A Hole Lot of Focus: Silicon Learns to Bend Ultraviolet Light

A team at the University of Stuttgart has made a flat patch of silicon that reflects and focuses ultraviolet light, by etching microscopic elliptical holes into it. The trick matters because silicon, the material of nearly every computer chip, normally eats UV light rather than steering it.

The work, titled "UV Mie-void reflective metasurfaces in silicon," was published in the peer-reviewed journal Advanced Photonics ^[1][3], with S. Arslan and colleagues as authors ^[2][3]. EurekAlert lists the publication date as November 2, 2026 ^[3]. The study is a laboratory demonstration of optics. It is not a finished sensor, and nobody has yet shown it inside a phone or on a satellite ^[1][2][3].

But the idea is a tidy one, and it answers a question that has bothered optics designers for a long time. Here is how it works, one question at a time.

Why is ultraviolet light so hard to steer?

Ultraviolet is light with a wavelength shorter than violet, beyond what the eye can see. The team worked at 325 nanometers. A nanometer is a billionth of a meter, so that is a few hundred times narrower than a human hair.

Short wavelengths are useful. They can resolve finer detail, and many biomolecules have characteristic UV absorption signatures, which the SPIE release notes could make compact UV optics useful in sensing ^[2][3]. SPIE is the optics society that publishes Advanced Photonics and issued the release.

The trouble is the materials. Silicon absorbs ultraviolet light strongly ^[1][2][3]. Conventional flat optics confine light inside little solid structures, and in silicon the light would be gobbled up before it could do anything useful.

Yet silicon is the platform the whole semiconductor industry knows how to pattern, at scale and cheaply. That is the pull in two directions: the best-understood material for making tiny things is a bad one for UV.

What is a metasurface?

A metasurface is a flat layer covered in nanoscale features that control light. Think of a lens. A glass lens focuses because its curve makes light at the edge travel a different distance than light at the center. A metasurface gets the same result without any curve. Each tiny feature delays the light that hits it by a chosen amount, and across the whole surface those delays add up to a focus.

The delay is called phase, and the pattern of delays is called a phase profile. Design the profile right and a flat sheet can behave like a curved mirror.

So what did the Stuttgart team do differently?

Most metasurfaces use solid pillars as their tiny building blocks. Light gets caught in each pillar and resonates there, building up and then re-emitting in a controlled way. That kind of trapping is called a Mie resonance. Pillars of silicon are exactly what you do not want for UV, because the light would sit inside the absorbing material.

The team turned the design inside out. Instead of pillars, they etched air-filled cavities into bulk silicon. They call them Mie voids. Much of the resonant light field sits inside the air cavity, where absorption is negligible, and not in the silicon around it ^[1][2][3].

In effect the light lives in the missing material. The silicon is the wall of the room, not the room itself.

There is a second point that sounds odd at first. The design does not remove silicon's absorption. The study description says that absorption actually helps, because it increases reflectivity at the boundary between air and silicon and so helps confine the light in the cavity ^[1][2][3]. A strong absorber makes a good wall. Metals reflect for a related reason, which is why a mirror can be made of something that is opaque.

What do the holes look like, and how do they focus?

The cavities are ellipses, about 470 nanometers long and 100 nanometers deep ^[1][3]. Lined up end to end, more than two thousand of them would stretch across a millimeter.

Each ellipse can be rotated. The rotation sets the phase of the reflected light, specifically light that is circularly polarized, meaning its electric field corkscrews as it travels. By changing the angle of the ellipses from place to place, the team encoded a focusing profile into the surface ^[1][3].

The holes themselves are all the same shape. What changes is which way they point.

The patterned area is about half a millimeter across, more than a thousand cavity lengths from edge to edge ^[1][2][3]. The researchers shone 325-nanometer light on it at a 45-degree angle. The surface sent the beam off at 90 degrees and focused it, behaving like a flat version of an off-axis parabolic mirror, the kind of curved mirror used to bring light to a point without blocking the beam ^[1][2][3].

How well did it focus?

The team made versions with focal lengths from five millimeters down to 1.25 millimeters. The smallest circular focal spots were about one micrometer across, roughly a thousandth of a millimeter ^[1][2][3].

Light cannot be focused to an arbitrarily small point. Waves spread out, and the smallest spot a given optic can make is called its diffraction limit. Up to a numerical aperture of 0.15, a measure of how steeply the optic bends light inward, the team's spots approached that limit ^[1][2][3]. In plain terms, for those designs the focus was about as sharp as physics allows.

They also tried a more exotic pattern. A spiral arrangement of phase produced beams carrying orbital angular momentum, whose intensity forms a doughnut shape with a dark center ^[1][3]. It is a showpiece of control. Such beams are of interest in optics research, and producing one in the UV from silicon shows the surface can do more than make a simple spot.

Why would anyone care outside an optics lab?

The manufacturing story is the quiet part of the news. According to the study descriptions, the metasurfaces were made on standard silicon substrates in a single lithography and etching step, and the sources say that is compatible with established semiconductor processing ^[1][2][3]. Lithography prints a pattern, etching cuts it in. Chip factories do both all day.

That is the route to the picture in the assignment: tiny, flat UV optics that could sit next to other electronics. The researchers and SPIE name compact high-resolution imaging, spectroscopy and sensing as possible uses ^[1][2][3].

A spectrometer splits light into its colors to read what a sample is made of. Shrinking the optics inside one is how instruments move from benchtops to handhelds, and a UV spectrometer that could read biomolecule signatures would be a useful thing to carry. The authors also say the approach may extend to deep and vacuum ultraviolet wavelengths and to other materials ^[1][2][3].

My view is that this is a good idea well executed, and that it is a long way from a phone. "May extend" is a prospect, not a result.

What has not been shown?

No count has been reported of how many devices were made, and no figure for how much of the incoming UV ends up in the focused beam, which is the number an instrument builder would ask for first ^[1][2][3]. No integrated sensor, spectrometer, phone component or satellite instrument has been reported either ^[1][2][3]. Nor has anyone shown the design working at wavelengths other than 325 nanometers ^[1][2][3].

The three accounts describe one study. The SPIE release, its EurekAlert posting and the Phys.org account all draw on the same material, so they are not three separate confirmations ^[1][2][3].

What to watch next

The questions that decide whether this leaves the lab are practical ones. How efficient is the mirror compared with the UV optics already on the market? Does it hold up from one fabricated batch to the next? Can it be joined to a UV detector on the same chip? And does the claim about deep ultraviolet survive an actual measurement at those wavelengths?

Silicon was supposed to be the wrong material for this job. The Stuttgart team did not change the silicon. They changed where the light lives, and the next papers will show how far that goes.