Lilit Gevorgyan,
Hovhannes Haroyan*,
Henrik Parsamyan
and
Khachatur Nerkararyan
Department of Microwave Physics, Yerevan State University, A. Manoogian 1, Yerevan, 0025, Armenia. E-mail: hharoyan@ysu.am
First published on 17th April 2023
A simple design of an ultra-broadband metamaterial absorber (MMA) of terahertz (THz) radiation based on vanadium dioxide (VO2) configurations is proposed. The system is composed of a top pattern representing orderly distributed VO2 strips, a dielectric spacer and an Au reflector. Theoretical analysis based on the electric dipole approximation is performed to characterize the absorption and scattering properties of an individual VO2 strip. The results then are used to design an MMA composed of such configurations. It is shown that the efficient absorption characteristics of the Au–insulator–VO2 metamaterial structure can be ensured in a broad spectrum of 0.66–1.84 THz with an absorption band relative to the center frequency reaching as high as 94.4%. The spectrum of the efficient absorption can be easily tuned via the corresponding choice of strip dimensions. Wide polarization and incidence angle tolerance for both transverse electric (TE) and transverse magnetic (TM) polarizations are ensured by adding an identical parallel layer rotated by 90 degrees in respect to the first one. Interference theory is applied to elucidate the absorption mechanism of the structure. The possibility of modulation of the electromagnetic response of MMA relying on the tunable THz optical properties of VO2 is demonstrated.
In this paper efficient broadband terahertz (THz) absorption of an elongated elliptical subwavelength configuration made of VO2 is investigated. The electric-dipole approximation is used to demonstrate the absorption enhancement of an elongated ellipsoid arising due to configurational resonance associated with a special combination of material properties and geometrical parameters. Finite element method-based simulations are conducted for a rectangular strip as an equivalent configuration for elongated oblate ellipsoid. It is shown that ordered arrangement of such strips on a surface of an Au-backed dielectric layer provides an absorption band exceeding 90% in a wide frequency range from 0.66 to 1.84 THz with the absorption bandwidth (BW) of ∼1.18 THz. The bandwidth relative to the center frequency of the absorption band (RBW) is ∼94.4%. Owing to controllable THz optical conductivity of VO2 the absorptance can be modulated from 0 up to 100% by varying the surrounding temperature. VO2 also provides an efficient platform for optimization and establishing prospective material characteristics for near-perfect absorption. Interference theory is applied to reveal the absorption mechanism.
p = ε0ε2VχE, | (1) |
![]() | (2) |
![]() | (3) |
The absorption and scattering cross-sections of a particle can be defined as follows:32
![]() | (4) |
![]() | (5) |
For quantitative analysis the absorption and scattering properties are normalized by the longitudinal geometrical cross-section area S = πayaz. The corresponding quantities called absorption and scattering efficiencies are:35
![]() | (6) |
![]() | (7) |
![]() | (8) |
Analytical relation (3) derived within the quasistatic limits represents the depolarization factor of an oblate ellipsoid whose scattering and absorption efficiencies are shown in Fig. 1 by red and black solid lines, respectively. VO2 is used as a strip material and the surrounding medium is air.
Both the absorption and scattering efficiencies gradually increase as the frequency increases, however, the absorption is noticeably larger than the scattering and is around 10 at frequencies above 1 THz, whereas the Qscat is about 0.3. This means that the functional scattering surface of the configuration is much smaller than its geometrical cross-section. However, from a standpoint of integration with planar surfaces, a strip-like configuration can be promising also due to the fact that it is more straightforward for possible fabrication. Hence, hereinafter we use a strip-like configuration with identical geometrical parameters as those of an ellipsoid: the side lengths of the strip di in each coordinate direction are related to the semi-axes ai of the ellipsoid by the relation di = 2ai, where i = x, y, z.
Three-dimensional full-wave numerical simulations based on the finite element method were conducted to analyze the absorption and scattering properties of a VO2 strip. Absorption and scattering efficiencies of a strip are shown in Fig. 1 by black and red dashed lines, respectively. While the scattering properties of both configurations are similar, certain differences are observed between the absorption spectra. In particular, the absorption efficiency spectrum of a strip has resonant nature with a maximum value of about 12 at 1 THz. Here one should note that although the difference between the lateral cross-section areas of a strip and an ellipsoid is small, the volume of a strip is almost twice as large as that of the ellipsoid, which results in an increased absorption efficiency. Comparison of the simulated absorption and scattering properties of VO2 strips with dy values of 80, 94 and 120 μm are shown in Fig. 1(b). Solid lines stand for the absorption and dashed lines – for the scattering efficiency. As expected, a spectral shift towards lower frequencies is observed as dy increases. At the same time, the peak value of the absorption efficiency is almost constant and is around 12. Moreover, the scattering efficiency is negligibly small even when at the largest considered case of dy = 120 μm and barely reaches 0.4 at the peak frequency of 0.8 THz. In these calculations, dx and dz are as for Fig. 1(a). Once the conditions for the enhanced absorption dominating over the scattering are established for an individual strip-like configuration, we used the results above to design a VO2–dielectric–metal metamaterial for near-perfect broadband absorption of the THz irradiation. The metamaterial is composed of planar layers of an Au reflector and a dielectric spacer with nd = 1.5. We want to mention that the dielectric properties of a number of polymer materials, such as polypropylene36 and polyethylene cyclic olefin copolymer37 are suitable to use as a spacer. The third layer represents an array of VO2 strips periodically distributed on the top surface of a dielectric layer. A three-dimensional view of a metamaterial and the top view of the unit cell are depicted in Fig. 2(a). tm and td are the thicknesses of the bottom reflector and spacer, respectively, px and py are periodicities along the indicated directions, and dx and dy are the lateral dimensions of the strip. We used the optimized dimensions for strips and changed the periodicities and spacer thickness to achieve an efficient broad absorption band corresponding to the absorbance equal to or larger than 90%. Generally, the absorptance of the metamaterial is defined as A = 1 − R − T, where R and T are reflectance and transmittance. The latter are defined via frequency-dependent scattering parameters as R = |S11(f)|2 and T = |S21(f)|2. Since the transmission through the structure is blocked by a 3 μm-thick bottom Au layer (T = 0), which is much larger than its skin-depth (δs,Au ∼ 100 nm around 1 THz), the absorption coefficient is expressed only through the reflectance as A(f) = 1 − R(f). The absorption performance of the proposed system is investigated under the y-polarized plane wave. Fig. 2(b) plots the absorption spectrum for spacer thicknesses of 28 μm (green, diamonds) 32 μm (red, rings), 36 μm (black, squares) and 40 μm (blue, triangles). The sizes of a strip are chosen as in Fig. 1. As the spacer thickness decreases, the absorptance at lower frequencies drops below 0.9 level and a rising tendency of the absorptance at higher frequencies is observed. For instance, the efficient absorption band (A ≥ 0.9) for td = 28 μm is ∼1.11 THz (from 0.94 up to 2.05 THz), and that of td = 40 μm is ∼0.72 THz (from 0.59 up to 1.31 THz).
A compromise between the efficient absorption bandwidth and average absorptance above 0.9 is achieved for td = 36 μm. In this case, efficient absorption is achieved from 0.66 up to 1.84 THz with the absorption BW ∼ 1.18 THz. The average absorptance is about 0.93. Another criterion for the absorption performance is the relative bandwidth defined as RBW = BW/fc × 100%, which is about 94.4%. Although at td = 32 μm the absolute absorption BW (1.22 THz) is slightly larger than that of td = 36 μm (1.18 THz), our choice of the optimal spacer thickness of 36 μm is conditioned by the larger RBW and average absorptance of about 94.4% and 0.93, respectively, in contrast to 90% and 0.92 of the former case. Note that the total thickness of the optimized structure is 39.2 μm, which is around one-sixth of the centre wavelength of the efficient absorption band.
The structural asymmetry of the unit cell element dictates certain polarization dependence of the absorptance. Shown in Fig. 2(c) in red (circles) and black (squares) are the absorption spectra of the PS under x- and y-polarized plane waves. It is noted that near-perfect broadband absorption is achieved only when the incident field polarization is parallel to the longest side of strips, whereas, under cross-polarized radiation, strips do not interact with electromagnetic waves which are completely reflected from the bottom Au layer. To find out how the strip-like pattern of the top interface affects the absorption characteristics of the system, in Fig. 2(c) we compared the absorptance of the main design of the MMA with that of a planar Au–dielectric–VO2 structure (from bottom to top) shown by a blue-triangles line. The absorptance of the dielectric spacer sandwiched by 3 μm-thick Au and 200 nm-thick VO2 layers is around 0.22 at the entire considered spectrum up to 2.5 THz, whereas introducing rectangular strips noticeably enhances the absorptance ensuring near-perfect absorption characteristics. The absorption here is mainly conditioned by the attenuation of the electromagnetic waves while passing forward and backward through a thin and lossy VO2 top layer.
The next step includes optimization of the lateral sizes of the unit cell while keeping the strip sizes constant. The absorption spectra for different values of periodicities px and py are shown in Fig. 3(a) and (b), respectively. The absorption performance alters in a similar manner when either of the periodicities is increased, namely, the bandwidth is decreased and the overall absorptance is increased. The values of px = 6 μm and py = 114 μm were found to be optimal to ensure a broad efficient absorption band. Starting from py = 114 μm, a solid broad efficient absorption band is formed, and at larger values of py the average absorption constantly increases and the BW decreases. Such a behaviour can be attributed to reducing the coupling between neighbouring strips when their distance increases, which results in dominant configuration resonances of individual strips. Since all strips are identical, the resonance conditions also match. When the distance between strips is too large to establish coupling between neighbouring strips with a different frequency of the coupled system, resonant absorption bandwidth decreases.
From the results presented in Fig. 1(b), one sees that the corresponding choice of the strip dimensions can alter the absorption spectrum. Fig. 3(c) sums up the tunability of the efficient absorptance spectrum of the MMA by varying the unit cell element sizes. First, a strip length is chosen with other geometrical parameters of the strip fixed and then the unit-cell dimensions were optimized to achieve a relatively broad absorption band. In particular, shown by red (triangles), black (squares) and blue (rings) lines (symbols) are the absorption spectra for the MMA with dy of 80, 94 and 120 μm, respectively. One notes that the efficient absorption spectrum gradually shifts to the higher frequencies as the strip length decreases, also resulting in the broadening of the absorption band. For instance, at dy = 120 μm the absorption band lays in 0.49–1.36 THz, with the corresponding BW (RBW) of 0.87 THz (94%). On the other hand, the efficient absorption band at dy = 80 μm expands from 0.92 up to 2.37 THz, with BW (RBW) of 1.45 THz (88%). These results clearly demonstrate the tunability of the suggested periodic structure, in all the considered cases achieving relative bandwidths around 90%. Finally, the top and bottom plots in Fig. 3(d) show the absorption spectra at different values of the real and imaginary parts of the spacer refractive index, respectively. Although the absorption BW slightly increases as the real part of the refractive index decreases, our choice of the real part of the substrate refractive index is 1.5 since it is characteristic of many commercially available polymers in the THz spectrum, such as polypropylene,36 high-density polyethylene,36 polyethylene cyclic olefin copolymer.37 On the other hand, the absorption spectrum is mainly independent of the spacer dielectric losses, hence supporting the practical implementation and widening the possible choice of spacer materials.
To examine the physical concept of the absorption mechanism and theoretically validate our results, we adopt interference theory based on the modified Fabry–Perot resonant model. In this theory, a metamaterial absorber is represented as a Fabry–Perot resonator, where the ground reflector serves as a perfect mirror and the top interface of strips is a partially reflective mirror.38 The schematic of the multiple reflections model within a metamaterial absorber is illustrated in Fig. 4(a). The plane electromagnetic wave impinging on the top interface of the metamaterial is partially reflected back to the air region and transmitted into the dielectric spacer with corresponding complex coefficients of r11 = |r11|exp(iφ11) and t21 = |t21|exp(iφ21). The latter waves are then totally reflected from the ground reflector and then divide into reflected and transmitted waves upon incidence on the top interface from the spacer side with the transmission coefficient of t12 = |t12|exp(iφ12) and reflection coefficient of r22 = |r22|exp(iφ22). Here φij represent the corresponding phase shift. The propagation phase is β = kh, where k is the wavenumber in the spacer and h is the path length of the beam inside the spacer. For a normal incidence h = td. According to the wave propagation theory in multilayer structures the total reflection coefficient of the layered system is determined as:38
![]() | (9) |
![]() | ||
Fig. 4 (a) The schematic of the multiple reflections model within a metamaterial absorber. (b) Comparison of the absorptance of the proposed MMA derived via numerical simulations (black) and calculated by the interference theory, eqn (11) (red-rings). (c) The real and imaginary parts of the relative effective wave impedance of the absorber (left axis) and absorptance spectrum (right axis). The highlighted region corresponds to the efficient absorption spectrum. |
Assuming that the reflection coefficient from the ground plane r23 = −1 and applying some mathematical manipulations with the interaction number m ≫ 1, the total reflection coefficient can be simplified to
![]() | (10) |
The absorption hence is conditioned by the destructive interference between all r11 and t12 waves (multiple transmissions from spacer–strips interface to air). Since the transmittance from the ground metal layer T = 0, the total absorptance of the metamaterial is defined as
A(ω) = 1 − R(ω) = 1 − |rtotal|2. | (11) |
The complex reflection/transmission coefficients used in the interference theory were calculated via the so-called decoupled model.38 The spectrum of the absorptance of the MMA calculated by eqn (7) of the interference theory and its comparison to that of obtained by simulation is shown in Fig. 4(b) by red-rings and solid black lines, respectively. One sees that the numerically and theoretically obtained results are in good agreement.
According to the impedance matching theory, a metamaterial structure perfectly absorbs the incident electromagnetic radiation when its complex effective wave impedance is matched to that of the wave impedance of the environment.31,38 Hence, when the total reflectance R(ω) of the structure approaches zero (A(ω) approaches 1), the wave impedance tends to match that of the free space. The relative impedance can be calculated by the following expression
![]() | (12) |
Two maxima are observed in the efficient absorption band at 0.8 THz and 1.8 THz both having absorptance peak values of ∼0.94. Shown by a blue line in Fig. 5(a) is the absorption spectrum of a system with Au strips, where a 4.56 × 107 S m−1 value for the THz conductivity was used.39 No broadband absorption is observed for highly conductive strip material but only several absorption peaks are present with the highest one at around 2.55 THz and a peak value of about 0.32. The dependence of the absorptance on the conductivity of VO2 (bottom axis, black) and temperature (top axis, red) are plotted in Fig. 5(b) at frequencies of 0.8 THz and 1.8 THz by dashed and solid lines, respectively. At both frequencies, the absorptance sharply increases by increasing the conductivity and starting from 1.14 × 105 S m−1 at 0.8 THz and 1.42 × 105 S m−1 at 1.8 THz the absorptance exceeds 90%. Since the conductivity of VO2 is proportional to the temperature, analogous dependence is also expected when the temperature is varied. Effective absorption is provided at temperatures above 336.9 and 335.6 K at frequencies of 0.8 and 1.8 THz, correspondingly. Thus, the absorptance can be effectively changed from 0.03 at (0.05) at 325 K to 0.95 (0.94) at 340 K when the frequency of the incident radiation is 0.8 THz (1.8 THz). This fact opens up the possibility to realize modulation of the amplitude of THz radiation. Moreover, the modulation depth can be tuned by controlling the temperature of the MMA structure.
In the above analysis, we chose an incident wave having a polarization parallel to the longest dimension of the strip to ensure efficient absorption. On the other hand, rotating the polarization by 90° in the lateral plane produces no absorption and all the electromagnetic waves are reflected. In other words, due to structural asymmetry, the absorption performance is sensitive to the incident field polarization. Polarization-insensitive absorption characteristics can be ensured by adding an additional parallel layer with strips oriented perpendicular to the first layer elements. The schematic of a super-cell of such an absorber and its absorption spectra under x- and y-polarized plane waves are shown in Fig. 6(a) and a two-dimensional map of the absorptance as a function of the frequency and polarization angle at normal incidence is shown in Fig. 6(b). Since the second layer of strips is closer to the Au reflector and is completely embedded in the dielectric spacer, a small difference is seen between absorption spectra of x- and y-polarized plane waves, as follows from Fig. 6(a). One sees that an efficient absorption band with a constant bandwidth is achieved for the entire range of polarization angles from 0° up to 90° [Fig. 6(b)]. The total optimized sizes of a supercell are td = 36 μm, px = 6 μm, py = 114 μm, dz = 200 nm, dx = 1 μm and dy = 94 μm, and the distance between two layers of strips is 3 μm. Finally, the performance of the polarization-insensitive design of the absorber under oblique incident TE and TM waves is shown in Fig. 7(a) and (b), respectively, with the corresponding illustrations of orientations of the electromagnetic field components.
![]() | ||
Fig. 7 Absorptance spectrum of the structure as a function of the incidence angle α in the case of (a) TE-polarized waves and (b) TM-polarized waves with the corresponding schematic illustrations. |
Under the oblique incident TE-polarized waves the system preserves its broadband absorption performance for angles of incidence up to 40°, and the further increase in α gradually decreases the absorptance. Under TM polarized waves a continuous efficient absorption band is observed at incidence angles up to 55°.
Table 1 compares the performance of the proposed MMA with several structures reported in the literature which work in the same spectrum. The comparison criteria are 90% absorption band, BW, RBW, unit cell sizes, materials used and structural layers, as well as a short description on the unit-cell configurations used.
Ref. | 90% absorption band (BW) in THz | RBW (%) | Unit cell sizes (μm3) | Materials (layers) | Structure |
---|---|---|---|---|---|
40 | 0.39–0.72 (0.33) | 60 | 174 × 174 × 70.52 | Au, SiO2, VO2 (4) | Metallic cross resonators on SiO2–VO2–SiO2 planar structure |
41 | 0.457–1 (0.543) | 74.5 | 150 × 150 × 50.4 | Aluminium, polyimide (3) | Two symmetrical L-shaped metallic resonators |
20 | 1.58–2.22 (0.642) | 33.7 | 120 × 120 × 20.8 | Au, polyimide (5) | Two layers of four different sized square metallic patches |
22 | 0.61–1.36 (0.75) | 76.1 | 120 × 120 × 66 | SiO2, VO2, Au (3) | Ring and cross-like resonators with different sizes combined in a unit-cell and embedded in a dielectric layer |
42 | 1.33–2.43 (1.1) | 58.5 | 100 × 100 × 13.2 | Au, SiO2 (3) | Two double square split-ring resonators combined with three double circular split-ring resonators with different orientations |
This paper | 0.66–1.84 (1.18) | 94.4 | 114 × 114 × 39.2 | VO2, polymer, Au (4) | Rectangular strips |
This journal is © The Royal Society of Chemistry 2023 |