How to solve the THz signal extraction challenge with optics and enable hardware manufacturability

At THz frequencies, efficient signal extraction from a semiconductor wafer becomes an enormous challenge. Without a solution, the technological concepts won’t survive outside the lab. Performance will degrade and calibration won’t provide a fix.

Whatever the solution, manufacturability is crucial, determining whether our concepts become real systems or remain demonstrations. We must treat manufacturability as a first class constraint.

Solving the challenge enables efficient production at array scale and smaller, lighter and higher-performance antenna and link solutions. They are useful for example in drone communications: longer links allow drone operation farther from the front line while reducing payload.
 

For a long time, RF engineers could take one thing for granted. The antenna was large, the chip was small, and getting the signal from one to the other was mostly an engineering detail. However, as frequencies increase, that intuition starts to change.

At mmWave and terahertz frequencies, antennas shrink with wavelength, while silicon stubbornly stays the same size. The result is a counterintuitive reversal: The antenna can become physically smaller than the chip it is attached to. This offers an obvious benefit, on chip-antennas, that can reduce the size of RF systems, but these are not traditionally so easy to implement.

The semiconductor substrate itself starts to dominate how electromagnetic energy behaves. At that point, extracting RF power out of the system, efficiently and predictably, becomes the real challenge. This is not yet a defining problem in the gigahertz range, but once wavelengths become short enough, it becomes unavoidable.
 

At first glance, this seems like an antenna problem. That is certainly how it is often approached. Improve the radiator, tweak the matching, adjust the geometry.

In practice, at mmWave and especially at THz, energy leaks into the substrate. Waves propagate along unintended paths. Loss mechanisms that were once small begin to dominate. Externally, the antenna may look fine, while internally a significant fraction of the energy never reaches free space.

The assumptions of classical RF design break down. Surprisingly, wave propagation, reflections, and impedance transitions start to resemble optical behaviour. Thinking in terms of rays, interfaces, and coatings suddenly becomes useful.

In optics, anti reflection coatings are well established. A carefully engineered layer, with thickness on the order of a quarter wavelength, can reduce reflections and guide energy where it is supposed to go.

For mmWave and THz hardware, we can apply the same principle.

The challenge is not only shaping the antenna but controlling how energy transitions from silicon to air. That transition happens across materials, interfaces, and thicknesses all comparable to the wavelength itself. In this regime, the substrate, passives, and whatever sits on top of them form a coupled wave system, where each component affects the others.

Here, engineered material layers enter the picture. Not as add-ons, but as a way to control energy propagation.

The dimensions between RF and visible-light optics are too small for conventional machining, yet too large and costly for classical microfabrication. Many elegant laboratory solutions fail to scale economically.

This is why engineered tapes are compelling. The material systems already exist in semiconductor manufacturing. For example, dicing tapes are routinely used during wafer sawing to hold the chips in place. Their thicknesses are well controlled.  their electromagnetic behaviour can be engineered through material and thickness selection to achieve the desired wavelength/frequency.

Most importantly, engineered tapes live in a manufacturing regime dealing with high volumes, large areas, repeatability and cost.

Seen this way, the tape is not a trick. It is a pragmatic way to implement quarter-wavelength-style control in a frequency range where traditional RF structures and optical fabrication methods struggle.

Another option is to shape the semiconductor itself, so it gradually transforms the impedance seen by the radiating field. In this view, the silicon is part of the electromagnetic structure guiding energy out of the chip, not just a carrier of circuitry.

These approaches work best together, reinforcing each other. Silicon shaping controls how energy propagates inside the structure, while materials above it control how energy exits into free space.

This becomes especially important at array scale, where even small inefficiencies add up rapidly. Power consumption increases, thermal margins shrink and performance degrades. Three-dimensional wavelength-scale silicon structures provide powerful control, but they are sensitive to process variation across large wafers and dense arrays. Planar material layers shift part of that control into a manufacturing regime where thickness, uniformity, and composition can be managed with highly scalable and repeatable processes.

Tape complements geometric shaping by providing repeatable wafer-level control at the silicon–air boundary.

At mmWave and THz, antennas can become smaller than chips. Signal extraction is no longer a detail but a prerequisite for scalable, manufacturable systems.

Manufacturability favours solutions that work with existing processes, materials, and handling methods. Using standard tapes to enable mmWave and THz chips is extremely practical. In that sense, manufacturability is not a limitation. It determines which physical ideas become real systems.
The air interface is no longer defined by a single component. It is defined by how well the entire hardware ecosystem works together to extract energy from silicon.

At VTT, we are doing this development in our laboratories. We offer licensable solutions for controlling wireless signals at these challenging frequencies as well as our expertise for companies in need of deployable, secure, high-capacity communication systems. Get in touch with us to explore how we can help.