A thermal photodetector that responds in 125 picoseconds could see every wavelength at once

Duke University

Traditional digital cameras see visible light. That is all. The silicon semiconductors at their core convert photons into electrical signals, but only within the narrow band of wavelengths that human eyes can also detect. Infrared, terahertz, ultraviolet: all invisible to the sensor, each requiring a different, specialized detector to capture.

Pyroelectric detectors take a fundamentally different approach. They generate electrical signals when heated by absorbed light, which means they can respond to any wavelength that carries enough energy to warm the material. The problem has always been speed. Making enough heat to produce a usable signal has required thick absorbers, and thick absorbers are slow because heat does not travel fast through bulk material. Commercial pyroelectric detectors operate in the nanosecond-to-microsecond range. Digital cameras operate in the picosecond range. That gap has kept thermal detectors out of high-speed imaging applications.

A team at Duke University has now closed that gap. Their pyroelectric photodetector, described in Advanced Functional Materials, operates at 2.8 GHz, producing an electrical signal in just 125 picoseconds. That is hundreds to thousands of times faster than any previous pyroelectric device.

Trapping light in 10 nanometers

The key is a structure called a metasurface. Precisely tailored silver nanocubes sit on a transparent film only 10 nanometers above a thin layer of gold. When light hits a nanocube, it excites the silver's electrons and traps the light's energy through a phenomenon called plasmonics. The frequency of light captured depends on the nanocubes' size and spacing, which can be tuned during fabrication.

This plasmonic trapping is so efficient that it generates enough heat in an extremely thin layer of pyroelectric material beneath the metasurface to produce a readable electrical signal. Because the pyroelectric layer is ultrathin, heat does not need to travel far. The entire thermal cycle, absorb light, heat up, generate current, happens in picoseconds.

The approach was first demonstrated by Maiken Mikkelsen's lab at Duke in 2019, but at that time the experimental setup could not measure the device's speed. Getting a speed measurement without spending hundreds of thousands of dollars on ultrafast electronics required ingenuity from PhD student Eunso Shin, who designed a setup using two distributed feedback lasers whose signals beat against each other at the device's operating frequency.

Design refinements that matter

The new version incorporates several changes from the 2019 prototype. The light-absorbing metasurface is circular rather than rectangular, maximizing exposure while minimizing the distance the electrical signal must travel to be read out. The team sourced even thinner pyroelectric layers from collaborators. And they upgraded the circuit design for relaying the electrical signals.

The device requires no external power source. It operates at room temperature. And it is thin enough to be integrated directly onto a chip. Those characteristics matter for applications where size, weight, and power are constrained: satellites, drones, portable instruments.

From one wavelength to many

The current device detects a single frequency of light, determined by the nanocube geometry. But the team is working on multi-frequency versions that layer several metasurfaces, each tuned to a different wavelength, into a single detector. A device that simultaneously captures visible, infrared, and terahertz light, and measures each one's polarization, would be a multispectral camera of a kind that does not currently exist.

The potential applications span a wide range. In medicine, multispectral imaging could aid skin cancer detection by revealing tissue properties invisible to the eye. In agriculture, satellite- or drone-mounted sensors could map crop health across large areas in real time, identifying which fields need water or fertilizer. In food safety, spectral signatures could flag contamination.

All of those applications are, as Mikkelsen puts it, still pretty far down the line. The physics works. The fabrication challenges of scaling from a laboratory device to a manufacturable product are substantial. But the team believes they can improve the already record-setting speed further by integrating the pyroelectric material directly into the gap between the nanocubes and the gold layer, and they are working to identify the fundamental kinetic limit of pyroelectric detection.

The question is no longer whether thermal detectors can be fast enough. The question is what you could see if they were.