This article investigates the validity of the physical optics/geometrical optics (PO/GO) approximation in modeling low-permittivity integrated lens antennas when there is significant illumination of the shadow region as a function of the feed and lens geometry. The comparative analysis reveals that for certain extended hemispherical lenses, feed illumination of the lens surface beyond the critical angle leads to significant discrepancies in radiation pattern and antenna gain between PO/GO predictions and full-wave simulations. These discrepancies are traced to the constructive far-field contribution of the shadow region currents relative to those originating from the top lens region below the critical angle. The measurements of fabricated extended hemispherical lenses confirm these findings, showing up to 2-dB gain enhancement over PO/GO predictions and showcasing the limitations of traditional high-frequency modeling techniques in capturing this phenomenon. The significant gain enhancement, along with the clean measured radiation patterns, highlights the potential of using lens antenna designs with feeds that provide strong shadow region illumination. These findings provide new insights into the development of high-performance integrated lens antennas for advanced communication and sensing applications.

Sub-THz wireless links targeting Tbps throughput face a fundamental tension between bandwidth and energy efficiency. Zero-Crossing Modulation (ZXM) is a promising approach to reduce receiver power consumption by combining 1-bit quantization with temporal oversampling. However, the resulting 1-bit receive chain introduces a strong nonlinearity, which limits the applicability of conventional linear multipleinput multiple-output (MIMO) spatial equalization techniques and makes the impact of coupling and crosstalk a primary system design constraint. This motivates the use of ZXM in combination with 1-bit quantization for a quasi-optical (QO) MIMO system, where the spatial inter-stream interference is very small. Hence, we present a system-level link-budget analysis for an 8×8 QO MIMO architecture at sub-THz, based on a signal-to-interference-plus-noise ratio (SINR) analysis that treats interstream interference as additional noise after 1-bit quantization. The end-to-end coupling matrix is modeled as the cascade of antenna coupling and RF front-end crosstalk contributions. Using this framework, we quantify when the link transitions from noise-limited to interference-limited operation, show SINR saturation with increasing transmit power, and derive feasible operating regimes over distance and spectral-efficiency targets for wideband 50 GHz operation. The results substantiate the feasibility of energy-efficient Tbps-class wireless connectivity.

This work presents a chessboard focal plane array (FPA) camera with state-of-the-art thermal and spatial resolution in the 200 600 GHz frequency range. The FPA is implemented in a 130-nm SiGe BiCMOS technology, where each antenna element is loaded with a direct detector based on heterojunction bipolar transistors (HBTs). The antenna and detector architecture, including the vias and biasing network, were optimized to achieve a noise-equivalent power (NEP) suitable for passive THz imaging. Overall, the estimated loss of the FPA is better than 4 dB between 350 and 600 GHz, of which 1.5 dB is due to ohmic losses in the FPA, 1 dB to mutual coupling between detectors, and 0.7 dB to the impedance mismatch between the detector and antenna. A prototype of 24 pixels was manufactured and mounted on the base of a silicon hyperhemispherical lens with an anti-reflection coating. Excellent spatial resolution is achieved through a tight element spacing in the fabricated FPA, which is only half the wavelength in silicon at 350 GHz and therefore consistent with the state-of-the-art. Its responsivity, noise, and radiation patterns were characterized using a quasi-optical measurement setup. The measured radiation patterns are within 1 dB of simulations, demonstrating that the integrated THz camera achieves excellent spatial resolution. Between 330 GHz and 500 GHz, the NEP was measured to be on the order of 10 pW/vHz. When considering the entire operational band, this NEP results in a noise-equivalent temperature difference (NETD) of the camera is 1.6 K for an integration time of 1 s per pixel, which is comparable to the state-of-the-art. While THz detectors with state-of-the-art sensitivity are limited to single-pixel designs, the presented work combines a multi-pixel implementation with competitive sensitivity.

A new generation of quasi-optical (QO) systems with multiple refractive and reflective components are required for several (sub)-millimeter applications such as the next generation of communication systems, sensing and security imagers, and instrumentation for far-infrared astronomy. Currently, there are no numerically efficient and accurate modeling tools available to design such systems tailored to their moderate sizes, in terms of wavelength, and wide operational bandwidths. In this article, a numerically efficient and accurate technique is proposed for designing QO systems containing a cascade of homogeneous dielectric lens components. The methodology is based on sequential geometrical optics (GO) combined with an analysis in reception mode. Thanks to the computational efficiency of the proposed technique, we were able to embed it as the kernel for multiobjective optimizers to design complex multilens QO systems. Here, two indicative examples of such geometries targeting specific requirements and scenarios are showcased. The overall optimization technique is validated against in-house and commercial physical optics (PO) codes, as well as full-wave simulations where possible, with excellent agreement in both efficiency terms and far-field beam patterns. Moreover, the proposed technique achieves orders of magnitude faster execution times with respect to the currently available commercial tools.

This work details the design and characterization of a wideband on-chip dual-polarized double-slot antenna for sub-THz frequencies, integrated in SiGe-BiCMOS technology. The proposed design features two perpendicularly arranged double-slots that provide dual-linear polarization and efficient illumination of dielectric lenses through the silicon substrate. To decrease the level of cross-polarization, a number of parasitic slots are added. Two feeding architectures have been studied, leading to two different prototypes. The first is designed in a 130 nm SiGe-BiCMOS process and is tailored for integration with single-ended circuit topologies. It achieves a maximum antenna efficiency of 63% in its operational bandwidth of 170-250 GHz, with a cross-polarized component lower than −22 dB up to 240 GHz. The second prototype is designed in a 90 nm process and is suited for differential circuits. A 55% fractional bandwidth is achieved around the central frequency of 287.5 GHz, with a maximum antenna efficiency of 56% and a cross-polarized component lower than −19.5 dB. The port isolation is better than −37 dB up to 300 GHz. Both prototypes are characterized in the 200-325 GHz frequency range, with very good agreement between the measurements and simulations.

This work presents a modified Marchand balun for sub-terahertz frequencies. By introducing an asymmetric ground shield in the coupler sections, low imbalance is achieved at the differential port. The proposed design is well-suited for on-wafer testing of differential power amplifiers and antennas. A prototype fabricated in a 90-nm SiGe-BiCMOS process is characterized in the 200–325 GHz range, showing good agreement with simulations. The measured amplitude and phase imbalances at 270 GHz are 0.1 dB and 2.5°, remaining lower than ±1 dB and 3.5° in the 240–290 GHz frequency range. The measured insertion loss is below 2.5 dB, while the simulated return loss in the single-ended port is better than 10 dB in the 220–340 GHz range.

This work presents an electrically-small lens that has been redesigned towards a flat interface. This way, the lens is easier to integrated, compared to an earlier introduced spherical core-shell lens concept. The lens is created from a single dielectric host material by conformally machining holes into the material. In this process, two artificial dielectric layers are created; The first layer is used for anti-reflection purposes, whereas the second is used to convert the spherical interface to a flat interface. The two layers enable the use of holes with lower aspect ratio drilling, compared to classical gradient-index lenses. The lens is designed to operate in the 140-170 GHz bandwidth, and a prototype with height of only 2.2 mm and diameter of 6.6 mm was fabricated and characterize. The prototype is small enough to fit in many integrated circuit packages. The flat lens was compared to a non-flat core lens in terms of pattern quality, return loss and dielectric loss, with only negligible performance degradation.

This contribution presents the assessment of RDL technology at sub-terahertz frequencies based on polybenzoxazole (PBO) polymers. A stack of two PBO layers of 10μm thickness with 3 metallization of 5μm copper is being under development. Vias as small as 10μm and a separation of 60μm are being explored. To assess the materials and capabilities of this technology, GCPW, stripline transmission line structures are being assessed, expecting losses in the order of 1.3dB/mm and 2.1dB/mm respectively. Moreover, two resonators have been designed to enhance the accuracy of the characterization.

Radiative near-field links have gained noticeable interests recently for high-data-rate wireless communication. Unlike far-field links, near-field links can have negligible path loss within hundreds of meters for electrically large antennas at high frequencies. In this work, we propose a multi-lens quasi-optical (QO) system for 100-m near-field backhaul communication at H-band. The QO system is designed with compact size (aspect ratio of 1.3:1) and high coupling efficiency of 82%. Moreover, the rotation of an auxiliary lens realizes beam scanning for the link alignment. The scan range is in the order of 1 m with less than 2 dB scanning coupling loss and scanning magnification of 14.5:1.