In-Ho Lee

This letter proposes an inverse design method for radome frequency selective surface (FSS) based on conditional generative adversarial network (cGAN). Wasserstein loss with gradient penalty (GP) is applied to enhance the optimization of the generative network. FSS design typically requires the integration of high-permittivity radomes, which complicates achieving multifunctional characteristics and leads to significant increases in computation time. The trained network facilitates the real-time generation of novel designs that closely approximate the target frequency responses, even in the absence of corresponding data in the existing dataset. The proposed network was validated through simulations and experimental measurements of the fabricated prototype. One of the generated structures achieved dual pass band with fractional bandwidths at the -2.5 dB transmission level of 5.13% in the K-band and 6.08% in the Ka-band, demonstrating high frequency selectivity.

Multiple quantum well (MQW) light‐emitting diodes (LEDs) provide precise wavelength control, making them ideal light sources. However, achieving simultaneous short‐wavelength infrared (SWIR, 1–3 µm) and mid‐wavelength infrared (MWIR, 3–5 µm) emission from a single LED presents significant technical challenges due to lattice mismatch and reduced quantum efficiency when MQW structures with distinct bandgap energies are integrated onto a single substrate. As a result, most LEDs typically operate in only one IR band. In this study, monolithic multi‐band MQW LEDs capable of simultaneous SWIR and MWIR emission are demonstrated. Strain engineering via Sb doping in the QWs induces well‐distributed local lattice distortions, such as modulations of atomic bond angles and lengths, leading to balanced strain compensation and coherent epitaxy with atomically sharp interfaces within the MQWs. Reducing the QW thickness of InAsSb enhances quantum confinement, enabling simultaneous SWIR and MWIR emission at 2.87 and 3.18 µm. To further extend the emission range, a simulation‐based fabrication feasibility map is developed, and an additional monolithic LED that emits simultaneously at 2.63 and 3.34 µm is fabricated. The monolithic integration of multi‐band emission into a single device not only reduces size and complexity but also facilitates multispectral analysis for future optoelectronic devices.

We propose a pixel-based metasurface design with asymmetry parameters with a large number of pixels used to construct the unit cell. Each square-shaped pixel has only two states, either metal or dielectric. Typically, the unit cell is designed to have a rotational symmetry. However, in this work, we introduce a set of additional parameters that can break the symmetry of the unit cell, closely associated with the physics of bound states in the continuum. These asymmetry parameters, defined by the scale of pixels in two dimensions, induce Fano resonances with high quality-factors in the transmission spectra and allow for the design of frequency-selective metasurfaces that are more specific to the general frequency-response characteristics.

In this letter, we propose multifunctional radome-based frequency selective surface (FSS) composites for stealth applications. The proposed composite simultaneously implements band stop characteristics in the X-band and band pass characteristics in the K- and Ka-bands. The composite consists of a sandwich structure, where two FSS layers are separated by a constant spacing. Thick dielectric materials with higher permittivity and loss tangent than air degrade the filtering performance. We thus performed conformational space annealing (CSA)-based optimization on both the FSS shape and dimension, resulting in a miniaturized pixelated FSS. This approach effectively addresses the downward shift in the resonance frequency by employing an FSS that is 22% smaller than a standalone FSS. We validated the proposed composite through simulations and measurements under varying polarizations and incident angles. The proposed composite achieved insertion losses greater than 20 dB across the entire X-band at a 45° oblique incidence in transverse electric (TE) mode. Under transverse magnetic (TM) mode at a 30° oblique incidence, the composite allowed transmission in the dual K-band and Ka-band while maintaining insertion losses below -18 dB within the stop-band.

In this paper, we use five types of deep-learning algorithms for denoising scanning electron microscope (SEM) measurement data. Denoising of SEM images is an important task since the images often suffer from noise, which can make it difficult to accurately interpret the data. We also investigate realistic SEM denoising characteristics using a variety of metrics to assess the quality of denoised images. Overall, we find that the trained generative models provide superior denoising performance and that it is crucial to objectively quantify the performance, just like in the scanning process itself. It is anticipated that the deep-learning based technique can accelerate image measurements, which can be utilized for very fast analytical investigations. We also demonstrate that the success of a generative model may depend on the appropriate assessment of noise characteristics in the specific image data analysis of interest. Moreover, it is addressed that denoising performance can be properly evaluated when a relevant metrics that aligns well with human visual systems.

In this letter, we propose a multifunctional frequency selective surface (FSS) designed for wide incident angles, optimized for X-band reflection and K-/Ka-band transmission. To verify its applicability, we analyzed the performance variations of the FSS with respect to polarizations and incidence angles. The proposed FSS facilitates space communication in the K/Ka-band, while simultaneously minimizing electromagnetic interference for X-band defense applications. The pixelated structures of the FSS are optimized to realize fine frequency selectivity in the target frequency bands. The proposed FSS is implemented in a double-layer to provide higher-order characteristics. To validate its performance, measurements are conducted in a free-space measurement system with spot-focusing lens horn antennas. The proposed structure demonstrates a -1.5 dB bandwidth from 17.41 to 20.97 GHz in the pass band of the K-band and 30.09 to 33.24 GHz in the pass band of the Ka-band, with fractional bandwidths of 18.6% and 9.9%, respectively. Additionally, it satisfies -15 dB blocking performance across the entire X-band.

We propose dual-band filters made from a single-layer metasurface for operating on the 28 and 39 GHz bands in mm-wave fifth generation (5G) communication systems. We use the conformational space annealing method, a global optimization technique, and test the usefulness of dual-band metasurfaces through design, fabrication, and measurement with 2916 square-shaped pixels per unit cell defined on a single layer. Each square-shaped pixel has only two states: either filled with metal or free of metal. We were able to obtain transmissive metasurfaces capable of performing three different functions, bandpass/bandpass, bandpass/band-stop, and band-stop/bandpass, by setting three different objective functions applicable to 5G communications. The designed filters have insertion loss of less than 3 dB in the passband region and greater than 20 dB in the stop band region. As a proof-of-concept, we fabricate very thin (50 μm) single-layer metasurfaces etched on only one side of a single dielectric substrate combined with a 5G antenna array to demonstrate high-frequency-selectivity with dual-band operation experimentally.