In a numerical investigation evaluating candidate hardware for ultra-high-speed wireless networks, researchers have designed a compact multiple-input multiple-output antenna system engineered specifically for terahertz-band operation across sixth-generation communications and advanced Internet of Things architectures. Through full-wave electromagnetic modeling, the two-port configuration established stable multi-resonance behavior across four distinct frequencies at 2.35, 3.60, 4.51, and 6.42 terahertz, spanning two ultra-wide operating frequency windows. By introducing a hybrid slot-elliptical unit cell alongside an integrated slotted decoupling barrier, the authors achieved a peak simulated antenna gain of 10.58 decibels and a total radiation efficiency of 85.82 percent while substantially suppressing inter-element mutual coupling down to minus 31.1 decibels. Furthermore, the overall physical footprint was maintained at a compact normalized area of 0.62 by 1.74 wavelengths, offering substantial spatial miniaturization.
Conventional microstrip patch antennas have long served as primary components throughout radio-frequency and microwave engineering due to their low profile, lightweight planar geometry, and straightforward integration alongside surrounding printed circuitry. However, traditional microstrip patches encounter severe operational bottlenecks when scaled into high-frequency domains, including inherently narrow operational impedance bandwidths, low radiation efficiency, diminished forward gain, and elevated dielectric losses. These physical impediments become particularly acute within the terahertz frequency spectrum, conventionally designated between 0.1 and 10 terahertz, where atmospheric attenuation, high signal propagation loss, and challenging transmission ranges hinder long-distance electromagnetic links. The researchers addressed these fundamental propagation obstacles by implementing substrate-level electromagnetic band gap engineering to constrain unwanted surface-wave modes without adding cumbersome external filtering apparatus.
Material selection played a decisive operational role in mitigating conductive and dielectric dissipation across the simulated terahertz radiating structure. The investigators utilized graphene, a two-dimensional carbon allotrope structured in an sp2-hybridized honeycomb lattice, for both the radiating patch elements and the underlying ground plane. In the electromagnetic simulation suite, graphene was characterized through a surface impedance boundary condition based on the complex conductivity formulation derived from the Kubo model and Drude approximations. To match representative experimental parameters for terahertz devices, the team established a chemical potential of 0.1 electron-volts, an electron relaxation time of 1 picosecond, an effective conductive thickness of 0.01 micrometers, and an ambient operating temperature of 300 kelvins. The conductive graphene layers were supported by a dielectric polyimide substrate possessing a physical thickness of 11 micrometers, a relative permittivity of 3.5, and a low dielectric loss tangent of 0.0027. This synthetic polymer substrate provides mechanical flexibility, high thermal resilience, and minimal dielectric attenuation across the intended spectrum.
Before assembling the multi-port array, the authors systematically developed and optimized a single-element microstrip patch radiator through standard analytical equations followed by parametric electromagnetic refinements. The base radiating patch was assigned physical dimensions of 55 micrometers by 65 micrometers centered upon a substrate footprint measuring 70 micrometers by 80 micrometers. Within the graphene patch, the team etched two mutually orthogonal cylindrical slots to stimulate additional electromagnetic resonance modes and broaden impedance bandwidth. Furthermore, two symmetric rectangular slots were cut along the lateral edges of the patch to redirect surface-current distributions and improve radiation matching. The radiator evolved across three progressive development steps: Step 1 utilized an unslotted graphene patch over a uniform periodic hole lattice, yielding a weak reflection coefficient of minus 10 decibels and modest gain of 5 decibels; Step 2 introduced patch truncation and slotting, deepening reflection to minus 20 decibels; and Step 3 incorporated hybrid elliptical perforations, dropping reflection to minus 46 decibels and raising gain to 10 decibels.
Comprehensive parametric sweeps revealed the precise operational sensitivity of the single-element radiator to geometric variations in the substrate lattice and microstrip feeding network. The researchers examined three distinct air hole diameters within the photonic band gap substrate, finding that a 5-micrometer hole diameter delivered an optimal reflection dip near minus 46 decibels and a peak gain of 10.5 decibels, outperforming test cases of 3 and 7 micrometers. Similarly, varying the center-to-center hole spacing across 8, 10, and 12 micrometers confirmed that a 10-micrometer pitch achieved superior impedance matching across the target bands. Adjusting the inset microstrip feed width identified 25 micrometers as the optimal dimension, while testing polyimide thicknesses of 10, 11, and 12 micrometers validated the 11-micrometer baseline. Comparative material benchmarks demonstrated that graphene significantly outperformed copper and gold patches, while the perforated photonic band gap substrate achieved an efficiency of 84.14 percent, markedly higher than the 40 to 65 percent recorded on a continuous homogeneous dielectric slab.
To transition from the isolated single radiator into a multi-channel diversity system, the research team evaluated five sequential physical configurations for the two-port multiple-input multiple-output array. Configuration 1 placed two identical radiators in a vertical orientation with matching polarization, which generated resonant dips down to minus 28 decibels but exhibited strong electromagnetic interaction with inter-port transmission reaching minus 9 decibels. Configuration 2 rotated one element by 90 degrees, slightly improving isolation to minus 13 decibels while shifting the resonance points. Configuration 3 aligned the radiators in opposite directions, producing an irregular reflection profile and an undesirable worst-case transmission peak of minus 5 decibels. Configuration 4 adopted a horizontal side-by-side arrangement, which deepened the reflection coefficient to minus 37 decibels but still yielded minus 12 decibels of mutual transmission. To overcome these inter-port coupling limitations, Configuration 5 inserted a 30-micrometer-wide slotted decoupling barrier featuring periodic slots measuring 8 micrometers by 0.6 micrometers between the horizontal patches, successfully driving transmission down to minus 31.1 decibels.
The finalized two-port array demonstrated exceptional dual-band multi-resonance capabilities across the investigated terahertz spectrum, meeting key performance criteria for broadband wireless links. Full-wave simulations confirmed four distinct reflection notches occurring at 2.35 terahertz with a depth of minus 51.3 decibels, 3.60 terahertz at minus 26.65 decibels, 4.51 terahertz at minus 51.4 decibels, and 6.42 terahertz at minus 41.52 decibels. Together, these resonance dips define two continuous operational bandwidths spanning 1.477 to 2.67 terahertz, representing a bandwidth of 1.19 terahertz, and 3.35 to 7.08 terahertz, representing an ultra-wide bandwidth of 3.72 terahertz. The transmission coefficient between the two excitation ports peaked at minus 31.1 decibels, verifying robust inter-element electromagnetic isolation. Furthermore, the combined assembly sustained a maximum simulated forward gain of 10.58 decibels alongside an overall radiation efficiency of 85.82 percent, confirming that the majority of input electrical energy is converted into radiated electromagnetic power rather than being lost to internal resistive or dielectric mechanisms.
To establish the operational validity of the design for advanced diversity systems, the authors calculated several standard multiple-input multiple-output metrics across the working frequency bands. Consequently, the calculated diversity gain reached 9.997 decibels, closely approaching the theoretical ideal ceiling of 10 decibels. The total active reflection coefficient, which characterizes cumulative reflected energy when both ports are simultaneously excited across arbitrary phase angles, remained below minus 10 decibels throughout the entire operating spectrum, recording prominent minima of minus 37 decibels at 2.35 terahertz and minus 36 decibels at 4.51 terahertz. The simulated channel capacity loss stayed beneath 0.45 bits per second per hertz, and the mean effective gain difference between the ports remained strictly within 3 decibels, verifying balanced reception and uniform power transfer.
Directional far-field radiation modeling and equivalent circuit synthesis further elucidated the electromagnetic behavior of the proposed terahertz antenna. At a representative resonance frequency of 2.35 terahertz, simulated far-field patterns revealed strong directional radiation, with the E-plane main lobe registering a magnitude of 13.2 decibel-volts per meter at 53.0 degrees with a 3-decibel angular beamwidth of 175.3 degrees. The corresponding H-plane pattern exhibited a main lobe magnitude of 18.7 decibel-volts per meter directed at 92.0 degrees with an angular width of 82.0 degrees and a sidelobe level of minus 8.3 decibels. To validate these full-wave numerical findings, the investigators constructed a lumped RLC equivalent circuit model in the Advanced Design System simulation environment, assigning parallel resonant branches to represent each operating mode alongside reactive coupling elements. Quantitative comparison between the CST full-wave results and the lumped circuit model revealed close agreement, recording a root-mean-square error of 3.37 decibels and a mean absolute error of 2.42 decibels across the full 1 to 7.2 terahertz evaluation span.
Despite the promising numerical metrics reported throughout the computational study, the authors emphasized several significant physical constraints, noting that physical prototypes have not yet been fabricated or experimentally validated. Future investigations outlined by the research team will focus on experimental fabrication procedures, developing specialized measurement testbeds for terahertz impedance and radiation characterization, and refining the underlying photonic band gap topology to enhance isolation further. Additionally, the team plans to scale the dual-element configuration into larger multiple-input multiple-output arrays, aiming to expand channel capacity and spatial beamforming capabilities across prospective sixth-generation communication networks and intelligent sensor systems.
Subject of Research: Terahertz graphene MIMO antenna design using photonic band gap substrate engineering
Article Title: Design and performance analysis of a PBG substrate MIMO antenna with RLC validation for 6G and IoT applications
Article References: Fahim, T. A., Riad, M. H., Das, I., Alkhaibari, A., Tiang, J.-J., Singh, N. S. S., & Haque, M. A. (2026). Design and performance analysis of a PBG substrate MIMO antenna with RLC validation for 6G and IoT applications. Discover Internet of Things, 6(1), Article 260. https://doi.org/10.1007/s43926-026-00543-z
Image Credits: AI Generated
DOI: 10.1007/s43926-026-00543-z
Keywords: terahertz communication, 6G networks, MIMO antenna, graphene, photonic band gap, polyimide substrate, equivalent circuit model, Internet of Things, Design, performance, analysis, substrate
News Source: Neil Sanderson. (November 8, 2026). Engineers Model Graphene Terahertz MIMO Antenna for Next-Generation Wireless Communication. Scienmag.
