Electromagnetic waves spanning 0.1 to 10 terahertz occupy a peculiar boundary region that combines the penetrating quality of radio waves with the directional propagation of light. This frequency band lies beyond the millimeter-wave range used in current fifth-generation (5G) mobile communication systems, and research continues to advance it as a promising candidate for future ultra-high-speed, high-capacity wireless communication and high-resolution sensing. However, efforts to integrate these waves into everyday practical devices have long been hindered by the physical constraints of both the generating devices and the optical systems around them.
At the root of the problem lies a mismatch between the output characteristics of terahertz sources and the shape of the resulting beam. Semiconductor devices such as resonant tunneling diodes (RTDs) have been developed as ultra-compact elements capable of continuous-wave terahertz oscillation at room temperature, but their physical dimensions are on the order of just a few millimeters. Semiconductor sources that excel at miniaturization generally suffer from low output power, and the electromagnetic waves radiated from the device's antenna aperture spread rapidly outward in a radial pattern. To deliver waves over a distance while maintaining sufficient power density, collimation—aligning the diverging wavefront into a parallel beam—is essential.
Conventional optical systems have addressed this by placing a thick bulk lens, machined from natural materials such as high-resistivity silicon, in front of the source to narrow the beam. However, because natural materials have refractive indices of only around 3.4, achieving sufficient refractive power requires the lens itself to be several millimeters to several centimeters thick. Even though the device itself is on the millimeter scale, the surrounding optical components end up far larger, making it structurally difficult to implement the system as a compact integrated module.
Metasurface technology, which uses artificial microstructures to freely manipulate the behavior of electromagnetic waves, has drawn attention as a way to break through this spatial constraint. A research group led by Associate Professor Kento Suzuki of the Institute of Engineering at Tokyo University of Agriculture and Technology (TUAT), together with Rohm Co., Ltd., designed a gradient-index (GRIN) metalens consisting of a two-dimensional array of metal patterns on an ultra-thin resin film, and conducted experiments combining it with a compact 0.3-terahertz-band oscillator.
Geometric arrangement of meta-atoms that control spatial refractive index
The physical thickness of the newly developed lens is just 51 micrometers—about half the thickness of ordinary copy paper. This thin film transforms the wavefront of terahertz waves as they propagate through space. What makes this possible is the design of the meta-atoms, the microscopic artificial structures that make up the lens.
In a uniform dielectric material found in nature, the refractive index inside the material is constant, and light is bent by creating an optical path difference through the curved shape of the lens surface. By contrast, the GRIN (Gradient-Index) structure adopted in this study is flat, yet designed so that the refractive index varies continuously from point to point. By setting the refractive index to increase in a stepwise manner from the outer edge of the lens toward its center, waves passing through the center are delayed, bringing them into phase alignment with waves passing through the outer regions. This precise correction of phase differences transforms a spherically spreading wavefront originating from a point source into a plane wave that propagates forward.
To achieve this optical property, the research group used a low-loss resin material, cyclo-olefin polymer (COP) film, as the substrate. On both the front and back surfaces, a total of 91 circular copper patterns are arranged periodically. By finely varying the diameter and periodicity of the metal patterns, the effective electromagnetic parameters of each region were independently controlled.
For terahertz waves with a wavelength of about 1 millimeter, the metal patterns making up the meta-atoms are fabricated at a scale sufficiently smaller than the wavelength. This structure was optimized so that both the permittivity and permeability simultaneously exhibit high values in response to the incident electromagnetic wave. As a result, at the operating frequency of 0.312 terahertz, the lens center achieved a refractive index of 6.2—a value that surpasses naturally occurring materials. Despite this large refractive index variation, reflectance was suppressed to 1.1%, while transmittance reached 94.3%. The design minimizes reflection loss at the interface while achieving powerful wavefront transformation within an extremely thin space.
The structural design methodology underlying this approach has been established as intellectual property through Japanese Patent No. 6596748, U.S. Patent No. 10686255, and Japanese Patent No. 7315983, among others.
Polarization symmetry and far-field pattern measurements
The developed metalens was evaluated by placing it 10 millimeters in front of a resonant tunneling diode (RTD) source oscillating continuously at 0.312 terahertz. The measurement system scanned the far-field radiation pattern emitted from the device, comparing the spatial intensity distribution with and without the lens.
The experimental results showed that terahertz waves passing through the metalens converged toward the central axis, with the beam's spread angle suppressed. Compared with the radiation pattern of the device alone, directivity in the main beam direction improved by up to 4.9 dB, corresponding to a factor of 3.1 (the directivity of the standalone device and the absolute values being compared are presented in the published materials as a relative differential ratio). This confirmed that the electromagnetic energy emitted by the oscillator was efficiently concentrated in the forward direction rather than dispersing into the surroundings.
| Development Generation / Paper | Distance from Source (Focal Length) | Directivity Improvement Factor (Note) | Polarization Dependence | Lens Thickness | Meta-Atom Structure |
|---|---|---|---|---|---|
| 2021, Optics Express | Several millimeters | 4.2× | Yes (polarization-dependent) | Tens of micrometers | Symmetric cut metal wire pairs |
| 2021, Applied Physics Express | 1.0 mm | 3.0× | Yes (polarization-dependent) | 24 μm | Fine metal patterns |
| February 2026, Applied Physics Express | 1.0 mm | 1.7× | No (polarization-independent) | 51 μm | Circular metal discs |
| This Work (IEEE PTL) | 10 mm | 3.1× (4.9 dB) | No (polarization-independent) | 51 μm | Front-and-back circular copper patterns |
Note: The directivity improvement factors in each paper are measurement results from the research group; the absolute values of the comparison baseline and comparison target are not directly presented in the published materials.
Another property confirmed in this measurement was polarization-independent operation. When the radiation pattern was re-measured with the lens rotated 90 degrees about the optical axis, the same directivity improvement was observed as in the initial orientation (0-degree rotation).
For lenses with anisotropic response to the vibration direction (polarization) of electromagnetic waves, the oscillation polarization axis of the source and the geometric orientation of the lens must be rotationally aligned with micron-level angular precision. However, because this lens operates through the rotational symmetry of the circular patterns placed on both its front and back surfaces, it does not depend on the polarization direction of the incident wave. This eliminates the need for strict rotational alignment during implementation, offering the advantage that the optical system can be assembled using only planar XY positioning.
The mechanical trade-off between integration distance and refractive power across design generations
The pursuit of terahertz wave control using metasurfaces has been a continuous effort to overcome the physical trade-off between miniaturization and optical performance. Comparing the series of studies previously published by the research group clarifies the engineering position occupied by this latest design.
The early GRIN metalens reported by the same group in Optics Express in 2021 achieved an extremely high refractive index exceeding 10, successfully boosting the directivity of the RTD oscillator by a factor of 4.2. However, the meta-atoms used in this design were rectangular cut metal wire pairs, which structurally responded only to linearly polarized light in a specific direction. In exchange for high gain, this design carried the implementation constraint of requiring precise polarization alignment.
A study aimed at eliminating polarization dependence while pursuing extremely close-range placement was published in Applied Physics Express (APEX) in February 2026. In this system, circular meta-atoms were adopted to achieve polarization independence, shrinking the distance between the oscillator and the lens to just 1.0 millimeter. However, bending a wave that spreads sharply over such an ultra-short focal length of 1.0 millimeter requires forming a steep refractive index gradient from the lens center to its outer edge. This demand for a steep phase change compressed the margin available for structural design; while the beam's half-power width was narrowed from 53 degrees to 20 degrees, the directivity improvement factor remained limited to 1.7×.
The achievement reported this time in IEEE Photonics Technology Letters (PTL) rebalances this trade-off by resetting the distance between the source and lens to 10 millimeters. By securing a longer propagation distance to relax the curvature of the wavefront, the metalens design optimized the refractive index variation of the meta-atoms while maintaining polarization independence, restoring a high directivity gain of 3.1× (4.9 dB). This represents an engineering compromise that recovers practical beam-focusing power while preserving extreme thinness.
Demonstrated boundaries and challenges toward future communication implementation
In evaluating the results of this research, it is essential to clearly distinguish what has been demonstrated as a laboratory-scale proof of principle from what remains unresolved.
First, the 3.1× figure obtained in this study is an indicator of the spatial concentration (directivity) of the far-field radiation pattern measured inside an anechoic chamber; it does not directly indicate the improvement in throughput or data transmission rate in an actual wireless communication link. No transmission/reception testing as a communication system, bit error rate (BER) measurement, or transmission evaluation using a modulated high-frequency signal input has been conducted.
Second, the demonstration experiment was carried out at a single frequency of 0.312 terahertz. Because metasurfaces control electromagnetic parameters through structural resonance, they tend to have a narrow operational bandwidth (fractional bandwidth). How far this lens can maintain uniform wavefront transformation across the ultra-wideband transmission of tens of gigahertz required for future terahertz communication remains to be confirmed through future broadband measurements.
Further consideration must also be given to the overall system size. Although the physical thickness of the lens itself is extremely thin at 51 micrometers, achieving the desired collimation effect requires securing a 10-millimeter propagation space between the lens and the oscillator. Incorporating this into a communication device module will require packaging technology capable of maintaining this 10-millimeter gap with low loss and high precision.
This research was supported by the Japan Science and Technology Agency (JST) FOREST Program (JPMJFR222I) and Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (JP24K01376, JP24K21617). Looking ahead, the authors suggest possibilities for applying this approach to various continuous-wave terahertz sources beyond RTDs, as well as extending it to higher-directivity antenna elements and highly functional flat optical elements capable of forming optical vortices and Airy beams. The outline of the role that flat metalenses can play as a foundational technology for accurately delivering weak radio waves in the ultra-high-frequency band to where they are needed is gradually coming into view.