Infrared cameras can detect things invisible to the naked eye, like gas leaks from pipelines, chemicals in the air, or heat escaping from buildings. However, advanced infrared light sensing still requires costly and large systems. Researchers at MIT have developed a chip-based optical device capable of dynamically controlling incoming infrared light, functioning as a tunable lens that enhances infrared camera capabilities. Each microscopic pixel of the lens can independently adjust infrared light, allowing for focus changes and detection of various signals without moving parts.
This innovation is detailed in a paper published in Nature Communications. The researchers demonstrated a lab-scale version using mostly standard manufacturing techniques in a semiconductor chip factory, indicating potential scalability for industrial production. This technology could lead to compact, adaptable infrared cameras for thermal imaging, chemical analysis, pollution monitoring, and new optical computing methods.
“This could provide more data as we explore space or assist with environmental protection by monitoring specific atmospheric compounds,” states first author Cosmin-Constantin Popescu PhD ’25. “Thermal imaging and military applications using night vision goggles are other possibilities. Many organic molecules absorb in the mid-infrared wavelength, and this system could detect them.”
Popescu collaborated with MIT PhD students Maarten Robbert Anton Peters and Khoi Phuong Dao; Dynasil scientists Oleg Maksimov and Harish Bhandari; University of Central Florida PhD candidate Kathleen Richardson and scientist Rashi Sharma; University of Washington Professor Arka Majumdar; Korea Advanced Institute of Science and Technology Associate Professor Hyun Jung Kim; MIT postdoc Rui Chen; Luigi Ranno PhD ’25; Brian Mills ’20, PhD ’26; Draper Laboratory scientist Dennis Calahan; MIT principal investigator Tian Gu; and Juejun Hu, MIT’s John F. Elliott Professor of Materials Science and Engineering.
In recent years, researchers have developed techniques to dynamically control light using microscopic patterns on transparent materials called “metasurfaces,” leading to more compact, programmable cameras and advanced optical devices. Hu’s group at MIT has been experimenting with metasurfaces that transition from solid to liquid when heated, enabling control over light interactions. In 2021, Hu and collaborators designed a miniature lens that could adjust its focus through such phase changes.
Their previous device functioned reliably but could only adjust focus uniformly across the material, typical of most metasurfaces. The new study aims to enhance this by allowing independent light control at each microscopic pixel. “Most active metasurfaces need wiring to each pixel, complicating wire routing,” Hu explains. “The best solution so far is one-dimensional pixel control with multiple wires.”
The researchers wanted their system to work with mid-infrared light, useful for detecting heat signatures and molecules like methane and propane. Mid-infrared devices already find gas leaks and study Earth’s atmosphere, with applications in defense and aerospace.
To develop their system, they adapted a method from displays, using two layers of copper wires arranged perpendicularly. Below the wires, doped silicon generates heat at wire intersections and sits atop the phase-change material. This heat alters each pixel’s structure, changing its interaction with infrared light. The silicon also includes a diode selector to prevent current leakage to neighboring pixels.
“We calculated that this architecture allows scaling to potentially millions of pixels without current issues,” Hu says. “The crossbar architecture is a scalable way to increase pixel-level metasurface switching. While not invented by us, it’s the first time used for active phase-change metasurfaces to achieve pixel-level control.”
Working with MIT.nano and a semiconductor chip factory, the researchers created a two-dimensional system with a 6-by-6 metasurface pixel array. Their system switched on and off reliably during tests. “This mesh architecture proved very resilient,” Popescu says. “You want it to switch many times, maybe tens of thousands, not just once.”
Integrating their design into existing semiconductor manufacturing could help transition from prototype to production. “Scaling requires a consistent process, making chip foundry manufacturing crucial,” Hu says. “Working with a semiconductor foundry offers well-defined process control and allows integrating components into a single efficient process.”
The team is expanding their pixel array and developing more robust systems to capture more infrared data. “When imaging, prior knowledge of what you’re seeking can help,” Hu notes. “You might look for a person in darkness or specific image features, and this system can highlight those.”
Hu mentions metasurfaces have been used to simulate computational neural networks driving AI systems, though practical applications may be distant. “This could enhance optical computing by encoding network weights in neural networks,” Hu explains. “Light interacting with the metasurface encodes information for computational results. Researchers have used this to emulate complex neural networks.”
The research received support from the U.S. Air Force, the U.S. National Science Foundation, the National Research Foundation of Korea, and the Draper Scholar Program.
Original Source: news.mit.edu
