A new photodetector from Duke University senses light across the entire electromagnetic spectrum and generates a signal in just 125 picoseconds. It needs no external power, works at room temperature, and could form the basis of a new generation of multispectral cameras for medicine, agriculture, and space.
Engineers at Duke have built a photodetector that can “see” across nearly the entire electromagnetic spectrum, while operating at speeds comparable to traditional semiconductor sensors.
Instead of relying on the photoelectric effect, which limits cameras to visible and near-infrared light, the device uses a plasmonic metasurface and an ultra-thin pyroelectric layer to convert light into heat and then into an electrical signal.
The result is a detector that responds in just 125 picoseconds (~2.8 GHz), hundreds to thousands of times faster than conventional broadband detectors, without needing cooling, external power, or specialized conditions.
This breakthrough could enable real-time multispectral imaging and new applications in cancer detection, remote sensing, and environmental monitoring, potentially creating the first practical sensor that is both fast and spectrum-agnostic.
Your phone’s camera cannot see the heat coming off your skin. It cannot see the water content of a wheat field from a drone, or the chemical signature of a tumour through tissue, or the faint infrared glow of a satellite target. That is not a software problem. It is a physics problem. Semiconductor detectors, the kind inside every digital camera ever made, are fundamentally limited to a narrow slice of the electromagnetic spectrum that roughly matches what the human eye can see.
A new device built by engineers at Duke University does not have that limitation. It can detect light from any part of the electromagnetic spectrum. And it does so faster than almost anyone thought a heat-based detector could manage, generating an electrical signal in just 125 picoseconds. That is hundreds to thousands of times faster than conventional devices of the same type. The results were published in Advanced Functional Materials.
The Problem With Seeing Everything
Most photodetectors work through a process called the photoelectric effect. Light of the right frequency strikes a semiconductor and directly knocks electrons loose, generating an electric current that a computer reads as a signal.
This is fast, efficient, and well understood. It is also the reason your camera cannot see infrared radiation, terahertz waves, X-rays at useful sensitivities, or any other part of the spectrum outside the narrow visible and near-infrared window where semiconductor band gaps happen to fall.
To detect other wavelengths, engineers have traditionally turned to a different class of device called a pyroelectric detector. Rather than converting light directly into electricity, a pyroelectric detector absorbs the light as heat, and then converts that heat into an electric signal through the pyroelectric effect, a property of certain crystals that generate a voltage when their temperature changes.
Pyroelectric detectors can in principle respond to any wavelength of light because they are responding to heat rather than to specific photon energies. But they have a serious problem. Heat moves slowly. To absorb enough light to generate a detectable signal, conventional pyroelectric detectors need thick absorbing layers, and thick layers are slow layers. Commercial pyroelectric detectors typically operate in the nanosecond to microsecond range, orders of magnitude slower than semiconductor detectors. That slowness limits them to low-frequency, low-speed applications where the broad spectral coverage is useful but raw speed is not required.
The Duke team’s device attacks that tradeoff directly.
Silver Nanocubes Floating Above a Sheet of Gold
The key to the new detector is a structure called a plasmonic metasurface, a concept that Professor Maiken Mikkelsen’s lab at Duke has been developing for several years. The metasurface consists of precisely engineered silver nanocubes placed on a transparent polymer film positioned just 10 nanometers above an ultrathin sheet of gold.
When light strikes one of the silver nanocubes, it excites the electrons in the silver into a collective oscillation called a plasmon. This plasmon traps the light’s energy at the surface of the nanocube, concentrating it into an extremely small volume rather than allowing it to scatter or transmit through. The specific frequency of light that gets trapped depends on the size of the nanocubes and the spacing between them, which can be engineered during fabrication to target any desired wavelength.
This plasmonic trapping is extraordinarily efficient. Because essentially all of the incoming light’s energy is absorbed in a very thin surface layer rather than distributed through a thick absorbing material, only an extremely thin layer of pyroelectric material is needed beneath the metasurface to convert that concentrated heat into a voltage signal.
Thin pyroelectric layer means fast thermal response. Fast thermal response means fast electrical signal.
“Commercial pyroelectric detectors aren’t very responsive, so they need a very bright light or very thick absorbers to work, which naturally makes them slow because heat doesn’t move that fast,” said Professor Mikkelsen in a statement released by Duke’s Pratt School of Engineering. “Our approach cleverly integrates near-perfect absorbers and super-thin pyroelectrics to achieve a response time of 125 picoseconds, which is a huge improvement for the field.”
From Rectangle to Circle: The Geometry Change That Helped Break the Record
The latest version of the device incorporates several refinements over the team’s earlier work, which had demonstrated the plasmonic metasurface approach but had not yet been optimised for maximum speed or equipped with a setup capable of measuring it.
The first change was geometric. The previous metasurface was rectangular. The new design is circular. This increases the total surface area exposed to incoming light while simultaneously reducing the maximum distance any generated electrical signal has to travel from the edge of the detector to the readout circuitry at the centre. Shorter signal path means less time lost in transit.
The team also worked with collaborators to source even thinner pyroelectric films than those used in previous iterations, reducing the thermal mass that the absorbed energy needs to heat before a signal is generated. The electronic circuitry used to capture and relay the voltage pulses was redesigned to handle the faster signals without distortion.
Measuring the resulting speed without access to test equipment costing hundreds of thousands of dollars required a creative approach from PhD student Eunso Shin, who led the experimental work. Shin built a measurement setup using two distributed feedback lasers. The two lasers produce slightly different frequencies, and where those frequencies interfere they create a beating pattern whose frequency equals the difference between the two lasers. By tuning the lasers so that this beat frequency matched the operating range of the detector, Shin could directly measure how quickly the device responded to incoming light without needing an oscilloscope rated for picosecond timing.
The measurements showed the detector operating at speeds up to 2.8 GHz, corresponding to a signal generation time of 125 picoseconds.
“Pyroelectric photodetectors commonly operate in the nano-to-microsecond range, so this is hundreds or thousands of times faster,” Shin said in the Duke statement.
What 125 Picoseconds Actually Means
A picosecond is one trillionth of a second. Light travels about 3.75 centimetres in 125 picoseconds. At 2.8 GHz operating speed, the detector is generating and clearing signals nearly three billion times per second.
For context, this puts the device into a speed regime comparable to silicon semiconductor photodetectors, which are the gold standard for fast optical detection in communications systems. The critical difference is that silicon detectors work only at visible and near-infrared wavelengths. This device works at any wavelength the metasurface is designed to absorb, from the terahertz band through the far infrared, mid-infrared, near-infrared, visible, ultraviolet, and potentially into X-ray regimes, all from a single platform that requires no external power supply and operates at ordinary room temperature.
No cooling. No specialised atmosphere. No external bias voltage. Those are significant practical advantages over competing broadband detector technologies. Detectors that can sense mid-infrared radiation typically require cooling to cryogenic temperatures to suppress thermal noise. Terahertz detectors are notoriously difficult to build at room temperature with meaningful sensitivity. The Duke device sidesteps both problems through the plasmonic concentration mechanism.
What It Could Be Used For
Mikkelsen was direct about the application horizon. “When you get into the ability to detect lots of frequencies at once, you open the door to so many different things,” she said in the Duke statement. “Cancer diagnosis, food safety, remote sensing vehicles. Those are all still pretty far down the line, but that’s the direction we’re heading in.”
The near-term application most directly enabled by the speed improvement is multispectral imaging at video or near-video frame rates. Current broadband pyroelectric detectors are slow enough that they are typically used as single-pixel sensors or in very low-frame-rate imaging systems. At 2.8 GHz operating speed, the Duke device can in principle support much faster readout, enabling real-time imaging across multiple spectral bands simultaneously.
The team’s roadmap points toward embedding the pyroelectric material and its readout circuitry directly into the gaps between the silver nanocubes and the gold film, rather than placing them in a separate layer beneath the metasurface. This would further reduce the thermal path length and push operating speeds higher still. A second direction involves fabricating multiple metasurfaces with different nanocube geometries on a single chip, allowing simultaneous detection of distinct wavelengths and polarisations without any mechanically tuned filters or beamsplitters.
For remote sensing applications, particularly on satellites, drones, and interplanetary spacecraft where power and mass budgets are tightly constrained, a broadband detector that operates at room temperature without external power is an obvious target for further development.
Sources
The paper, “Metasurface-Enhanced Thermal Photodetector Operating at Gigahertz Frequencies,” was published online December 11, 2025 in Advanced Functional Materials by Eunso Shin, Maiken Mikkelsen, and colleagues at Duke University’s Department of Electrical and Computer Engineering.
Quotes in this article are drawn from a press release issued by Duke University’s Pratt School of Engineering on March 4, 2026.
Ray Jackson holds a BSc in Electrical Engineering from the University of Manitoba and a PhD in Physics from Carleton University. His reporting interests include Current and Future Technologies, Engineering and Artificial Intelligence.