As industrial systems move toward more capable sensing, imaging, and wireless technologies, component design at terahertz frequencies presents a distinct engineering challenge: structures are physically small, while material losses and manufacturing tolerances can have a substantial effect. A planar terahertz antenna array fabricated from gold-plated, micromachined silicon offers one approach to that challenge. Demonstrated at 300 GHz, the design uses air as the electromagnetic medium and a channelized coplanar waveguide (CCPW) to create a broadside-radiating array.
The work is a component-level antenna demonstration, not evidence of a deployed industrial system. Its value for automation engineers and technical decision-makers lies in the design and fabrication strategies it illustrates, particularly for compact, highly integrated equipment where antenna packaging and signal losses matter.
Why use an air-filled antenna at terahertz frequencies?
Terahertz systems are of interest for applications including broadband wireless communications, detection, and high-resolution imaging. At these frequencies, dielectric loss can become a significant design concern. In a conventional structure, electromagnetic energy may interact with dielectric substrate material. The antenna described here instead confines its fields to air channels: silicon provides the mechanical structure, but its exposed surfaces are covered with gold so the electromagnetic wave does not encounter the silicon dielectric.
This distinction is central to the design. The antenna is not simply made from silicon; it uses micromachined silicon as a supporting framework for an air-filled electromagnetic path. The authors’ reported simulations show radiation efficiencies from 87.9% to 97.8% across the simulated band, alongside broadside gains from 10.6 to 13.8 dBi. These are results for the reported antenna prototype and should not be read as general performance guarantees for other terahertz designs.
How the CCPW structure creates an antenna array
The radiating structure begins with an air-filled CCPW. In its unmodified form, the fringing electric fields on opposite sides of the center conductor have opposing directions. The resulting field symmetry supports a resonant condition but does not provide the desired in-phase radiating aperture.
The design changes that field pattern by placing metallic blocking stubs periodically and alternately in the gaps beside the center conductor. The stubs suppress selected portions of the field, leaving slots where the remaining fields point in the same direction. Each section behaves like an open cavity radiating from its open side; arranged in sequence, the sections form an array with an in-phase aperture and broadside radiation.
A four-unit array was used to demonstrate the principle. The radiating slots are spaced by 0.5 mm, described in the design as half a free-space wavelength at 300 GHz. The field distributions were similar at the examined frequencies of 250, 275, 300, and 325 GHz, indicating that the intended in-phase behavior was maintained across that range in the modeled structure.
Hierarchical micromachining reduces the layer count
The array comprises two silicon layers, each processed as two sublayers. This hierarchical approach allows different features to be formed from opposite sides of a silicon piece rather than requiring a separate substrate layer for every structural level. Compared with the researchers’ earlier design, the approach reduced the required silicon substrate layers by one.
Fabrication combines anisotropic wet etching, through-wafer dry etching, gold plating, optical alignment, and wafer bonding. Wet etching forms cavities with angled sidewalls; dry etching creates the radiating slots with nearly vertical sidewalls. The process sequence matters for one of the layers: the radiating window is made by wet etching before the slots are formed by dry etching. The other layer’s air space and coupling slot are both produced by wet etching.
After etching, the surfaces and sidewalls are electroplated with gold, then the two layers are aligned and bonded. The resulting assembly includes the air-filled antenna and a coupling slot for feeding it. In the test configuration, a standard 300 GHz WR-3 waveguide supplied the input, with a copper spacer connecting the waveguide and antenna and helping protect the silicon wafers.
What the prototype measurements show
In simulation, the −10 dB impedance bandwidth was 71.1 GHz, spanning 238.4 to 309.5 GHz. The measured bandwidth exceeded 117.7 GHz, from 222.3 GHz to beyond the test system’s 340 GHz upper limit. The measured radiation patterns and gains broadly agreed with the simulations, although the authors noted discrepancies associated with the measurement environment, alignment, and available receiving and calibration antennas.
These measurement conditions are important when interpreting the results. The reported prototype confirms the feasibility of the structure and fabrication strategy, but the measured bandwidth should not be treated as a complete characterization beyond the equipment’s available frequency range. Similarly, the test setup’s limitations make the measured results less direct than an idealized laboratory comparison.
Implications for integration-focused design
For automation and instrumentation teams evaluating emerging high-frequency components, the work highlights several practical design considerations:
- Manage dielectric interaction: an air-filled path can address dielectric loss by keeping the electromagnetic field away from the silicon substrate.
- Design fabrication into the architecture: hierarchical etching can create multiple structural levels while reducing the number of silicon layers.
- Account for assembly and alignment: optical alignment and wafer bonding are integral to producing the final antenna, not secondary packaging details.
- Plan the system interface: the demonstrated prototype uses a waveguide feed, while the paper notes that a ground-signal-ground input could support integration with other ICs within a package.
The last point connects the antenna design to antenna-in-package (AiP) development. Silicon’s compatibility with integrated circuits and the antenna’s compact structure may support future packaged implementations, but the reported work demonstrates the antenna concept rather than a complete integrated package.
Conclusion
This planar terahertz antenna array combines field shaping in a CCPW with gold-plated silicon micromachining to create an air-filled radiating structure. Its two-layer, hierarchically processed construction addresses both electromagnetic loss and fabrication complexity, while prototype testing verifies operation around 300 GHz. For technical decision-makers, the practical takeaway is a design pathway worth understanding for future high-frequency integration—not a ready-made industrial solution. System-level adoption would still require integration, interface, and application-specific validation.