High-isolation antenna array using SIW and realized with a graphene layer for sub-terahertz wireless applications

This paper presents the results of a study on developing an effective technique to increase the performance characteristics of antenna arrays for sub-THz integrated circuit applications. This is essential to compensate the limited power available from sub-THz sources. Although conventional array structures can provide a solution to enhance the radiation-gain performance however in the case of small-sized array structures the radiation properties can be adversely affected by mutual coupling that exists between the radiating elements. It is demonstrated here the effectiveness of using SIW technology to suppress surface wave propagations and near field mutual coupling effects. Prototype of 2 × 3 antenna arrays were designed and constructed on a polyimide dielectric substrate with thickness of 125 μm for operation across 0.19–0.20 THz. The dimensions of the array were 20 × 13.5 × 0.125 mm3. Metallization of the antenna was coated with 500 nm layer of Graphene. With the proposed technique the isolation between the radiating elements was improved on average by 22.5 dB compared to a reference array antenna with no SIW isolation. The performance of the array was enhanced by transforming the patch to exhibit metamaterial characteristics. This was achieved by embedding the patch antennas in the array with sub-wavelength slots. Compared to the reference array the metamaterial inspired structure exhibits improvement in isolation, radiation gain and efficiency on average by 28 dB, 6.3 dBi, and 34%, respectively. These results show the viability of proposed approach in developing antenna arrays for application in sub-THz integrated circuits.


Scientific Reports
| (2021) 11:10218 | https://doi.org/10.1038/s41598-021-87712-y www.nature.com/scientificreports/ amplifiers constituting the RF front-end of sub-terahertz communication system have restricted output power and bandwidth. Existing technology relies on mixing optical sources essentially lasers to generate low to moderate power-levels at sub-THz band. Receivers at such frequencies rely on direct detection using Schottky-diodes, which is not band selective and has a low dynamic range and sensitivity 4 . At sub-THz frequencies metals exhibit a lower conductivity than at DC. This phenomenon enhances field infiltration within the metal at sub-THz and adversely affect the radiation efficiency of metallic antennas 5 . In addition to the low conductivity of metal at these frequencies, the effect of small geometric parameters of metallic antennas, specifically the width or radius of metallic traces which are smaller than 0.1 micron, must be considered. This is because numerical analysis reveals that small scale antennas of sub-100 nano-meters compared to their millimeter-sized counterparts results in significantly low radiation efficiency at sub-THz frequency due to the high surface resistance of metallic traces 6 . Moreover, because of the method employed for conductor deposition in the fabrication of ultra-thin metallic traces of sub-nano-meter thickness, the magnitude of the metal's conductivity becomes considerably lower than that of the bulk material. The low conductivity is attributed to the following factors: grain boundary scattering, surface scattering, and surface roughness 7,8 . These factors can considerably reduce the theoretically predicted radiation efficiency of antennas at sub-terahertz frequencies.
Recently various sub-THz antenna arrays have been investigated for THz spectroscopy based on slots 9 , printed dipoles 10 or patch antennas 11 however the radiation characteristics of these antennas are limited for practical applications. Hence there is demand for antennas at these frequencies with improved radiation gain and efficiency characteristics. Although high performance wideband horn antennas are now available 12 , they are bulky in structure even at very high frequencies. Instead, low profile planar antennas are desirable for ease of integration with RF-circuits to realize compact transceivers 13,14 .   www.nature.com/scientificreports/ In this study the isolation, radiation-gain and efficiency of sub-terahertz array structure is improved by employing a combination of substrate integrated waveguide (SIW) and metamaterial (MTM) inspired technologies. The prototype antenna arrays were designed and analysed using numerical simulation tools, and the optimized designs fabricated and measured. The array structures were coated with a thin layer of Graphene as it supports the propagation of surface plasmon polaritons enabling it to operate in the sub-terahertz frequency band 15,16 . In 17 it has been shown that plasmons in graphene layer strongly confine electromagnetic energy, and in 18 it has been shown that the application of a voltage to the graphene layer affects the phase characteristics of the antenna. These results show Graphene material opens exciting prospects for antenna development in the terahertz frequency band.

Sub-terahertz antenna based on SIW Technology
Geometries of the standard reference patch and the proposed SIW-loaded patch antenna are shown in Fig. 1. The patch antenna was constructed on a polyimide dielectric substrate (Kaneka, Apical NP, USA) with thickness of 125 μm, dielectric constant ( ε r ) of 3.5, and loss-tangent (tangδ ) of 0.0027. To produce the graphene patch, a CO2 laser (Universal Laser 230 VLS) was used to scribe the front and rear polyimide surfaces and thereby convert this into a graphene film, using the same experimental conditions as discussed in 19 . To produce the vias, the laser power was increased from 8.1 W (as used for the standard patch on the front face) to 15 W, and the laser scribe was repeated twice in these areas, in order to 'punch' through the polyimide surface. A cross section of the via taken using an SEM (Hitachi SU5000 field emission scanning electron microscope) can be observed in Fig. 2. The proposed antenna includes SIW isolator which is framed around the square patch. Dimensions of the antenna are given in Table 1. Reflection-coefficient response for both antennas over 0.194-0.196 THz is shown in Fig. 3. The results reveal SIW isolation improves the impedance match of the antenna, which is pronounced at the resonance frequencies of 0.1942 THz and 0.1954 THz. Surface current density distribution over the two antennas at their dominant resonant modes in Fig. 4 show the effectiveness of the SIW isolator in suppressing the propagating surface waves over the structure. Suppression of surface waves is desirable in antenna arrays to mitigate mutual coupling between adjacent radiating elements and thereby improve the antenna's radiation characteristics. Radiation gain and efficiency plots of both antennas at the resonant frequencies in Fig. 5 show the SIW isolator significantly enhances the radiation-gain and efficiency properties over the reference antenna, due to reduction in substrate loss.

SIW loaded antenna with slots
Metasurface characteristics are introduced into the patch antenna by the inclusion of periodic array of subwavelength length slots of varying length to realize a wider impedance bandwidth, as shown in Fig. 6 20 . These slots exposing the substrate essentially manipulate the electromagnetic (EM) response of the surface. The EM waves impinging on the metasurface induce electric and magnetic dipole moments, which are related to the effective permittivity and permeability of the composite medium. Detailed theoretical analysis of this type of structure is provided in 21 . To confirm the antenna supports backward waves the CST Microwave Studio, which is a 3D electromagnetic simulation tool, was used to obtain its dispersion characteristics that was computed using 22 : where β is the wave propagation constant along the direction of propagation, and p is the length of the unit-cell. Figure 7 shows how the phase through the structure varies with frequency. The dispersion diagram slope is Over this frequency range the signal group and phase velocities are oriented along opposite directions resulting in a backward wave, which is characteristic of metamaterials. This is due to the sub-wavelength slots that control the electromagnetic response of the surface to produce a homogeneous response at 'macroscopic' level. The simplified equivalent circuit of the SIW-loaded slotted antenna in Fig. 8 will be referred to hereon as SIW-loaded MTM. The lumped element values in the equivalent circuit were extracted using the well-established pseudo-inverse technique from S-parameters 22,23 . According to 22  The array of periodic slots etched on the patch generate multiple resonances whose frequencies can be determined from the dispersion curve when the following condition is satisfied 20 : It can be shown from Eq. (2), there are three resonance frequencies which can be excited including the ± 1st-order resonances and the zeroth-order resonance at f se . These resonant frequencies are verified from the reflection-coefficient response of the antenna in Fig. 9, which are at f r−1 : 0.1941 THz, f r0 : 0.1950 THz, and f r+1 : 0.1957 THz. The reflection-coefficient shows significant improvement in impedance match between 0.194 and 0.196 THz with the inclusion of the slots.
The current density distribution in xy-plane at the three resonant frequencies is shown in Fig. 10. The current distribution at f r−1 : 0.1941 THz and f r+1 : 0.1957 THz are very similar; however, it is significantly muted at f r0 : 0.1950 THz. The size of the antenna is 3.89λ 0 × 4.54λ 0 × 0.08λ 0 when operated at f r−1 : 0.1941 THz.
Comparison of the radiation plots of the reference, SIW-loaded, and SIW-loaded slotted antennas are shown in Fig. 11. It is evident that the SIW-loaded MTM patch antenna exhibits significantly improved performance over the other two antennas. The optimum radiation gain of the SIW-loaded MTM antenna at resonance modes − 1st (f r-1 : 0.1941 THz), zeroth (f r0 : 0.1950 THz), and + 1st (f r+1 : 0.1957 THz) are 8.05 dBi @87.7°, 8.0 dBi @51.6°, and 7.92 dBi @59.3°, respectively. Besides the gain, the radiation efficiency at resonance modes − 1st, zeroth, and + 1st are 45%, 45%, and 47%, respectively. In the case of f r−1 : 0.1941 THz and f r+1 : 0.1957 THz, the gain of the reference antenna is comparable to the other two antennas, but this is over a very narrow azimuth range between 12° and 26°. Total dimensions of the three antennas are identical.

Array configuration of the proposed SIW-loaded MTM antenna for sub-THz applications
In this section 2 × 3 antenna arrays are realized based on the standard reference, SIW-loaded, and SIW-loaded MTM patch antenna. The main issue encountered in an array structure is the adverse effect of mutual coupling between the radiating elements constituting the array. Several techniques previously studied and proposed are exemplified in [23][24][25][26][27][28] but the size of the array matrix investigated in all cases reported was limited to 1 × 2. The techniques proposed previously include the use of air gap between radiator and ground-plane or the implementation of ground-plane defection. In all cases reported the isolation achieved is limited and is restricted over narrow frequency band.   www.nature.com/scientificreports/ In small-sized antenna array structures it is very difficult to supress the mutual coupling effects resulting from surface wave interaction between the radiation elements due to their proximity. Proposed here is a simple but effective mutual coupling reduction technique based on SIW isolator as described in "SIW loaded antenna with slots" section. The prototype of the SIW-loaded MTM antenna is shown in Fig. 12, where the SIW isolator is wrapped around each patch to significantly reduce propagation of surface waves and minimise substrate loss. The prototype antenna array was coated with 500 nm thick film of Graphene. The identical patches in the array have dimensions of 3 × 3 × 0.125 mm 3 . The dimension of the 2 × 3 antenna array is 20 × 13.5 × 0.125 mm 3 . Dimensions of the optimized structural parameters are tabulated in Table 2.

Results and discussions
The simulated and measured impedance bandwidth of the standard reference, SIW-loaded, and SIW-loaded MTM 2 × 3 antenna arrays are shown in Fig. 13. CST Microwave Studio was used obtain the simulation results. Keysight PNA Network Analyzer with a frequency extender was used to measure the reflection-coefficient of the antenna. These results clearly show that the SIW isolator has a significant impact on the impedance matching and bandwidth of the array over the standard reference antenna array. Further improvement is achieved by incorporating sub-wavelength slots in the patch antennas. The experimental results show that with SIW-loading the impedance matching is improved by 10 dB on average from 190 to 200 GHz over the reference antenna array. By incorporating both SIW-loading and sub-wavelength slots the impedance matching also improves on average by 14 dB from 190 to 200 GHz compared to the reference array antenna. There is generally good agreement between the simulated and measured results. The discrepancy in the results is attributed to manufacturing tolerance and unaccounted loss in the simulation model.
The transmission-coefficient response of the standard reference, SIW-loaded, and SIW-loaded MTM 2 × 3 antenna array are shown in Fig. 14. These results demonstrate the effectiveness of the proposed technique of using SIW isolator in suppressing the surface-wave propagations and reducing substrate loss to mitigate mutual coupling effects without increasing the overall size of the array structure. Experimental results show that compared to the reference antenna array the SIW-loaded array is shown to improve the isolation on average by 17 dB from 190 to 200 GHz. Also, by incorporating sub-wavelength slots the SIW-loaded MTM array is shown to suppress mutual coupling on average by more than 28 dB across 190 GHz to 200 GHz compared to the reference array antenna.
The simulated and measured radiation gain and efficiency of the standard reference, SIW-loaded, and SIWloaded MTM (CRLH) array are shown in Fig. 15. Experimental results show that compared to the reference array      The set-up to measure the far-field radiation patterns of the SIW-loaded MTM antenna arrays is shown in Fig. 16. In transmit mode we used a 2-way power splitter in conjunction with two 3-way power splitters. It was necessary to ensure the lengths were such that the phase at the antenna terminals were identical. This arrangement was used to split equally the input sub-THz signal applied to the 2 × 3 array. Conversely, in the receive mode we used two 3-way power combiners in combination with a 2-way power combiner. This arrangement combined the received signal power from the 2 × 3 array. The output from the array in receive mode was connected to Keysight E4448A spectrum analyser. Keysight E8244A signal generator was used to produce a signal that was adjustable in the frequency range 38-40 GHz with enough power (10-20 dBm) for the mixer to generate a set of harmonics simultaneously. The 5th harmonic was used as it was in the desired frequency range of 190-200 GHz.   Table 3. The comparison shows that the proposed antenna array has comparable gain and radiation efficiency to the cited references, which confirm the effectiveness of the proposed approach. It is less complex and cost effective to implement and realize in practice for mass production, which makes it a viable candidate for applications to sub-terahertz wireless systems.

Conclusion
We have demonstrated the use of a Graphene-based antenna at sub-THz and the effectiveness of the proposed technique of framing each radiator with SIW. This technique is shown here to reduce surface wave propagations and suppressing near-field mutual coupling between the radiators in an array. The consequence of this is significant improvement in the array's performance in terms of impedance matching and radiation performance which is achieved without compromising the footprint of the antenna array. Further improvement in performance was demonstrated by incorporating sub-wavelength slots in the patch radiators. The feed mechanism employed is however considered to be unfeasible for a large array, where aperture coupling would be more appropriate.