7800). Optical characterization of reflectance spectra was performed using FTIR micro-spectrometers (MFT-2000, and FT/IR-6300,IRT-7000 Jasco).