betaFeedback

WorksPhysics and Astronomy

Optimal Design of Dallenbach Absorbers Under Broadband Broad-Angle Illumination

Abstract

The classical scenario where a single plane-wave field impinge a Dallenbach absorber is well studied both theoretically and experimentally. However, occasionally a spectrum of plane-waves impinges the absorber. Such a scenario occurs for example if an antenna is located adjacent to the absorbing layer. In this paper, for this scenario we obtain the absorbing performance bound and design an optimized layered absorber that approaches the bound. In a numerical demonstration, we explore a realistic case where a dipole antenna is placed in the vicinity of a finite, electrically thin, Dallenbach absorber backed by a PEC plane in the 6G frequency range. In the absence of the absorbing layer covering the PEC plane, severe scattering from the plane distorts the radiated fields. These distortions are robustly mitigated by the specifically tailored optimal absorber to yield a more desired radiation pattern. Additionally, we propose a metamaterial realization that emulates the required properties of the absorbing layer for all field polarizations.

Introduction

Wave absorbers play an important role in wave engineering [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], as they are being used in a variety of electromagnetic applications such as reducing reflections [12, 13, 14, 15], enhanced accuracy of antenna measurements in anechoic chambers [16, 17, 18], decreasing undesired emissions of discrete electric components and printed circuit boards (PCBs) [19, 20, 21] and to capture light in solar cells [22, 23], to name a few. Among the different types of wave absorbers, Dallenbach layer, a lossy substance backed by a perfect electric conductor (PEC) sheet, is widely used due to its compatibility with planar geometries [1]. For the case of a normal plane wave incidence a known sum rule [24] imposes a constraint between the absorption efficacy, thickness and the operating frequency band. This bound is unavoidably affected by the direction of arrival of the impinging wave and by its polarization. Thus, extensions for oblique incidence have been derived [25].

Occasionally the impinging wave is more complicated spectrally, involving a spectrum of plane waves, arriving at different directions. This is the case, for instance when a source is located adjacent to the absorbing layer, in that case the spectral content comprising the field is much broader than in the case of a single plane wave incidence, which is applicable in strictly far-field scenarios. In this case, the design problem should be formulated by taking into account the relative contribution to the absorption, i.e. “absorption significance,” of different spectral components. The need to tailor properly the absorption for different spectral components, implies that the absorber has to have a suitably designed non-local behaviour. This can naturally be achieved in a planar layered media [26].

In the following, we first extend the canonical single plane wave absorber bound to the case of spectrally wide field, and include the absorption significance merit in the derived bound. This will be used as our design benchmark. Later, we verify numerically that the new bound is tight when integrating along the entire frequency bandwidth. Lastly, we set an optimization problem to obtain the optimal design parameters, namely the permittivity and electric conductivity σ\sigma of the layered Dallenbach absorber. While being restricted to the new bound, the optimal solution maximizes the contribution within the desired bandwidth while simultaneously minimizing it outside. Our approach’s applicability is demonstrated through numerical examples involving a half-wavelength dipole antenna positioned above a large, but finite, ideal ground plane (perfectly electric conductor - PEC). This configuration leads to the emergence of severe interference pattern in the radiated field, resulting in some extreme gain variations both frequency and angle dependent that may restrict the use of such physical layout. In order to solve this issue and suppress undesired reflections caused by the ground plane, we design a layered Dallenbach absorber in the 6G frequency range, f∈[7,15]f \in[7,15] GHz by taking into account the absorption significance of the antenna’s radiation pattern. Specifically we consider two distinct cases when the dipole is placed horizontally, (a) the hight of dipole antenna above the plane is larger than a wavelength and therefore the radiation pattern suffers from the emergence of grating lobes; while in the second case (b) the dipole antenna is positioned at a hight that is much smaller than a wavelength above the plane and therefore the radiation pattern is not uniform which causes severe gain variations. Additionally, to investigate the polarization dependence of the designed absorber, as a third example (c) we replaced the horizontal dipole in case (a) with a vertical dipole. By designing a specifically optimized absorber, we demonstrate that the severe distortions in the radiation pattern, as seen in cases (a), (b) or (c), can be effectively mitigated, resulting in a more desirable radiation pattern. Subsequently, we discuss some metamaterial realization aspects for the proposed absorbers that are applied for fabrication with PCB technology.

Mathematical Formulation

Consider a Dallenbach planar layer consists of a PEC sheet covered by layered dielectric absorber with electrical permittivity (ϵ\epsilon), magnetic permeability (μ\mu) and electrical conductivity (σ\sigma) and surrounded by vacuum (ϵ0\epsilon_0, μ0\mu_0). The structure extends to infinitely in the xx and yy directions and has a finite thickness dd in the zz direction. The absorber is impinged by a wave field that is composed of an angular spectrum of propagating oblique incidence plane waves with either transverse electric (TE) or magnetic (TM) polarizations. See Figure 1 for illustration. Due to the interaction with the Dallenbach layer, some of the wave field is reflected, while the rest is absorbed within the layer. Note that the PEC boundary enforces that there is no transmitted wave beyond the absorbing layer. Assuming that a single, oblique incidence plane wave with angle θ\theta impinging the absorber, the following sum-rule is known to bound the absorption efficiency [24, 25],

Figure 1. Illustration of the problem. A Dallenbach layer with thickness dd is impinged by a wave field that is composed of multiple propagating plane waves within the angular domain θ∈[θ1,θ2]\theta\in[\theta_1,\theta_2]. As a result of the interaction with the Dallenbach layer, a portion of the wave field is reflected backwards, while the remaining portion is absorbed inside the layer. The wave field may consist of either one or both of the TE and TM polarizations.

∣∫0∞ln⁡∣ρ(λ,θ)∣ dλ∣≤2π2μsd{cos⁡(θ),TE1/cos⁡(θ),TM(1)\left|\int_{0}^{\infty}\ln|\rho(\lambda,\theta)|\,d\lambda\right| \leq2\pi^{2}\mu_s d \begin{cases} \cos(\theta), & \text{TE} \\ 1/\cos(\theta), & \text{TM} \end{cases} \tag*{(1)}

where ρ\rho is the reflection coefficient at the interface between the Dallenbach absorber and its surroundings, λ\lambda is the free space wavelength and μs\mu_s is the static relative permeability. In the following, we shall denote the right-hand-side of Eq. (1) by RB(θ)R_B(\theta). The expression in Eq. (1) clearly illustrates the strong dependence of the sum-rule bound on the field polarization as well as on the direction of arrival of the impinging wave. Therefore, it is clear that Eq. (1) cannot be directly applied for fields comprising more than a single plane wave, i.e. with a spectrum of plane waves, such as one expects in the presence of a finite source adjacent to the absorber. In order to accommodate for the extended spectral content we propose the following: First, we multiply both sides of Eq. (1) by a normalized non-negative weight function W(θ)W(\theta) [27] that denotes for the “absorption significance” of the spectral components. Next, we integrate with respect to the angle of incidence (θ\theta). In addition, to accommodate for real-world applications where systems are typically designed to operate at a specific finite frequency bandwidth, we truncate the originally infinite integration wavelength domain into a finite domain [λ1,λ2][\lambda_1,\lambda_2], where λ1\lambda_1 [λ2][\lambda_2] denote the minimal [maximal][\text{maximal}] free-space wavelength of interest. These steps lead to,

∫θ1θ2W(θ)∣∫λ1λ2ln⁡∣ρ(λ,θ)∣ dλ∣dθ≤\int_{\theta_1}^{\theta_2} W(\theta)\left|\int_{\lambda_1}^{\lambda_2}\ln|\rho(\lambda,\theta)|\,d\lambda\right|d\theta\leq
∫θ1θ2W(θ)∣∫0∞ln⁡∣ρ(λ,θ)∣ dλ∣dθ≤∫θ1θ2W(θ)RB(θ) dθ,(2)\int_{\theta_1}^{\theta_2} W(\theta)\left|\int_{0}^{\infty}\ln|\rho(\lambda,\theta)|\,d\lambda\right|d\theta\leq\int_{\theta_1}^{\theta_2}W(\theta)R_B(\theta)\,d\theta, \tag*{(2)}

where [θ1,θ2][\theta_1,\theta_2] are the minimal and maximal desired angles of incidence, respectively and ∫θ1θ2W(θ)dθ=1\int_{\theta_1}^{\theta_2}W(\theta)d\theta=1, W(θ)≠0 ∀ θ∈[θ1,θ2]W(\theta)\ne0\ \forall\ \theta\in[\theta_1,\theta_2]. The expression in Eq. (2) describes the appropriate weighted sum rule (bound) for wave fields composed of a spectrum (multiple) of oblique incidence plane waves. The right hand side of Eq. (2) bounds from above the left and middle sides for any selection of the properties of the layer parameters {ϵ,σ}\{\epsilon,\sigma\} that are embedded in ρ\rho. This gives rise to an optimization problem for designing a Dallenbach absorber. By employing Eq. (2) as a WW-weighted L1L^1 norm on θ\theta space, we formulate the following optimization problem to obtain the absorber’s characteristic parameters,

min⁡ϵ,σ ∫θ1θ2∣W(θ)RB−W(θ)∣∫λ1λ2ln⁡∣ρ(λ,θ)∣ dλ∣∣dθ(3)\underset{\epsilon,\sigma}{\min}\ \int_{\theta_1}^{\theta_2}\left|W(\theta)R_B-W(\theta)\left|\int_{\lambda_1}^{\lambda_2}\ln|\rho(\lambda,\theta)|\,d\lambda\right|\right|d\theta \tag*{(3)}

subject to ρ(λ,θ)=Zin(λ,θ)−η0(θ)Zin(λ,θ)+η0(θ)\rho(\lambda,\theta)=\frac{Z_{\mathrm{in}}(\lambda,\theta)-\eta_0(\theta)}{Z_{\mathrm{in}}(\lambda,\theta)+\eta_0(\theta)}.

The impedance ZinZ_{\mathrm{in}} refers to the input impedance at the interface between the absorbing layer and its surroundings, while η0\eta_0 is the free space (TE and TM) characteristic impedance (see [28, 29, 30]). An alternative (equivalent) way to express Eq. (3) as a maximization problem is to normalize the cost function by dividing Eq. (2) by its right-hand side. The resulting value of the expression is between 0 and 1, where a score of 1 indicates the highest achievable value of the cost function, i.e., tightest bound, for any given set of design parameters (ϵ,σ\epsilon,\sigma). The alternative optimization problem reads,

max⁡ϵ,σ ∫θ1θ2W(θ)∣∫λ1λ2ln⁡∣ρ(λ,θ)∣ dλ∣dθ∫θ1θ2W(θ)RB(θ) dθ(4)\underset{\epsilon,\sigma}{\max}\ \frac{\int_{\theta_1}^{\theta_2}W(\theta)\left|\int_{\lambda_1}^{\lambda_2}\ln|\rho(\lambda,\theta)|\,d\lambda\right|d\theta}{\int_{\theta_1}^{\theta_2}W(\theta)R_B(\theta)\,d\theta} \tag*{(4)}

subject to ρ(λ,θ)=Zin(λ,θ)−η0(θ)Zin(λ,θ)+η0(θ)\rho(\lambda,\theta)=\frac{Z_{\mathrm{in}}(\lambda,\theta)-\eta_0(\theta)}{Z_{\mathrm{in}}(\lambda,\theta)+\eta_0(\theta)}.

In the following sections we use Eqs. (2)–(4) to explore the tightness of the new sum rule and to design a practical Dallenbach absorber that operates within a finite but broad frequency range and wide angular spectrum of incidence waves.

The Tightness of the Sum Rule

For the sake of concreteness, here, we consider the case of a wave field with an infinite frequency support within the angular domain [θ1,θ2]=[0,π/3][\theta_1,\theta_2] = [0,\pi/3]. For that purpose we consider a non-magnetic layer (μ=μ0\mu= \mu_0, μs=1\mu_s = 1) with thickness d=0.4d = 0.4 [m]. The layer is composed of causal materials satisfying Kramers-Kroning relations, with wavelength dependent permittivity ϵ(λ)\epsilon(\lambda) and electric conductivity σ(λ)\sigma(\lambda) having a Lorentzian form [31],

ϵ(λ)=ϵ0(1+A1+jλrelλ−(λresλ)2)(5)\epsilon(\lambda) = \epsilon_0 \left(1+\frac{A}{1+j\frac{\lambda_{\mathrm{rel}}}{\lambda}-\left(\frac{\lambda_{\mathrm{res}}}{\lambda}\right)^2}\right) \tag*{(5)}
σ(λ)=σs1+j2πc0τσλ,\sigma(\lambda) = \frac{\sigma_s}{1+j\frac{2\pi c_0\tau_\sigma}{\lambda}},

where AA denotes the strength of the dielectric response, λres[λrel]\lambda_{\mathrm{res}}[\lambda_{\mathrm{rel}}] is the resonant [relaxation] wavelength, σs\sigma_s is the static (long wavelength) electric conductivity, c0=1/ϵ0μ0c_0 = 1/\sqrt{\epsilon_0\mu_0} is the speed of light in vacuum and τσ\tau_\sigma is a time constant of the conductor. By combining Ampere’s law with Ohm’s law and Eq. (5), an effective electric permittivity, ϵeff(λ)\epsilon_{\mathrm{eff}}(\lambda), can be expressed. This effective permittivity takes into account both material polarization and conductance effects and is given by

ϵeff(λ)=ϵ(λ)+σ(λ)j2πc0λ−1.(6)\epsilon_{\mathrm{eff}}(\lambda) = \epsilon(\lambda) + \frac{\sigma(\lambda)}{j2\pi c_0\lambda^{-1}}. \tag*{(6)}

For the sake of concreteness, we set the following parameters: A=1A = 1, λres=λrel=1\lambda_{\mathrm{res}} = \lambda_{\mathrm{rel}} = 1 [m], τσ=10−12\tau_\sigma= 10^{-12} [sec], while the only remaining free-parameter is σs\sigma_s that will be scanned over a wide range of values to demonstrate its effect on the tightness of the bound. In addition, we define three typical weight functions {W1(θ),W2(θ),W3(θ)}\{W_1(\theta), W_2(\theta), W_3(\theta)\} within the angular domain θ∈[0,π/3]\theta\in[0,\pi/3] (see Figure 2 (a)),

Figure 2. Tightness of the bound for several weight functions. In accordance with Eq. (2), the ratio is bounded from above by 1 and is non-negative. The ratio is calculated for a wide range of σs\sigma_s, while holding all other parameters in Eq. (5) constant. (a) The three weight functions. (b) TE polarization. (c) TM polarization.

W1(θ)=3π,W_1(\theta) = \frac{3}{\pi},
W2(θ)=4.5496 θ5(7)W_2(\theta) = 4.5496\,\theta^5 \tag*{(7)}
W3(θ)=5.015 e−10∣θ−π/6∣,W_3(\theta) = 5.015\,e^{-10\lvert\theta-\pi/6\rvert},

and zero elsewhere. The three weight functions (recall, these enforce the “absorption significance” of each spectral component) were selected to have specific characteristics. Firstly, W1(θ)W_1(\theta) is uniformly distributed, indicating that all angles hold equal importance. Secondly, W2(θ)W_2(\theta) prioritizes larger angles, where extreme values are observed in the sum rule (see Eqs. (1) and (2)). Lastly, W3(θ)W_3(\theta) exhibits symmetry around θ=π/6\theta= \pi/6 and follows an exponential distribution. We define the tightness of the bound as the ratio in the argument of the optimization problem in Eq. (4),

T=∫θ1θ2Wi(θ)∣∫0∞ln⁡∣ρ(λ,θ)∣ dλ∣ dθ∫θ1θ2Wi(θ)RB⁡(θ) dθ,0≤T≤1.(8)T = \frac{\int_{\theta_1}^{\theta_2} W_i(\theta)\left|\int_0^\infty\ln\left|\rho(\lambda,\theta)\right|\,d\lambda\right|\,d\theta}{\int_{\theta_1}^{\theta_2} W_i(\theta)\operatorname{RB}(\theta)\,d\theta}, \qquad0 \leq T \leq1. \tag*{(8)}

Figures 2 (b) and (c) depict the evaluation of TT for multiple values of σs\sigma_s and for each Wi(θ)W_i(\theta) with i∈1,2,3i \in1,2,3 and for both TE and TM polarizations. The integral in the numerator should be performed over an infinite wavelength range. This is obviously impossible numerically. Instead, the integration is performed over an extended but finite wavelength range λ∈[0.001,104]\lambda\in[0.001,10^4] [m] and converges to the infinite integration result is verified. The evaluation of TT in the numerator uses the exact value of the reflection coefficient, ρ(λ,θ)\rho(\lambda,\theta), that was calculated using the expression given in Eqs. (3)–(4) where the impedance ZinZ_{\mathrm{in}} is given by the transmission-line model of the layered media [28, 29]. A tight bound with values approaching 1, as shown in Figure 2(b,Figure 2c), can be observed for any of the three weight functions considered, provided that the free parameter σs\sigma_s is properly selected. It is important to note that the scanning process for finding optimal parameters can be further fine-tuned by including additional free-parameters, such as AA, λres\lambda_{\mathrm{res}}, λrel\lambda_{\mathrm{rel}}, and τσ\tau_\sigma or even to incorporate multiple resonance terms in the Lorentzian model. By exploring different combinations of these parameters, it is possible to identify alternative sets that may also maximize the ratio.

Optimal Absorber Design

In practical systems, absorption performance is typically optimized within a specific bandwidth λ∈[λ1,λ2]\lambda\in[\lambda_1,\lambda_2], λ1>0\lambda_1 > 0 and λ2<∞\lambda_2 < \infty. In light of the absorption efficiency sum-rule, it is clear that to maximize absorption within a certain frequency range it is essential to minimize it elsewhere. The optimization approach developed here is formulated to achieve this goal. Because of the design’s field polarization sensitivity, the optimization needs to be conducted separately for TE and TM polarizations. The optimization process that we consider allows also for the design of a stratified inhomogeneous Dallenbach layers. In that case, we consider a piecewise homogeneous design, where, the layer of thickness dd is truncated into NN parallel sub-layers with identical thickness, each of which is homogeneous (see Figure 3 for illustration). For any value of NN we aim to find optimal design parameters (ϵ,σ)(\epsilon,\sigma) that maximizes the tightness, TT, in Eq. (8). For each sub-layer, we consider the following two models for the design parameters:

A piecewise homogeneous design where the non-magnetic absorbing layer is truncated into N parallel sub-layers

Figure 3. A piecewise homogeneous design where the non-magnetic absorbing layer of thickness dd is truncated into NN parallel sub-layers with identical thickness di=d/Nd_i=d/N (i∈[1,N]i\in[1,N]), each of which is homogeneous with ϵi,σi\epsilon_i,\sigma_i.

  • Model 1 - constant parameters. We use Eq. (5) with λrel=λres≈0\lambda_{\mathrm{rel}}=\lambda_{\mathrm{res}}\approx0 and 2πc0τσλ≪1\frac{2\pi c_0\tau\sigma}{\lambda}\ll1, resulting in an approximately constant ϵ=ϵ0ϵr\epsilon=\epsilon_0\epsilon_r and σ=σs\sigma=\sigma_s, where ϵr=1+A\epsilon_r=1+A is the relative permittivity. In this model, there are two design parameters, ϵr\epsilon_r and σ\sigma, which provide only a conductive loss mechanism.

  • Model 2 - frequency dependent parameters. We use Eq. (5) with ϵ(λ)\epsilon(\lambda) providing three design parameters, AA, λrel\lambda_{\mathrm{rel}}, and λres\lambda_{\mathrm{res}}. To reduce the number of design parameters, it is convenient to set σs=0\sigma_s=0, providing only a polarization loss mechanism.

Model 1 is widely used in microwave engineering, such as for example in transmission-lines where the parameters usually vary slowly within the desired frequency band to maintain wideband impedance matching. In this model the material resonances are typically located at much higher frequencies than these of interest, as opposed to Model 2 where they are within (or near the edge) of the operational region. Both models satisfy Kramers-Kronig relation over the entire infinite frequency range. Note that other models, such as incorporating electric conductivity σ(λ)\sigma(\lambda) in Model 2 or incorporating multi-resonance polarization response, may also be applicable. However, as the complexity of the model increases, with additional design parameters per sub-layer, the optimization process becomes more involved while the relative improvement decreases. The

optimization problem to obtain the absorber’s parameters giving the tightest design is non-convex. To solve this optimization problem, we adopt a four-level search approach. First forming a relatively coarse calculation grid. In Model 1 realization, the parameters ϵr\epsilon_r and σ\sigma of each layer are uniformly distributed with MM points in the ranges ϵr,min⁡≤ϵr≤ϵr,max⁡\epsilon_{r,\min}\leq\epsilon_r\leq\epsilon_{r,\max} and 0≤σ≤σmax⁡0\leq\sigma\leq\sigma_{\max} [S/m], for some parameters {ϵr,min⁡,ϵr,max⁡,σmax⁡}\{\epsilon_{r,\min},\epsilon_{r,\max},\sigma_{\max}\}. In Model 2 realization, the parameters of each layer AA, λrel\lambda_{\mathrm{rel}} and λres\lambda_{\mathrm{res}} are uniformly distributed with MM points in the ranges Amin⁡≤A≤Amax⁡A_{\min}\leq A\leq A_{\max} and λmin⁡≤λrel,λres≤λmax⁡\lambda_{\min}\leq\lambda_{\mathrm{rel}},\lambda_{\mathrm{res}}\leq\lambda_{\max} [m] for some parameters {Amin⁡,Amax⁡,λmin⁡,λmax⁡}\{A_{\min},A_{\max},\lambda_{\min},\lambda_{\max}\}. Note that the lower and upper values of the grid may vary based on the requirements, such as frequency bandwidth, angular domain, etc. Overall, with the parameters of NN sub-layers, the total number of calculation grid points in Model 1 is M2NM^{2N}, and for Model 2 is M3NM^{3N}. To ensure memory limitations are not exceeded, for a given NN, we choose MM such to limit the number of grid points to no more than 500 million. For each calculation grid point, in the first calculation level, we evaluate the cost function in Eq. (3) (or its equivalent in Eq. (4)). In the second step, we focus on the top-performing results from the first-level, coarse grid, calculation step, limiting our search to no more than 1% of the total number of grid points. To refine our search further, we perform a local search around each candidate solution of the first-level, evaluating the cost function at steps that are smaller by factor 10 than in the first-level. Once we have explored possible refinement in the vicinity of the top-performing results of the previous step, the most optimal solution obtained so far is set as an input to the third step. In the third step, we employ a subsequent round of local search, which involves exploring the immediate surroundings of the optimal solution using smaller step sizes, specifically reduced by a factor of 25 compared to the initial step. This process allows us to fine-tune the optimal solution and achieve even greater refinement. As a final step we perform the optimization procedure described in steps 11–33 three additional times, adjusting arbitrarily the parameters {εr,min⁡,εr,max⁡,σmax⁡,Amin⁡,Amax⁡,λmin⁡,λmax⁡}\{\varepsilon_{r,\min}, \varepsilon_{r,\max}, \sigma_{\max}, A_{\min}, A_{\max}, \lambda_{\min}, \lambda_{\max}\} in the first step. This step aims to improve the optimized results, as the initial grid points are modified. The entire process is done for both TE and TM polarizations and for several values of NN. In the following section, we illustrate the design procedure for a metamaterial structure to emulate the necessary effective parameters derived from the optimization process. For this purpose, employing Model 1 is more practical due to its more feasible requirements in the quasistatic regime.

Practical numerical example - metamaterial absorber for 6G applications

In this section we demonstrate by numerical examples how the absorber’s design introduced in the previous sections helps to overcome fundamental issues with antennas that are placed in the vicinity of a large but finite reflective surfaces (modelled as PEC ground plane).

Horizontal dipole above an optimized Dallenbach layer

We start by noting that the H-plane radiation pattern of a half-wavelength dipole which is inherently omnidirectional in free space, exhibits variable radiation performance when parallel to a large ground plane [32]. This comes due to a reflected field from the ground plane that can be interpreted as a contribution of an image source that is anti-symmetrically located beyond the PEC. Note that this intuitive description neglects edge diffraction effects resulting from the finite nature of the ground plane.

In the first example, the dipole is positioned at h=1.25λmax⁡h = 1.25\lambda_{\max} above a circular finite ground plane of radius of R=5λmax⁡R = 5\lambda_{\max} (λmax⁡=0.0429\lambda_{\max} = 0.0429[m], corresponding to the free space wavelength at f=7f = 7[GHz]). The non-negligible distance however >λ>\lambda (in the band) between the dipole and the ground plane induces the presence of grating lobes that severely distort the radiation field by introducing nulls in the field radiation pattern. To address this issue and mitigate the impact of the distorting reflected wave, we aim to optimize a Dallenbach absorber with a thickness of d=0.005d = 0.005[m] in the 6G frequency range f∈[7,15]f \in[7,15][GHz]. Physical layout of the system is depicted in Figure 4(a)–(Figure 4b), showing side and top views, respectively.

Figure 4. Problem illustration. A half-wavelength dipole is placed above, in parallel to a Dallenbach absorber (h=0.0536h = 0.0536[m], corresponding to 1.25λmax⁡1.25\lambda_{\max}). (a) Side view. (b) Top view. The gray rectangle illustrates a lumped port that is set to excite the structure. (c) and (d) present the H-plane realized gain obtained from HFSS for f=7f = 7GHz and f=11f = 11GHz, respectively. Radiation in free space (blue), the radiating element in the presence of an infinite ground plane (dotted green) and finite ground plane (dark red), the element in the presence of an optimal Dallenbach absorber with N=1N = 1 (black), and N=2N = 2 (dotted purple).

It is important to note that as the frequency increases, more grating lobes emerge while the distance hh between the radiating element and the ground plane remains constant. We use a cosine weight function, W(θ)=cos⁡(θ)W(\theta) = \cos(\theta), over θ∈[0,π/2]\theta\in[0,\pi/2] to denote the absorption significance. Given the predominant TE polarization aligned with the wire in the far-field of the half-wavelength dipole antenna, we apply the optimization approach, detailed in Sec. IV with one or two absorbing layers (N=1,2N = 1,2) to determine optimal values of εr\varepsilon_r and σ\sigma for suppressing TE polarized reflected waves. The optimized values are : εr=2.807\varepsilon_r = 2.807, σ=0.996\sigma= 0.996[S/m] for N=1N = 1 (resulting in a score of 0.4213 in Eq. (4)), and εr={3.7627,5.7119}\varepsilon_r = \{3.7627,5.7119\}, σ={0,5.2881}\sigma= \{0,5.2881\}[S/m] for N=2N = 2 (yielding a score of 0.4855 in Eq. (4)). These results were obtained with the grid constraints εr,min⁡=1\varepsilon_{r,\min} = 1, εr,max⁡=5\varepsilon_{r,\max} = 5, σmax⁡=1\sigma_{\max} = 1[S/m] (for N=1N = 1), and εr,min⁡=2\varepsilon_{r,\min} = 2, εr,max⁡=7\varepsilon_{r,\max} = 7, σmax⁡=6\sigma_{\max} = 6[S/m] (for $N = 2). To assess the system’s performance, both with and without the absorbing layer, a full-wave electromagnetic simulation was carried out using the High-Frequency Structure Simulator (HFSS) [33]. A lumped port was employed to excite the half-wavelength dipole. Given the narrow frequency band nature of the radiating element which is in contrast with the extended range of the 6G frequency band, a scaled redesign was applied to match the element’s length (ll) with the corresponding free space wavelength (l∼λ/2l \sim\lambda/2) at each frequency. This involved varying the length of the radiating element to maintain the desired proportional relationship with the frequency. Figure 4 (c) and (Figure 4d) display the numerically obtained realized gain (H-plane) from HFSS for f=7f = 7GHz and f=11f = 11GHz, respectively. Each figure features 5 lines representing the realized gain for different scenarios: free space radiation of the half-wavelength dipole only (blue), the dipole in the presence of an infinite ground plane (dotted green), the dipole in the presence of a finite ground plane (dark red), the dipole in the presence of an optimal Dallenbach absorber with N=1N = 1 (black), and N=2N = 2 (dotted purple). As seen in Figure 4 (c) and (Figure 4d), the lack of an absorbing layer results in multiple angles where the realized gain is nullified. This is due to destructive interference between the direct ray and the reflections caused by the finite ground plane. In contrast, as we introduce the electrically thin optimized Dallenbach absorber, the severe effect of the reflected waves from the ground plane, causing the nulls, is mitigated in such a manner as to nearly eliminating any trace of the nulls at higher frequencies further into the band to give antenna’s performance that is closer to what is expected in free space. To explore and assess the effect of the absorber on the dipole’s radiation pattern, we introduce a flatness parameter (FF) as follows,

F≜[12π∫0π/2[G(θ)GFSD(θ)+GFSD(θ)G(θ)]2 dθ]12,F≥1(9)F \triangleq\left[\frac{1}{2\pi}\int_{0}^{\pi/2}\left[\frac{G(\theta)}{G_{\mathrm{FSD}}(\theta)}+\frac{G_{\mathrm{FSD}}(\theta)}{G(\theta)}\right]^2\,d\theta\right]^{\frac{1}{2}}, \qquad F \ge1 \tag*{(9)}

where GFSDG_{\mathrm{FSD}} is the realized gain of the free-space dipole and GG represents the realized gain under examination for finite GND, infinite GND and optimal absorber designs. Applying Eq. (9) with the simulation results in Figure 4 (c) reveals that without the absorber (finite GND) F=8.359F = 8.359. With the absorber F=1.2977F = 1.2977 for N=1N = 1 and F=1.2528F = 1.2528 for N=2N = 2, indicating an improvement of about 84.7%. Similarly, in Figure 4 (d) without the absorber (finite GND) F=4.5312F = 4.5312. With the absorber F=1.0686F = 1.0686 for N=1N = 1 and F=1.0659F = 1.0659 for N=2N = 2, indicating an improvement of about 76.4%. Similar results to those shown in Figure 4 (c) and (d) were achieved at frequencies of f=9,13,15f = 9, 13, 15[GHz].

In the first example, we examined the case where the radiating element was positioned more than 1.25λ1.25\lambda away from the absorber, which was utilized to demonstrate the reduction of grating lobes that caused nulls to appear in the radiation pattern in specific directions. Next, as a second example, we explore whether the same absorber that was designed in the first example, above, can effectively reduce reflections when the dipole is placed in the immediate vicinity of the surface (h≪λh \ll\lambda). In this case, grating lobes do not occur. However, the gain varies rapidly with θ\theta, which is undesirable for quasi-omnidirectional and quasi-isotropic antennas [34, 35, 36]. A common performance metric for such antennas is “gain variation”, defined as the difference between the maximum and minimum gain the angular band measured in decibels. In our scenario, the large PEC ground plane enforces zero radiation at grazing angles (θ∼π/2\theta\sim\pi/2) and in the blocked lower hemisphere (θ∈[π/2,π]\theta\in[\pi/2, \pi]) and therefore inducing some gain variation in upper radiation hemisphere. An efficient wave absorber may significantly reduce the gain variation in the upper hemisphere. This is not a trivial task at all, as omni-directional and isotropic antennas are typically small compared to the operating wavelength and their radiation pattern is sensitive when placed in the vicinity of large metallic reflectors. To that end, Figure 5(a) depicts a typical normalized gain radiation pattern (in db) at f=13f = 13GHz with h=0.007h = 0.007[m], corresponding to h/λ=0.3033h/\lambda= 0.3033 for the five cases (as in Figure 4) similar patterns are obtained for different frequencies in the band. It should be noted that due to the close proximity of the antenna to the finite ground plane (dark red line) and the strong electromagnetic coupling, the maximum of the gain is shifted from the boresight (z axis) direction and increase the gain variation. Figure 5(b) depicts the gain variation in the upper radiation hemisphere at the different frequencies in the 6G band. It is readily noted that for both absorbers realizations (N=1,2N = 1, 2, solid-black and dotted-purple lines, respectively) the gain variation is smaller by at least 4 [db] along nearly all the 6G band with respect to the finite ground plane with no absorber case (solid dark red line). Note that smaller gain variation imply for relatively more uniform angular pattern, therefore the use of a Dallenbach absorber is favorable as it provide better results as to the no-absorber case. It should be emphasized that in a free-space case, where the pattern is omni-directional, the gain variation is 0 [db]. This is the most desired case

Figure 5. (a) A typical normalized gain radiation pattern (in dB) at f=13GHzf = 13\mathrm{GHz} with h=0.007[m]h = 0.007[\mathrm{m}], corresponding to h/λ=0.3033h/\lambda= 0.3033 for the five cases (as in Figure 4). (b) depicts the gain variation in the upper radiation hemisphere at the different frequencies in the 6G band. It is readily noted the for both absorbers’ realizations (N=1,2N = 1, 2, solid-black and dotted-purple lines, respectively). In both figures the blue line represents the free space case of the half-wavelength dipole, the dotted green line shows the dipole in the presence of an infinite ground plane, the dark red line indicates the dipole with a finite ground plane, and the black and dotted purple lines depict the dipole in the presence of an optimal Dallenbach absorber with N=1N = 1 and N=2N = 2, respectively.

however due to the practically constrained physical space for the construction of the optimal absorber a complete mitigation of the gain variation is limited however this example indicate that “correction” in the gain variation is indeed feasible.

Vertical dipole above an optimized Dallenbach layer

Here, we aim to verify whether the previously designed absorber can reduce reflections caused by a half-wavelength vertical dipole placed at a height of h=1.25λmax⁡h = 1.25\lambda_{\max} above the same circular finite ground plane. It is important to note that at this stage, no further optimization steps are performed. Instead, we use the previously optimized parameters: ϵr=2.807\epsilon_r = 2.807, σ=0.996\sigma= 0.996[S/m] for N=1N = 1 and ϵr={3.7627,5.7119}\epsilon_r = \{3.7627, 5.7119\}, σ={0.5,0.2881}\sigma= \{0.5, 0.2881\}[S/m] for N=2N = 2. Figures 6 (a) and (b) show the side view and top view of the geometry, respectively. Figure 4 (c) and (d) show the numerically obtained realized gain (Z-X plane) from HFSS at frequencies of f=7f = 7, GHz and f=11f = 11, GHz, respectively. Each figure includes five lines representing the realized gain for various scenarios, as described previously: free-space radiation of the half-wavelength dipole only (blue), the dipole in the presence of an infinite ground plane (dotted green), the dipole in the presence of a finite ground plane (dark red), the dipole in the presence of an optimal Dallenbach absorber with N=1N = 1 (black), and N=2N = 2 (dotted purple). It is evident that the absorber effectively mitigates reflections, as the realized gain in the presence of the absorber closely matches the gain observed when the dipole is placed in free space. To quantify this effect, the FF parameter, defined in Eq. (9), is used. In Figure 4 (c) without the absorber (finite GND) F=53.09F = 53.09. With the absorber F=3.8884F = 3.8884 for N=1N = 1 and F=2.4733F = 2.4733 for N=2N = 2, indicating an improvement of 92.67% and 95.34%, respectively. Similarly, in Figure 4 (d) without the absorber (finite GND) F=78.8501F = 78.8501. With the absorber F=3.1208F = 3.1208 for N=1N = 1 and F=1.9654F = 1.9654 for N=2N = 2, indicating an improvement of 96.04% and 97.5%, respectively. Similar results to those presented in Figures 6 (c) and (d) were obtained at frequencies of f=9,13,15[GHz]f = 9, 13, 15[\mathrm{GHz}].

Metamaterial realization of the optimized layer parameters

Our design, as discussed in the previous section, is aimed for the microwave range where the proper lossy dielectric dispersion is dictated by Model 1 that was introduced in Sec. IV. Thus, we seek to realize the desired complex relative permittivity (see Eq. (6))

ϵdesired(f)=ϵr+σj2πfϵ0.(10)\epsilon_{\mathrm{desired}}(f) = \epsilon_r + \frac{\sigma}{j2\pi f\epsilon_0}. \tag*{(10)}

This design challenge is addressed in this section, where we provide composite metamaterial layer that emulated effectively the desired effective parameters (ϵr,σ)(\epsilon_r,\sigma) obtained through the optimization carried out in the previous section. This design should achieve the desired permittivity profile given in Eq. (10) for TE and TM waves to incorporate with the broad spectral range and the two dipole orientations, horizontal and vertical with respect to the absorbing layer. To achieve this goal, we propose a method for designing a metamaterial structure with minimal frequency and spatial dispersion within the quasistatic regime. We start with a 2D conductive wired metamaterial structure, where the effective permittivity was previously defined in [37]. In our recent work [38], we revisit the calculations outlined in [37], specifically for normal incidence waves. This involves the introduction of a resistor (R′R') connecting the two arms of the wire within each unit cell, embedded in a host material with relative permittivity ϵh\epsilon_h, resulting in an effective relative permittivity ϵanalytical\epsilon_{\mathrm{analytical}},

ϵanalytical(f)=ϵh−1ϵ0(2πf)2a2L~−jϵ0(2πf)a2R~(11)\epsilon_{\mathrm{analytical}}(f) = \epsilon_h - \frac{1}{\epsilon_0(2\pi f)^2a^2\tilde{L}-j\epsilon_0(2\pi f)a^2\tilde{R}} \tag*{(11)}

where a=1[mm]a = 1[\mathrm{mm}] represents the dimension of the cubical unit cell, r0=0.15[mm]r_0 = 0.15[\mathrm{mm}] is the wire radius, L~=μ0/(2π)ln⁡[a2/(4r0(a−r0))]\tilde{L} = \mu_0/(2\pi)\ln\left[a^2/(4r_0(a-r_0))\right] and R~=R′/a[Ω/m]\tilde{R} = R'/a[\Omega/\mathrm{m}] are the inductance and resistance per unit length, respectively. Embedding loaded conductive wires in a host material can pose implementation challenges. Hence, we leverage PCB technology [39, 40] by substituting the wires with flat strips, each having a width of 2r02r_0. The resistors can be implemented in PCB technology as embedded resistors, such as for example [41]. The host material serves as the substrate for the PCB. However, substituting the metamaterial loaded conductive wires embedded in a host material with a PCB is non-trivial, and the design process involves the following steps.

Design procedure

Step 1 - Fitting to Eq. (11). A parametric sweep was conducted for the wired metamaterial on ϵh\epsilon_h and R′R' to minimize the root mean squared error (RMSE) between ϵdesired\epsilon_{\mathrm{desired}} and ϵanalytical\epsilon_{\mathrm{analytical}}. The optimal values were found to be ϵh=2.9\epsilon_h = 2.9, R′=1000[Ω]R' = 1000[\Omega].

Step 2 - Realizing an infinite absorbing metasurface structure in HFSS using strips instead of wires for practical purposes. The structure (without the PEC back-plane) comprises of 5 unit cells (shown to be a sufficient number of unit cells [42]), with a total thickness of 5a=d=0.005[m]5a = d = 0.005[\mathrm{m}] along the ZZ direction. Periodic boundary conditions were implemented on four faces of the structure, specifically at the XY plane. Additionally, two Floquet ports were assigned to the remaining faces, positioned at a distance of 60.5[mm]60.5[\mathrm{mm}] from the layer. The structure is excited by the fundamental Floquet wave-field, where the electric field aligns with the strip’s direction and propagates normally to the layer, see Figure 7 (a) for illustration. In this step, the strips are located at the center of the unit cell. Note that the distance between the Floquet ports and the absorbing structure was reduced for the purpose of providing a clearer visualization in Figure 7 (a). Furthermore a de-embedding process that shifts and calibrates the numerical reference plane of the simulation (the “Flouquet ports” in Figure 7 (a),(Figure 7d)) into the interface plane between the vacuum and the absorber was applied. At this stage, we conducted a sweep on ϵh\epsilon_h and compared the scattering matrix, specifically S11S_{11} and S21S_{21} following the re-calibration, with those obtained when substituting the structure with a bulk material having the required parameters ϵr\epsilon_r and σ\sigma. This sweeping process revealed that the optimal host material value should be tuned to ϵh=2.3\epsilon_h = 2.3. Note that the fundamental requirement for homogenization, ensuring that the unit cell dimension aa is significantly smaller than the minimal operating wavelength within the material, given by λm=0.02/2.3=13.1876[mm]\lambda_m = 0.02/\sqrt{2.3} = 13.1876[\mathrm{mm}] is satisfied.

Figure 7. Design and simulation layout in HFSS. (a) The initial simulation, used to extract the necessary host material properties, featured two Floquet ports defined at both ends of the structure. Periodic boundary conditions were enforced on the other four faces of the structure. (b) Side view of the stackup of a single unit cell, based on ROGERS RT-duroid 5880 lamination and ROGERS 3001 bonding material. (c) Oblique view of the unit cell. (d) The simulation setup of the final structure (similar to Figure 7(a)), incorporating the stackup described in (b). A parametric sweep was conducted on R′R' to determine the optimal parameter for the design.

Step 3 - Search for appropriate raw and bonding PCB materials. ROGERS offer RT-duroid 5880 laminate material with ϵd=2.2\epsilon_d = 2.2, tan⁡δd=0.0009\tan\delta_d = 0.0009 and ROGERS 3001 bonding material with ϵb=2.28,tan⁡δb=0.003\epsilon_b = 2.28,\tan\delta_b = 0.003 [43]. The stackup for the structure is established by combining these two materials with copper (simulated as PEC) traces. Figure 7 (b) illustrates the stackup for a single unit cell (side view). Figure 7 (c) depicts an oblique view of the unit cell structure. During the manufacturing process the bonding material covers the thin metal-

lic strips. Consequently, the thickness of each unit cell is 1.0033[mm]1.0033[\mathrm{mm}], resulting in a total absorber thickness of 5.0165[mm]5.0165[\mathrm{mm}]. Figure 7 (d) illustrates the final structure, incorporating the simulation definitions outlined in Figure 7 (a) and the absorber layout depicted in Figure 7 (b) and (c). Unlike Figure 7 (a), both horizontal and vertical loaded wires are introduced to ensure polarization insensitivity, specifically for electric fields within the wire plane (TE polarization). Importantly, the wires are positioned on separate planes, with a distance of 0.381[mm]0.381[\mathrm{mm}], as indicated in the stackup. An additional round of parametric sweep is carried out on the absorbing structure, varying the resistance R′R' in a manner similar to the preceding step. The electric field polarization was maintained as in the previous sweep. The corrected optimal observed value is R′=1025[Ω]R' = 1025[\Omega].

Step 4 - Homogenization of the final metamaterial design using scattering data extracted by full wave simulations. The metamaterial shown in Figure 7 (c) undergoes homogenization, where its effective parameters are derived from the simulated scattering matrix. The methodology for this process was initially introduced for normal incidence in [42]. Here, we aim to extract the effective parameters not only for normal incidences, but for TE and TM oblique incidences. Therefore, we take inspiration from existing works [44, 45] and extract the effective wave parameters under oblique incidence (see Appendix A for additional information). An alternative approach, though not implemented here, involves directly extracting the permittivity and permeability tensor coefficients using an extended field-sampling method [46]. It is important to note that the technique employed to extract the effective parameters in this work requires knowledge of the scattering matrix at the interface between the metamaterial and its surroundings, therefore the de-embedding process. Given the small size of the sample in comparison to the operating wavelength, there is no branching problem [47]. The proposed metamaterial shown in Figure 7 (c) lacks symmetry in the X-Y plane. Consequently, scanning solely over θ\theta is inadequate for accurately extracting its effective parameters; a scan across the X-Y plane is also required. This critical insight is explored and addressed further in the following discussion.

Our first objective is to identify the essential parameters for TE polarization, specifically when the electric field is confined to the XY plane (the plane of the wires). While the general electric field may consist of three components, i.e., E⃗=Exx^+Eyy^+Ezz^\vec{E} = E_x\hat{x} + E_y\hat{y} + E_z\hat{z}, TE polarization satisfy to the following conditions:

E⃗⋅n^=0,E⃗⋅k^=0.(12)\vec{E} \cdot\hat{n} = 0,\qquad\vec{E} \cdot\hat{k} = 0. \tag*{(12)}

Here, n^=z^\hat{n} = \hat{z} represents the normal to the layer, and k^=sin⁡(θ)cos⁡(ϕ)x^+sin⁡(θ)sin⁡(ϕ)y^+cos⁡(θ)z^\hat{k} = \sin(\theta)\cos(\phi)\hat{x} + \sin(\theta)\sin(\phi)\hat{y} + \cos(\theta)\hat{z} denotes the direction of propagation for the incident wave, and ϕ\phi denotes the azimuthal angle in the X-Y plane, while θ\theta denotes an elevation angle measured with respect to the normal. Note that n^\hat{n} and k^\hat{k} set the plane of incidence. The first condition implies that Ez=0E_z = 0, confirming the electric field’s polarization within the plane of the wires. The subsequent condition implies that Excos⁡(ϕ)+Eysin⁡(ϕ)=0E_x\cos(\phi) + E_y\sin(\phi) = 0 (for θ≠0\theta\ne0), which sets the relation between ExE_x and EyE_y. We define three distinct cases to extract the effective parameters: In the first case, when ϕ=0\phi= 0, the electric field is y^\hat{y} polarized, denoted as 1{sup}st Pol. In the second case, for $\phi = \pi/2$, the electric field becomes $\hat{x}$ polarized, identified as 2{sup}nd Pol. Lastly, in the third case with ϕ=π/4\phi= \pi/4, the electric field exhibits diagonal polarization at 45° from the xx and yy axes, denoted as 3{sup}rd Pol. {xref}Figure 8 presents the homogenization results for TE polarization, with the desired (see {xref}Eq. (10)) and analytical (see {xref}Eq. (11)) permittivities represented by blue and red dashed lines, respectively, which are desired to be independent of the incidence wave’s polarization. Subfigures (a) and (b) illustrate the real and imaginary parts of the extracted relative permittivity for the 1{sup}st Pol at θ=0∘\theta= 0^\circ (black), θ=30∘\theta= 30^\circ (dash-dotted purple), and θ=60∘\theta= 60^\circ (dotted green). Subfigures (c) and (d) depict the corresponding results for the 2{sup}nd Pol, while subfigures (e) and (f) show the results for the 3{sup}rd Pol. As shown in Figure 8, the designed metamaterial achieves relative (real) permittivities in the range ϵr∈[2.6,3]\epsilon_r \in[2.6, 3] across the entire frequency range of operation, while its imaginary part closely matches the desired profile. To provide the complete picture, Figure 9 depicts the extracted permeability μr\mu_r for the three excitations, compared to the desired value of μr=1\mu_r = 1. The real part of the permeability ranges from 0.919 to 1.02, while the imaginary part remains small.

Figure 8. Comparison between the desired, analytical and extracted permittivities (TE polarization). The desired (blue) and analytical (red) are irrespective of the incidence wave’s polarization. The extracted results for θ=0∘\theta= 0^\circ (black), θ=30∘\theta= 30^\circ (dashed purple) and θ=60∘\theta= 60^\circ (dotted green) are presented. (a),(b) Real and imaginary parts under the 1st Pol. (c),(d) Similar to (a),(b) however for the 2nd Pol. (e), (f) Similar to (a),(b) however for the 3rd Pol.

Figure 9. Comparison between the desired and extracted permeabilities (TE polarization). The desired (blue) and analytical (red) are irrespective of the incidence wave’s polarization. The extracted results for θ=0∘\theta= 0^\circ (black), θ=30∘\theta= 30^\circ (dashed purple) and θ=60∘\theta= 60^\circ (dotted green) are presented. (a),(b) Real and imaginary parts under the 1st Pol. (c),(d) Similar to (a),(b) however for the 2nd Pol. (e), (f) Similar to (a),(b) however for the 3rd Pol.

Our next objective is to extract the parameters of the metamaterial structure under TM illumination. Specifically, the magnetic field is confined to the XY plane (the plane of the wires). The magnetic field of such polarization satisfies the relations in Eq. (12) with the replacement E→HE \to H. The polarization definition for the TM case is defined similarly to the TE case. When ϕ=0\phi= 0, the magnetic field is y^\hat{y} polarized, denoted as 1{sup}st Pol. In the second case, for $\phi = \pi/2$, the electric field becomes $\hat{x}$ polarized, identified as 2{sup}nd Pol. Lastly, in the third case with ϕ=π/4\phi= \pi/4, the electric field exhibits diagonal polarization at 45° from the xx and yy axes, denoted as 3{sup}rd Pol. {xref}Figures 10 and 11` depict the extracted relative permittivities and permeabilities for all three polarizations, respectively. The blue line shows the desired result. The extracted results are presented for three values of θ\theta: a solid black line for θ=0∘\theta= 0^\circ, a dashed purple line for θ=30∘\theta= 30^\circ, and a dotted green line for θ=60∘\theta= 60^\circ. To ensure that the imaginary part of the permeability corresponds its real part (and also to avoid numerical errors), i.e. satisfies Kramers-Kronig relations, we used a Lorentzian model for the permeability. The extracted complex permittivity and permeability exhibit good alignment with the desired results. However, at larger angles (e.g., θ=60∘\theta= 60^\circ), the alignment weakens for the TM case, in contrast to the TE case, where the alignment remains strong even at large angles.

Dallenbach absorber performance

In the preceding discussion, we introduced a homogenization approach for a metamaterial absorber designed for PCB manufacturing. The objective of this section is to validate the performance of the proposed Dallenbach absorber, assessing both its bandwidth and angular spectrum. To achieve this, we eliminated the back Floquet port from the simulation setup and introduced a PEC back sheet, resulting in a simulation with only one port. The simulation setup is depicted in Figure 12 (a). The reflection coefficient (S11S_{11}) was computed for all three polarizations (both for TE and TM). For each polarization, the reflection coefficient was determined across a grid of 101 equally spaced frequencies ranging from 7 to 15 GHz and 91 equally spaced elevation angles spanning from 0 to π/2\pi/2. Figure 12 (b) illustrates the reflection for TE polarized wave, 20log⁡10∣S11∣20 \log_{10} |S_{11}|, from the designed Dallenbach absorber. The graph presents HFSS simulation results for the 3rd Pol, with similar results observed for other polarizations and when the metamaterial was replaced by a bulk material containing the optimal ϵ\epsilon, σ\sigma obtained in Sec. IV. The maximum error between the optimized bulk and its metamaterial realization (for all three polarizations) was −24-24 db.

Figure 12. Reflection and absorption characteristics of the Dallenbach metamaterial absorber. (a) Simulation setup in HFSS. (b) TE polarization reflection coefficient, expressed as 20log⁡10∣S11∣20\log_{10}|S_{11}|, is computed across a range of frequencies between 7 and 15 GHz, along with elevation angles spanning from 0 to π/2\pi/2. (c) The absorption coefficient, directly derived from the reflection coefficient, is given by A=1−∣S11∣2A=1-|S_{11}|^{2} and presented on a logarithmic scale. (d) TM reflection map (analogous to (b)). (e) TM absorption map (analogous to (c)).

As an additional validation step, we used the effective permittivity and permeability from Figure 8 and Figure 9 to create a hypothetical isotropic bulk material. The reflection and absorption coefficients were similar to those shown in Figure 12, so we have not included them here. Figure 12 (c) illustrates the absorption coefficient, denoted as A=1−∣S11∣2A = 1 - |S_{11}|^2, represented on a logarithmic scale (10log⁡10A10\log_{10} A). Figure 12(d) and (Figure 12e) show the corresponding results for the TM case. The maximum error between the optimized bulk material and the proposed metamaterial is approximately −10-10 dB, as the metamaterial exhibits increased frequency dependence under TM polarization at larger angles (e.g., θ>50∘\theta> 50^\circ). Despite the higher error, the absorption performance remains satisfactory. The efficiency of the proposed Dallenbach metamaterial absorber is evident in both frequency and angular spectrums, as depicted in Figure 12.

Image not extracted

Figure 10. Comparison between the desired, analytical and extracted permittivities (TM polarization). The desired (blue) is irrespective of the incidence wave’s polarization. The extracted results for θ=0∘\theta= 0^\circ (black), θ=30∘\theta= 30^\circ (dashed purple) and θ=60∘\theta= 60^\circ (dotted green) are presented. (a),(b) Real and imaginary parts under the 1st Pol. (c),(d) Similar to (a),(b) however for the 2nd Pol. (e), (f) Similar to (a),(b) however for the 3rd Pol.

Conclusion

In this work, we considered the problem of Dallenbach absorbing layers subjected to a wide angular spectrum of oblique incidence plane waves with TE or TM polarizations. A scenario that is typical in near field absorbing applications. We derived a modified sum-rule bound and showed that it is tight in a large parameter range. Moreover, we used this bound derivation in order to formulate an optimization problem for the design of a Dallenbach absorber that is applicable for spectrally wide and broadband impinging wave. A finite PCB realization using effective field theory and backed by full-wave simulations has been proposed. We demonstrated the efficacy of the absorber in three examples by placing a

Image not extracted

Figure 11. Comparison between the desired and extracted permeabilities (TM polarization). The desired (blue) is irrespective of the incidence wave’s polarization. The extracted results for θ=0∘\theta= 0^\circ (black), θ=30∘\theta= 30^\circ (dashed purple) and θ=60∘\theta= 60^\circ (dotted green) are presented. (a),(b) Real and imaginary parts under the 1st Pol. (c),(d) Similar to (a),(b) however for the 2nd Pol. (e), (f) Similar to (a),(b) however for the 3rd Pol.

dipole antenna operating in the 6G frequency band above a Dallenbach absorber. In the first example the dipole antenna was placed horizontally more than a wavelength above the ground plane, showing that it can be used to prevent major nulls in the radiation pattern due to destructive interference between direct and reflected waves in the far-field, and in a wide frequency range suitable for 6G communication applications. In the second example, we demonstrated that when the dipole antenna is placed in the immediate vicinity of the absorber, the gain variation is reduced bringing the pattern to be more uniform as is desired for wide communication coverage. In the third example we examined the case of a vertically-oriented dipole antenna placed about a wavelength above the ground plane (similarly to the first case). In all these scenarios, the designed optimal absorber significantly improved the radiation properties compared to the case where the radiators are placed near a conductor without the proposed optimal absorber. These examples demonstrate the applicability of the suggested bound and optimal design approach for timely real-world scenarios.

Acknowledgments

C. F. would like to thank to the Darom Scholarships and High-tech, Bio-tech and Chemo-tech Scholarships, to Yaakov ben-Yitzhak Hacohen excellence Scholarship and to IEEE Antennas and Propagation Society Fellowship (APSF). This research was supported by the Israel Science Foundation (grant No. 1457/23).

Formulas for extracting effective parameters

This appendix provides the formulas for parameters extraction under TE and TM polarizations. The extraction expressions are inspired by, but not identical to [44]. This is since we use [45] to correct a seemingly typographical inconsistency in [44] that applies for TM polarization. The two port system is described using the S-parameters. The normal wave vector component kzk_z and the generalized impedances (ξTE,ξTM\xi_{\mathrm{TE}},\xi_{\mathrm{TM}}) are given by,

kzd=±cos⁡−1(kz(1−S112)+kzS212kz(1−S11)+kz(1+S11))+2πm(13)k_zd=\pm\cos^{-1}\left(\frac{k_z(1-S_{11}^{2})+k_zS_{21}^{2}}{k_z(1-S_{11})+k_z(1+S_{11})}\right)+2\pi m \tag*{(13)}
ξTE=kz(1−S11)2−(S21)2(1+S11)2−(S21)2(14)\xi_{\mathrm{TE}}=k_z\sqrt{\frac{(1-S_{11})^{2}-(S_{21})^{2}}{(1+S_{11})^{2}-(S_{21})^{2}}} \tag*{(14)}
ξTM=1kz(1−S11)2−(S21)2(1+S11)2−(S21)2(15)\xi_{\mathrm{TM}}=\frac{1}{k_z}\sqrt{\frac{(1-S_{11})^{2}-(S_{21})^{2}}{(1+S_{11})^{2}-(S_{21})^{2}}} \tag*{(15)}

where mm is an integer, and is equal to zero for thin layers. The effective relative parameters are given by,

μr=kzξTE,ϵr=c2ω2kt2+kz2μr,TE(16)\mu_r=\frac{k_z}{\xi_{\mathrm{TE}}},\quad\epsilon_r=\frac{c^2}{\omega^2}\frac{k_t^2+k_z^2}{\mu_r},\qquad\mathrm{TE} \tag*{(16)}
ϵr=kzξTM,μr=c2ω2kt2+kz2ϵr,TM(17)\epsilon_r=k_z\xi_{\mathrm{TM}},\quad\mu_r=\frac{c^2}{\omega^2}\frac{k_t^2+k_z^2}{\epsilon_r},\qquad\mathrm{TM} \tag*{(17)}

where ktk_t is the transverse component of the wave vector (on the plane normal to zz).

References

References

  1. [1]G. Ruck, Radar Cross Section Handbook. Springer, 1970.DOI
  2. [2]C. Sohl, M. Gustafsson, and G. Kristensson, “Physical limitations on metamaterials: restrictions on scattering and absorption over a frequency interval,” Journal of Physics D: Applied Physics, vol. 40, no. 22, pp. 7146–7151, 2007.DOI
  3. [3]N. I. Landy, S. Sajuyigbe, J. J. Mock, D. R. Smith, and W. J. Padilla, “Perfect Metamaterial Absorber,” Physical Review Letters, vol. 100, no. 20, 2008.pubmed.ncbi.nlm.nih.gov/18518577
  4. [4]C. Sohl, C. Larsson, M. Gustafsson, and G. Kristensson, “A scattering and absorption identity for metamaterials: Experimental results and comparison with theory,” Journal of Applied Physics, vol. 103, no. 5, p. 054906, 2008.DOI
  5. [5]Y. Ra’di C.R. Simovski, and S.A. Tretyakov, “Thin Perfect Absorbers for Electromagnetic Waves: Theory, Design, and Realizations,” Physical Review Applied, vol. 3, no. 3, 2015.
  6. [6]D. E. Fernandes and M. G. Silveirinha, “Topological Origin of Electromagnetic Energy Sinks,” Physical Review Applied, vol. 12, no. 1, 2019.DOI
  7. [7]A. Krasnok, D. Baranov, H. Li, M.-A. Miri, F. Monticone, and A. Alù, “Anomalies in light scattering,” Advances in Optics and Photonics, vol. 11, no. 4, p. 892, 2019.
  8. [8]M. Gustafsson, K. Schab, L. Jelinek, and M. Capek, “Upper bounds on absorption and scattering,” New Journal of Physics, vol. 22, no. 7, p. 073013, 2020.
  9. [9]K. Schab, A. Rothschild, K. Nguyen, M. Capek, L. Jelinek, and M. Gustafsson, “Trade-offs in absorption and scattering by nanophotonic structures,” Optics Express, vol. 28, no. 24, p. 36584, 2020.DOI
  10. [10]S. Qu and P. Sheng, “Microwave and Acoustic Absorption Metamaterials,” Physical Review Applied, vol. 17, no. 4, 2022.DOI
  11. [11]M. I. Abdelrahman and F. Monticone, “How Thin and Efficient Can a Metasurface Reflector Be? Universal Bounds on Reflection for Any Direction and Polarization,” Advanced Optical Materials, vol. 11, no. 4, p. 2201782, 2022.arxiv.org/abs/2208.05533
  12. [12]D. Ye et al., “Ultrawideband Dispersion Control of a Metamaterial Surface for Perfectly-Matched-Layer-Like Absorption,” Physical Review Letters, vol. 111, no. 18, 2013.pubmed.ncbi.nlm.nih.gov/24237561
  13. [13]Z. Cao et al., “Backend-Balanced-Impedance Concept for Reverse Design of Ultra-Wideband Absorber,” IEEE Transactions on Antennas and Propagation, vol. 70, no. 11, pp. 11217–11222, 2022.DOI
  14. [14]A. Parameswaran, A. A. Ovhal, D. Kundu, H. S. Sonalikar, J. Singh, and D. Singh, “A Low-Profile Ultra-Wideband Absorber Using Lumped Resistor-Loaded Cross Dipoles With Resonant Nodes,” IEEE Transactions on Electromagnetic Compatibility, vol. 64, no. 5, pp. 1758–1766, 2022.DOI
  15. [15]Z.-C. Lin, Y. Zhang, L. Li, Y.-T. Zhao, J. Chen, and K.-D. Xu, “Extremely Wideband Metamaterial Absorber Using Spatial Lossy Transmission Lines and Resistively Loaded High Impedance Surface,” IEEE Transactions on Microwave Theory and Techniques, pp. 1–10, 2023.DOI
  16. [16]C. L. Holloway, R. R. DeLyser, R. F. German, P. McKenna, and M. Kanda, “Comparison of electromagnetic absorber used in anechoic and semi-anechoic chambers for emissions and immunity testing of digital devices,” IEEE Transactions on Electromagnetic Compatibility, vol. 39, no. 1, pp. 33–47, 1997.DOI
  17. [17]B.-K. . Chung and H.-T. . Chuah, “Modeling of RF Absorber for Application in the Design of Anechoic Chamber,” Progress In Electromagnetics Research, vol. 43, pp. 273–285, 2003.
  18. [18]C. Méjean et al., “Electromagnetic absorber composite made of aramid fibers loaded epoxy foam for anechoic chamber application,” Materials Science and Engineering: B, vol. 220, pp. 59–65, 2017.DOI
  19. [19]Y. Arien, P. Dixon, M. Khorami, A. Degraeve, and D. Pissoort, “Study on the reduction of heatsink radiation by combining grounding pins and absorbing materials,” 2015 IEEE International Symposium on Electromagnetic Compatibility (EMC), 2015.DOI
  20. [20]J. Xing et al., “Ultra-Thin SSPP-Based Sheet for Suppressing Microwave Radiated Emission in SiP Modules,” IEEE Transactions on Electromagnetic Compatibility, vol. 65, pp. 1–10, 2022.DOI
  21. [21]A. Khoshniat and R. Abhari, “Metamaterial Absorbers for Lining System Shield Box and Packaging: Cavity Analysis and Equivalent Material Design,” IEEE Transactions on Electromagnetic Compatibility, vol. 63, no. 4, pp. 1007–1014, 2021.DOI
  22. [22]C. Hägglund and S. P. Apell, “Plasmonic Near-Field Absorbers for Ultrathin Solar Cells,” The Journal of Physical Chemistry Letters, vol. 3, no. 10, pp. 1275–1285, 2012.
  23. [23]B. Mulla and C. Sabah, “Perfect metamaterial absorber design for solar cell applications,” Waves in Random and Complex Media, vol. 25, no. 3, pp. 382–392, 2015.DOI
  24. [24]K. N. Rozanov, “Ultimate thickness to bandwidth ratio of radar absorbers,” IEEE Trans. Ant. Prop., vol. 48, no. 8, pp. 1230–1234, 2000.DOI
  25. [25]J. P. Doane, K. S. Kertel, and J. L. Volakis, “Matching bandwidth limits for arrays backed by a conducting ground plane,” IEEE Trans. Ant. Prop., vol. 61, no. 5, pp. 2511–2518, 2013.DOI
  26. [26]L. B. Felsen and N. Marcuvitz, Radiation and Scattering of Waves. Wiley, 1994.
  27. [27]Note that both sides of Eq. (1) are positive.
  28. [28]R. B. Adler, L. J. Chu, and R. M. Fano, Electromagnetic Energy Transmission and Radiation, Hoboken, NJ, USA: Wiley, 1968.
  29. [29]R. E. Collin, Foundations for Microwave Engineering. McGraw-Hill, New York, 1966.
  30. [30]D. M. Pozar, Microwave Engineering. 4th ed. John Wiley & Sons, Hoboken, NJ, 2011.
  31. [31]J. D. Jackson, Classical Electrodynamics. John Wiley & Sons, Hoboken, 2007.
  32. [32]C. A. Balanis, Antenna theory analysis and design. Hoboken, New Jersey Wiley, 2016.
  33. [33]https://www.ansys.com/products/electronics/ansys-hfss
  34. [34]P.C. Bybi, G. Augustin, B. Jitha , C. K. Aanandan, K. Vasudevan and P. Mohanan, “A quasi-omnidirectional antenna for modern wireless communication gadgets,” IEEE Ant. Wire. Propag. Lett, vol.7, 2008.DOI
  35. [35]S. I. H. Shah, S. M. Radha, P. Park, and I-J Yoon, “Recent advancements in quasi-isotropic antennas: A review,” IEEE Access, vol. 9, 2021.DOI
  36. [36]Z. Zhang, S. Liao, Y. Yang, W. Che and Q. Xue, “Low-profile and shared aperture dual-polarized omnidirectional antenna by reusing structure of annular quasi-dipole array,” IEEE Trans. Ant. Prop., vol. 70, no. 9, 2022.DOI
  37. [37]Sergei Tretyakov, Analytical modeling in applied electromagnetics. Boston: Artech House, 2003.
  38. [38]C. Firestein, A. Shlivinski, and Y. Hadad, “Absorption and Scattering by a Temporally Switched Lossy Layer: Going beyond the Rozanov Bound,” Phys. Rev. Appl., vol. 17, no. 1, 2022.
  39. [39]M. R. Abdul-Gafoor, H. K. Smith, A. A. Kishk and A. W. Glisson, “Simple and efficient full-wave modeling of electromagnetic coupling in realistic RF multilayer PCB layouts,” IEEE Transactions on Microwave Theory and Techniques, vol. 50, no. 6, 2002.DOI
  40. [40]D. L. Jones, PCB Design Tutorial, 2004: chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://alfadex.com/wp-content/uploads/2014/02/PCBDesignTutorialRevA.pdf
  41. [41]chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://koaeurope.de/wp-content/uploads/XR73-Embedded- Resistors-2016-12-15.pdf
  42. [42]D. R. Smith, S. Schultz, P. Markoš, and C. M. Soukoulis, “Determination of effective permittivity and permeability of metamaterials from reflection and transmission coefficients,” Phys. Rev. B, vol. 65, no. 19, 2002.DOI
  43. [43]https://www.rogerscorp.com/advanced-electronics-solutions
  44. [44]C. Menzel, C. Rockstuhl, T. Paul, F. Lederer, and T. Pertsch, “Retrieving effective parameters for metamaterials at oblique incidence,” Physical Review B, vol. 77, no. 19, 2008.DOI
  45. [45]D. Cohen, R. Shavit, “Bi-anisotropic metamaterials effective constitutive parameters extraction using oblique incidence S-parameter method,” IEEE Transactions on Antennas and Propagation, vol. 63, no. 5, 2015.
  46. [46]R. Watanabe, M. Iwanaga and T. Ishihara, “s-polarization Brewster’s angle of stratified metal-dielectric metamaterial in optical regime,” Phys. Stat. Sol. (b), vol. 245, 2008.DOI
  47. [47]Z. Szabo, G.-H. Park, R. Hedge, and E.-P. Li, “A Unique Extraction of Metamaterial Parameters Based on Kramers–Kronig Relationship,” IEEE Transactions on Microwave Theory and Technique, vol. 58, no. 10, pp. 2646–2653, 2010.DOI

Paper details

vlnc:W_ugj2skqz
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents