Glass Penetrating Transparent Surface with Square Loop/Cross Double-Layer FSS for Indoor Millimeter-Wave Communications

Article information

J. Electromagn. Eng. Sci. 2025;25(5):491-499

Publication date (electronic) : 2025 September 30

doi : https://doi.org/10.26866/jees.2025.5.r.321

Thinh Tien Nguyen, Chang Won Jung

Graduate School of Nano IT Design Fusion, Seoul National University of Science and Technology, Seoul, Korea

^*Corresponding Author: Chang Won Jung (e-mail: changwoj@seoultech.ac.kr)

Received 2024 October 18; Revised 2025 January 9; Accepted 2025 February 2.

Abstract

I. Introduction

Over the past few decades, frequency selective surfaces (FSSs) have been deployed for radar cross-section reduction purposes in military applications (e.g., aircraft, absorbent metamaterials) [1, 2]. Given their ability to control electromagnetic waves, FSSs have come to play important roles in microwave and optics systems as spatial filters [3–8], radomes [9, 10], absorption material [11, 12], polarization convertors [13], and transmit array antennas [14, 15]. To meet the requirements of these applications, narrowband characteristics, indicating a high-quality factor (Q-factor), are usually required to increase frequency selectivity [3]. In addition, the transparency aspect of FSSs has motivated many researchers to develop various applications, including those that improve signal communication [16, 17], electromagnetic shielding [18–20], and frequency selection on glass [4], among others. In this regard, most reports have investigated single-band applications [4, 16, 18], though some have been geared toward multi-band applications [17, 19, 21].

At present, several transparent FSS designs are being developed for use in the millimeter-wave (mmWave) band [12, 16–18, 22]. For example, a bandstop FSS based on an acrylic substrate was designed using the transparent electro-textile technique for on-chip applications within the frequency range of 27–35 GHz. This design is highly transparent, inexpensive, and waterproof, but its limitation lies in its poor out-band rejection region [18]. In 2021, the latest publication on transparent FSS research presented by Chen et al. [17] achieved low insertion loss (IL) and high transparency at mmWave frequencies (27.5–29.5 GHz). However, these FSSs had to be designed using expensive window glass materials (i.e., quartz), and the fabrication cost increased with meshing, similar to another proposed FSS [16].

In the context of 5G wireless communication, research on transparent FSSs applied to conventional glass surfaces, known as glass penetrating transparent surfaces (GPTS) [4, 17, 22–24], has identified many potential applications. These include improving signal quality and reducing out-of-band interference between communication devices in buildings and densely populated areas, thereby creating a “cleaner” electromagnetic environment [17]. In [23], we analyzed a simple single-layer FSS structure, but it was limited by its relatively high IL coefficient and operating bandwidth, which led to a low Q-factor and reduced the design’s frequency selection effectiveness. Furthermore, in [24], the authors investigated the optimization and analysis of metal mesh on glass surface, significantly improving transparency. However, the problem of high cost and challenges in industrial-scale implementation remained significant barriers.

In this work, we propose a novel design for a high Q-factor GPTS, designed based on a cascaded structure of conventional window glass, for signal selection in the n257 band. To thoroughly address undesired signals, an FSS_2 layer is designed based on dual-square loop units [10, 25], along with an additional PET layer solution, which was introduced in an earlier work [22] to dramatically improve IL. Simulations of the proposed GPTSs were conducted using CST Microwave Studio software. They were fabricated by the PETCCL-etching method, and then experimentally validated. The measurement results showed that the proposed GPTSs achieve low IL, a narrow bandwidth, a low passband ripple level (PRL), and a high outof-band rejection rate.

II. Design of the Proposed GPTS Structure on Window

Fig. 1 shows the proof of concept for signal selectivity using the proposed GPTS FSS in the desired n257 band, indicating that signal quality improves after the signal is transmitted through the window glass for 5G indoor communications.

Fig. 1

Concept of the GPTS with signal selectivity through window glass for 5G indoor communications.

1. GPTS Unit-Cell Configuration

The schematic diagram of the proposed GPTS and its physical dimensions, as shown in Fig. 2, were carefully optimized. The structure comprises a bottom layer made of conventional glass (ɛ_{r_Glass} = 7.38 (1–j0.06)), a 0.1 mm-thin PET layer (ɛ_{r_PET} = 3 (1–j0.005)) on top that is enclosed by two separate FSS units, and a 1.5 mm-thick PET layer between the FSS_2 layer and the glass layer. With regard to the design, the FSS_1 layer was designed based on a combination of two arrow-corner and cross-shaped units, while the structure for the FSS_2 layer was based on a combination of dual-square loop units [10].

Fig. 2

The proposed GPTS (h₁ = 0.1 mm, h₂ = 1.5 mm, h₃ = 2 mm, l = 3 mm, l₁ = 1 mm, l₂ = 1.6 mm, w = w₁ = w₂ = w₃ = 0.1 mm).

Fig. 3 presents an analysis of the equivalent circuit theory through which the behavior of the proposed GPTS structure can be predicted.

Fig. 3

Equivalent circuit analysis model for the proposed GPTS unit cell (C₁ = 0.038 pF, C₂ = 0.005 pF, C_m = 0.078 pF, L₁ = 0.59 nH, L₂ = 0.52 nH, L₃ = 0.5 nH, L₄ = 0.55 nH, Z_s1 = 217.7 Ω, Z_s2 = 138.8 Ω).

This structure can be described as an inductor in a parallel configuration and an LC resonator in series. Specifically, L₁ and C₁ represent the arrow angles of the FSS_1 layer, and L₂ denotes the cross-shaped element. In the FSS_2 layer, L₃ and L₄ describe the square-outer and square-inner loop elements, respectively, while C₂ is generated by coupling between the dual-square loop elements. Notably, each dielectric layer can be modeled by the two short segments of a transmission line, with the characteristic impedance expressed as Zs=Z0/ɛr, the exact value of which will depend on the permittivity (ɛ_r) of each dielectric, where Z₀ is the free space impedance (i.e., 377Ω). The initial circuit parameter values were approximated using formulas from an earlier work [20] and then optimized using the schematic tool in CST software, as noted in the caption for Fig. 3. The value of |S₂₁| for the equivalent circuit model (ECM) is also depicted in Fig. 4(a), along with the results obtained from the GPTS for TE and TM polarizations, and glass simulation. In addition, the center frequency at the normal incidence angle (i.e., 0°) was determined to be partially reduced by the capacitance (C_m) between the two layers, expressed as follows:

Fig. 4

Comparison of transmission responses (|S₂₁|): (a) ECM, GPTS for TE and TM polarizations, and glass simulations; (b) simulation of the structure with and without the FSS_2 layer on glass and glass at 0°.

(1) fc≈12π×L1+L2L1×L2(C1+Cm).

Fig. 4(a) exhibits the transmission properties of the TE and TM polarizations at 0°. Notably, the polarization-insensitive characteristic is demonstrated by the identical transmission values [3]. Furthermore, the improvement in frequency filtering when adding the FSS_2 layer is clearly depicted in Fig. 4(b). Applying a combination of a high-pass filter and a band-stop filter, represented by the corresponding lumped components L₃, C₂, and L₄ in the ECM, narrowed the bandwidth for the entire structure. As mentioned before, the frequency of the transmission zero point of the proposed GPTS at 34.1 GHz was determined using the following equation:

(2) ftz≈12π×L3+L4L3×L4(C2+Cm).

2. Parametric Study and Analysis

As illustrated in Fig. 5(a), in the numerical simulation, the bandwidth of the proposed GPTS becomes narrower as the width of the arrow angle (w₁) increases, while the resonant frequency shows signs of shifting to the left. With regard to the ECM, as depicted in Fig. 5(b), L₁ gradually decreases with an increase in w₁, while C₁ remains relatively stable. This shift in the transmission zero point can be attributed to the increase in the arrow angle area, consequently leading to an increase in C_m formed between the two FSS layers. Moreover, as shown in Fig. 6(a), the enhancement in bandwidth narrowing and attenuation of out-of-band signals becomes evident with the increase in width of the outer square (w₂). In addition, Fig. 6(b) shows that L₃ and C₂ exhibit opposite changes with increasing w₂. Subsequently, the theoretical optical transparency (OT) was determined based on an equation presented in the literature [26–29].

Fig. 5

Performance of the proposed GPTS at various w₁ values: (a) S₂₁ and (b) relationship between L₁ and C₁.

Fig. 6

Performance of the proposed GPTS at various w₂ values: (a) S₂₁ and (b) relationship between L₃ and C₂.

Notably, structural meshing [16] can be applied to this structure to attain higher transparency. However, the trade-off between fabrication costs and achieving the desired transparency is an issue that must be accounted for. Based on the above information, an optimal structure was selected, with w₁ and w₂ values of 0.1 mm achieving an acceptable theoretical OT value (i.e., 70.02%), along with stable FSS performance.

The simulated |S₂₁| results of the proposed GPTS for both TE and TM polarizations at various incidence angles (θ) are shown in Fig. 7. The simulated 3-dB bandwidth of the FSS ranges from 26.1 to 30.2 GHz, thus covering the desired n257 band. Usually, for TE polarization, the characteristic impedance changes with Z₀cos(θ), while for TM polarization, it is equivalent to Z₀cos(θ) as the value of θ changes [30]. Fig. 7(a) shows that for TE polarization, the 3-dB bandwidth becomes narrower as the value of θ increases. Meanwhile, as depicted in Fig. 7(b), the 3-dB bandwidth under TM polarization is significantly expanded within the operating frequency range. This is demonstrated by the 3-dB bandwidth range at θ = 45°, which is 26.3–29.5 GHz and 25.1–29.8 GHz for the TE and TM polarizations, respectively. Additionally, it is observed that the out-of-band frequencies of TE polarization are significantly improved compared to TM polarization. This phenomenon can be explained by the increased mutual coupling between the two FSS layers in the structure as the characteristic impedance decreases.

Fig. 7

Simulated transmission response (|S₂₁|) of the proposed GPTS: (a) TE and (b) TM polarizations at various incident angles.

3. Improvement of Transmission Response (|S₂₁|)

In Fig. 8, an analytical model with an additional PET_3 layer built into the bottom of the proposed GPTS (GPTS2) is presented. The PET_3 layer was added to solve the issue of phase-shift limitation caused by the FSS layers, thereby resulting in a significant improvement in the S₂₁ coefficient. Simulation results showed identical outcomes for ECM analysis and for both TE and TM polarizations at 0°, with |S₂₁| of the glass and glass/PET at 28 GHz being 1.25 dB and 2.05 dB, respectively.

Fig. 8

Equivalent circuit analysis model for the proposed GPTS2: (a) the ECM, and (b) ECM and full-wave simulation results at 0°.

Fig. 9 presents the simulated results for the transmission response of the proposed GPTS2, showing that its Q-factor value at 0° improves dramatically when the 3-dB bandwidth is narrowed to 27.2–29.3 GHz. Meanwhile, the absolute value of the minimum value of the transmission response (|S₂₁|_min) is 0.84 dB, at which point channel efficiency through the FSS reached its peak. However, the out-of-band region appears to increase only slightly. For TE polarization, the 3-dB bandwidth narrows as the value of θ increases. As for TM polarization, the 3-dB bandwidth in the operating frequency band expands slightly and is more stable than in the case of the proposed GPTS structure.

Fig. 9

Simulated transmission response (|S₂₁|) of the proposed GPTS2: (a) TE and (b) TM polarizations at various incident angles.

III. Experimental Verification and Discussion

Fig. 10 shows the measurement setup. Keysight VNA E8364B was employed to validate the performance of the proposed GPTSs. The measurement procedure is summarized as follows: First, two fabricated using the PETCCL-etching method [31–33], as shown in Fig. 10(b) and 10(c), respectively, consisting of 100 × 100 unit cells with a size of 30 cm × 30 cm. The FSS samples were implanted into an aluminum wall to reduce multipath reflection and diffraction [3]. Second, two model SAS-588 active horn antennas, characterized by a broadside gain of 14.6 dBi and a beamwidth of approximately 30° in both the E- and H-planes, were positioned 1 meter apart in the Ka-band (26.5–40 GHz), thus satisfying the conditions for far-field measurement. In addition, the measured cable loss within the operating frequency range was approximately 1.5 dB. A comparison of the OT of the fabricated GPTS with that of glass is shown in Fig. 10(d).

Fig. 10

Photograph of (a) the measurement setup; fabricated samples of (b) the FSS_1 layer and (c) the FSS_2 layer; and (d) the OT of glass and the proposed GPTS.

Fig. 11 shows the measured results for OT within the spectral range of 400–750 nm. It can be observed that the fabricated GPTS and GPTS2 reach about 69.1% and 65.7%, respectively, at a 550 nm wavelength, which is in good agreement with the theoretical OT. In contrast, the corresponding transparencies of glass and PET are 93.3% and 95.6%, respectively.

Fig. 11

Measured optical transparency of substrates, including glass, PET, proposed GPTS, and GPTS on glass/PET.

Fig. 12 shows the measured transmission responses for both TE and TM polarizations of GPTS and GPTS2 at 0° and 45°. The 3-dB bandwidths of GPTS and GPTS2 at 0° are 26.3–30 GHz and 27.4–29.1 GHz, respectively. Furthermore, the 3-dB bandwidths of the GPTS for TE and TM polarizations at 45° are 26.8–29.6 GHz and 25.4–30.1 GHz, while the corresponding values for the GPTS2 are 27.5–28.9 GHz and 26.5–29.2 GHz. Additionally, the |S₂₁|_min values yield measurements of 2.31 and 0.98 dB for GPTS and GPTS2 at 28 GHz, respectively, indicating an improvement of about 0.72 dB over conventional glass. Compared to the simulation results mentioned earlier, the measured frequency range indicates a slight rightward shift. In particular, at 0°, the resonant frequency exhibits a marginal increase of approximately 1.07% for GPTS and 0.56% for GPTS2, with the corresponding shifts for TE polarization at 45° being about 0.72% and 0.89%.

Fig. 12

Simulated transmission response (|S₂₁|) for (a) TE and (b) TM polarizations at 0° and 45° of the proposed GPTS and GPTS2.

Interestingly, the resonant frequency for TM polarization at 45° is consistent with the measured value observed at 0°. This shift can primarily be attributed to the formation of small air gaps between the cascaded FSS structure and the glass surface.

Additionally, it may have partly resulted from the actual loss of the substrates (i.e., PET and glass), as well as non-uniform transmit-receive power from the horn antenna pair during actual measurements.

Table 1 compares the performance results obtained in this work with those obtained for several existing designs. It is evident that the proposed designs have the narrowest 3-dB fractional bandwidth, a low IL value at θ = 0°, and the highest Q-factor, along with acceptable transparency, favorable size values, and low PRL. Furthermore, they allow for saving fabrication costs by using metal lines while exhibiting polarization insensitivity up to θ = 45° in the mmWave band.

Table 1

Comparison with reported transparent FSSs

IV. Conclusion

In this paper, a novel high-Q-factor GPTS capable of operating in the n257 band is proposed. The equivalent circuit method was employed to theoretically predict the behavior of the FSSs. The simulation and measurement results were observed to be in good agreement with the theoretical calculations. Furthermore, an additional PET-layer solution was implemented to improve the transmission response of the proposed GPTS, achieving an |S₂₁|_min value of 0.98 dB in the frequency range of 26.5–29.5 GHz, with a polarization insensitivity under oblique incidence up to 45° for both TE and TM polarizations. Overall, owing to their outstanding transmission performance, the proposed GPTS structures can be considered promising candidates for signal filtering in 5G indoor mmWave communications.

Article information Continued