A multiband patch antenna with embedded square split ring resonators in non-Homogeneous substrate

Received Apr 20, 2020 Revised Dec 5, 2020 Accepted Mar 16, 2021 The authors have attempted to influence an embedded square split ring resonator (SSRR) response in a stacked non-homogeneous substrate to demonstrate a quad-band antenna. The purpose is to produce multiband operations of a microstrip patch antenna. The highlighted factor is the effect of embedding an SSRR and the differing relative permittivity of the substrate on the side length of the SSRR. The analysis shows that a non-homogeneous dual substrate patch produces multiple bands compared to a single substrate patch antenna without any parameter change. A dual substrate antenna fabricated using FR4 and Rogers RT/Duroid 5880 copper clad sheets with a dimension of 85.6x54x0.908 mm (0.314λ0x0.198λ0x0.003λ0). The antenna resonates at 1.1, 2.45, 3.65 and 5.25 GHz in the L-, Sand C-bands. It is possible to employ the patch antenna in WLAN (dual-band) and WiMAX applications and suitable for mobile broadcast service at 1.1 GHz. The authors compare the simulated and measured results of a prototype in the article. The maximum measured gain is 5.48 dBi at 1.1 GHz and 4.025 dBi at 3.65 GHz. The measured bandwidth is 60 MHz (1.2%) at 5.25 GHz.


INTRODUCTION
The need to develop novel antennas for modern generation electronic devices poses a challenge to antenna engineers. Creating new and innovative designs to support wideband/multiband operations is essential. Performance in a compact size is a crucial factor in making an antenna appealing. The various WLAN bands, WiMAX and RFID, are of interest for many applications concerning mobile electronic devices, item identification, tracking, access control and other such applications.
A report is available on a microstrip antenna on an FR4 substrate of 0.8 mm thick with wideband characteristics and decent gain values. The antenna operates from 0.8 to 6.05 GHz and has a bandwidth of 5.25 GHz bandwidth [1]. A meandered stripline feed and H/U-type slits on the radiating plane are used in this design along with defected ground to obtain broadband characteristics. In literature, we see a great deal of effort in studying the various approaches to improve bandwidth in a compact form factor and facilitate a patch antenna's operation in multiple bands. The metallic conductors' magnetic behaviour report revolutionised devices' design in radio frequency and optical applications [2]. The authors discuss an antenna using a triangular split ring resonator [3] with a low measured gain. A triple-band antenna is reported with a reconfigurability feature and split ring resonator in [4] with a gain of 3 dBi. A T-shaped structure loaded with meander line stubs on a small area operates in multiple bands (L/S/C bands) and has decent bandwidths and a with some superior performance characteristics under their RT/Duroid product line. Researchers have used these substrate materials to develop their designs extensively. The design of multi-substrate antennas can be with bonded structures [13], [14], and [16]. It could also be a superstrate [7], [27], [28]. In this paper, a dual substrate antenna designed with FR4 (εr=4.4, tan δ=0.02) and Rogers RT/Duroid 5880 (εr=2.2, tan δ=0.00019) as the substrate material is studied. The dimension of the antenna is 85.6x54x0.908 mm 3 (0.314λ0x0.198λ0x0.003λ0). The top layer is of FR4, which is 0.4 mm thick (Layer 1), and the bottom layer is RT/Duroid 5880, which is 0.508 mm thick (Layer 2). The first stage of development was designing the patch antenna on a single substrate FR4 material of thickness 0.8mm. Figure 1(a) shows the antenna schematic. This antenna resonated at only one frequency (3.9 GHz). In the next step, the addition of 0.508 mm thick layer of Rogers RT/Duroid 5880 done below the FR4 layer (0.4mm thick). This structure yielded resonance at 2.45/3.6/5.21/5.93/6.16 GHz. In the next step, an SSRR array incorporated between the substrate layers leading to a change of the resonant bands to four frequencies (1.1/2.45/3.65/5.25 GHz) and at the same time suppressing the undesired frequency bands. The later sections in the article have a detailed discussion on this aspect.
The patch dimensions are calculated based on formulas for deriving the same as described in [29]. The transmission line model adopted to calculate the dimensions of the patch antenna. The effective relative permittivity ( ) of the antenna when the ration w/h>1 is given in (1) The width of the patch (wp) is related to the resonant frequency (fr), the velocity of light in free space (v0) and relativity permittivity (εr) by the relationship in (2), Consider the dominant mode of excitation (TM10) of the patch antenna, and the patch length is given by (3) as, The design utilises inset feed to obtain wideband impedance matching. The calculation of the inset length (y0) into the patch utilising in (4), Where, Rin (y=y0)=50 Ω and Rin (y=0) is the impedance at the edge of the patch. After obtaining the initial values, the dimensions iteratively modified to obtain the best possible EM solver results for the desired band of frequencies. The list of dimensions of the antenna and SSRR is in Table 1. Figure 2

Design of square split ring resonator
A metamaterial unit cell's performance and an array have been described based on the Kramers-Kronig relationship in [30]. Parameter retrieval based on the S-parameters of a unit cell metamaterial is discussed in detail [31], [32]. The above work explores structures exhibiting tuneable magnetic conductivity built from nonmagnetic conducting sheets. H. Chen et al. [33] refers to S-shaped resonators acting as lefthanded materials without the aid of a wire to generate double negative values of EM parameters.
The etching of a 3x3 SSRR array on the Rogers RT/Duroid 5880 substrate's top surface is as per the design in the proposed work. Available literature talks about different types of resonant structures like circular, rectangular, triangular and other such shapes Shalini Sah and Malay R Tripath [34], Chaturvedi and Raghavan [35] to cite a couple of research articles. The reason for selecting a square resonator is that it requires a smaller dimension (side length) compared to a circular or any other such shape. Studies show that aligned split gaps of the rings in the same direction generates dual resonances in close range. Proximity double resonance is otherwise is not possible in the complementary orientation of the split gaps. As shown in (5) gives the average length of the conductor lavg [36], where is the average relative permittivity of the dual substrate layer [37], N is the number of turns of the rings. The expression for effective relative permittivity (εrsub) of the substrates (7), With the help of the above equations, description of the capacitance and inductance of a metallic inclusion on a dielectric substrate is possible by the below-given as [36], Where h=h1 + h2 is the total height of the substrate. ρ is the fill factor for the metallic inclusion on the dielectric substrate (ρ=(N-1) (w+s)/[l1-(N-1) (w+s)], K is the elliptic integral of the first kind. The variable k defines the lower integer number of the ratio [s/(w+s)]. LSSRR is the inductance, and CSSRR is the capacitance of the metallic inclusion, fr is the resonant frequency of the SSRR structure based on the equivalent quasistatic model defined by the equations above. The calculated value of fr as per (11) Figure 3 shows the unit cell in a waveguide medium set up to extract the S-parameters of the SSRR to evaluate the EM response, and the tutorial describes the method for extraction of S-parameters of a unit cell. The tutorial highlights the procedure for retrieval of permeability, permittivity and refractive index [38]. The extraction of the EM parameters is possible by the Nicholson-Ross-Weir (NRW) method or transmission-reflection (TR) method (two popular techniques of a couple of methods available). We make use of (12) and (13) to compute the refractive index (n) and impedance (z) of the metamaterial, Once the refractive index and impedance is known, the magnetic permeability (μ) of the SSRR is calculated by (14), The plot of permeability response of the SSRR is in Figure 4. The real part of the structure's permeability is negative at 2.362 GHz and 3.628 GHz exhibiting the single negative material characteristic. The permeability response is nearly zero for a considerable range of frequencies in the 5-5.5 GHz band, as seen in Figure 4(a). Figure 4 (b) describes the response of the SSRR in a configuration of a three-cell unit and 3x3 array. The mutual coupling factor between the rings contributes to the extended behaviour of the threecell unit's permeability response and 3x3 array.
The use of two different substrates has resulted in smaller dimensions of the metallic inclusion's side length. A discussion on dielectric composition's effect on resonant frequencies of a metallic inclusion in the conference article [39] highlights the importance of dielectric composition. Embedding the SSRR structure between the two layers helps lower the resonance frequency than the same etched on a single substrate. The resonant frequency for a split ring resonator with a side length of the outer ring l=8.5 mm for a single cell is tabulated in Table 2 for different substrate material selections.  The resonant frequency values obtained indicate that the effect of non-homogeneous substrates on the resonant frequency is considerable. The result would be more profound at higher frequencies. Table 2 compares the proposed SSRR configuration and SRR response in terms of their resonance frequencies. As the reaction of metallic inclusions clearly illustrates that the SSRR produces dual resonance compared to SRR due to the combined effect of per unit capacitance and the gap capacitance. SRR produces a single lower resonant frequency for the given side length dimension. Hence, it is advantageous to incorporate in cases where a single resonance is required. Whereas the proposed SSRR produces dual resonance for the same structure due to the combined capacitive effect as stated above. Further size reduction is possible by selecting one material with higher relative permittivity (~10) and the other with a lower relative permittivity (~2). A 40% reduction in the resonance frequency in a few cases is achievable [39].

Reflection coefficient (S11)
A commercial solver based on the finite element method (FEM) (Ansys HFSS) assesses the antenna design numerically. Figure 5 The total thickness of the antenna substrate is 0.908 mm. We see in the second case that the antenna resonates at six frequencies (1.1/2.45/3.6/5.21/6.16 GHz) even though the radiating patch dimension is unaltered. It is due to the change in the effective relative permittivity of the substrate. The third case is that of a dual substrate antenna with embedded SSRR. The antenna resonates at four frequencies (1.1/2.45/3.65/5.25 GHz). In comparing the second and third case, the suppression of the frequency band's upper end (6.16 GHz) is due to the embedded SSRR. It is due to the attenuation induced by the real part of the embedded SSRR array's permeability (refer to Figure 4). Another factor is the improved reflection coefficient levels of the antenna with the SSRR compared to the one without SSRR due to metallic inclusion.
The reflection coefficient S11 of the fabricated dual substrate patch antenna with the embedded SSRR was measured using an Anritsu vector network analyser (VNA-S820E). Figure 5 Table 3 tabulates the prototype antenna's measured bandwidth and the antenna's simulated bandwidth with and without an SSRR array. It is a proven fact that the bandwidth of a conventional microstrip patch antenna is directly proportional to the substrate's thickness. The antenna's impedance bandwidth is lesser than other designs discussed in the literature, mainly due to two reasons. First, the antenna uses a relatively thin substrate (0.908 mm) compared to different models, which use 1.6 mm or higher thickness values. Second, the proposed design does not incorporate bandwidth enhancement techniques like meander lines or defective ground structure. We obtain better bandwidths for the nonhomogenous structure compared to a single substrate antenna of comparable design parameters.
Parametric study of the antenna carried out by varying the slot length Sl1 (varied between 22 mm to 28 mm). The results of the parametric analysis for the reflection coefficient is in Figure 5(b). The parametric study yielded the best results for a slot length of 25 mm for Sl1.

Gain and efficiency of the antenna
The antenna response characterised in an anechoic chamber. The radiation measurement is done in a chamber of dimension 7x4x3 m 3 . A distance of 5.5 m separates the AUT from the source that is a broadband horn antenna of linear polarisation. The antenna exhibits decent gain levels in the four frequency bands of interest. The plot of simulated and measured antenna gain is in Figure 6(a). Table 4 tabulates the gain values at the different resonant frequencies. The highlight of the design is decent gain values at different resonant frequencies by a thin substrate structure. It is due to the use of a dual substrate and the embedded SSRR. The use of low loss substrates like Rogers RT/Duroid 5880 and the SSRR enhances the patch antenna's gain.
The plot in Figure 6(b) depicts the microstrip patch antenna's radiation efficiency with embedded SSRR, and the efficiency of the antenna is 47% at 3.65 and 5.25 GHz. The efficiency at 1.1 and 2.45 GHz is 42.5 and 41%, respectively. The dual substrate antennas (with and without SSRR) shows a marked improvement in efficiency compared to the single substrate antenna printed on FR4 substrate (9% at 3.9 GHz). The use of second substrates aids in improving the overall radiation efficiency across the various resonance frequencies.   Figure 7 shows the surface current distribution obtained from the HFSS simulation. The current distribution shows how the structure surface is excited by the current. The major contributors are the slots etched to provide longer current paths and the feedline of the patch. The current concentrates along the length of the slots (Sl1, Sl2 & Sl3). We observe that the current intensity is higher along the edges of slot Sl1 at almost all the frequencies. We position the slots Sl2 & Sl3 selectively by design to influence the current distribution at 1.   Figure 8 in appendix, depicts the radiation pattern in the antenna's E and H planes at 1.1 GHz, 2.45 GHz, 3.65 GHz, and 5.25 GHz. The radiation pattern of the antenna exhibits a good correlation between the simulated and measured values. The pattern is dipolar in H-plane at 2.45 GHz, whereas it is omnidirectional in all other resonant frequencies (both E & H). The kinks/distortions observed in the measured value can be because of the coaxial connector's soldering with the patch, surface currents induced in the metallic inclusions in the multilayer structure influencing the radiation. The pattern measurement constrained by a protruded mounting mechanism hindering the values in particular radial.

Discussion
A list of a comparative study of a few of the research articles available concerning specific key parameters like frequency bands, substrate height, gain and number of substrates is in Table 5. The comparison brings out the uniqueness of the proposed structure visa-a-vie the ones published to date. The designed antenna operates in multiple frequency bands without alteration in the radiating patch's physical dimension. Secondly, it exhibits proper levels of gain on a thin substrate across all the resonant frequencies.
Thirdly, an embedded SSRR array suppresses undesired frequencies and enhance radiation performance in other frequency bands. The most crucial factor is the technique of alignment and bonding of the two dielectric substrates. There is a chance of a shift in resonant frequency if the substrates are not appropriately aligned.

CONCLUSION
The article does a detailed study of a microstrip patch antenna made of non-homogeneous substrate materials with an embedded SSRR array operating in 1.1, 2.45, 3.65 and 5.25 GHz bands. The numerical analysis and measurement of the prototype have shown very close agreement in the various performance parameters. The antenna's evolution from a single substrate to a dual substrate with embedded SSRR indicates improved performance indices of the vital parameters related to the antenna. The dual substrate feature aids in reducing the side length of the SSRR and makes it capable of resonating at lower frequencies, thereby bringing about a kind of frequency reconfigurability and compactness. The antenna produces maximum gain measured is 5.48 dBi at 1.1 GHz and the maximum bandwidth obtained is 60 MHz (1.2%) at 5.25 MHz on a thin substrate (0.908 mm). The proposed antenna produces output only in the required bands and suppresses unwanted resonances. The dual substrate antenna's performance in L-, S-and C-bands is encouraging and can be used in various wireless applications. Broadside radiation is noted in 1.1/3.65/5.25 GHz, while dipolar radiation pattern in H-plane at 2.45 GHz. The proposed design reduces 48% in patch length and 55% in patch width of the antenna.