Skip to Content

Better thermal imaging, thanks to a small device

With help from a small transistor, a team of researchers led by Professor Fengnian Xia figured out a way to make a type of thermal imaging technology dramatically more accurate. The results are published in Nature Sensors

Why it matters

Robots, drones, self-driving vehicles and other autonomous devices rely on thermal sensing and imaging to navigate the spaces they travel in. It’s also used in many other technologies, including night vision, remote thermometers, and rescue operations.

The problem

Thermal imaging devices detect the heat emitted by objects and then create a visual heat map. There are two common types of thermal imaging technology; both have their downsides. One uses photon detection, which is very accurate but also needs to be kept in extremely cold temperatures. It’s also very expensive, and is typically used in military purposes and other specialized usages. The more commonly used form of thermal imaging technology uses devices known as microbolometers, which convert heat to a detectable electrical signal. Portable and cheap, they’re used in several industries, including firefighting and surveillance. However, its performance significantly lags behind that of cooled photon detectors. That’s largely because of limitations of the materials typically used for this technology, vanadium oxide and amorphous silicon. They both have a low temperature coefficient of resistance (TCR) - that is, how much electrical resistance changes in a material in response to variations in heat.
 

How they solved it

The researchers applied a two-terminal NPN transistor, a device that amplifies a circuit. Jiazhen Chen, a PhD student and lead author of the study, explained that rather than changing the materials used for the technology, the research team used the transistor to enhance the existing material. Essentially, the transistor creates a carrier feedback loop to enhance its temperature dependence and further makes it programmable. This increases the materials’ TCR from about 10% to up to 150% per Kelvin, allowing the devices to better detect heat emitted by objects in its surroundings. 

Next up, Chen said, the researchers will build actual devices using the technology, specifically ones with additional components that can detect mid-infrared heat. The researchers will also implement this technology on a silicon platform, a material widely used in computer chips.

Image: Artistic view of the carrier feedback process in a transistor, which leads to strong and programmable temperature dependent properties. Courtesy of the Xia Lab.

More Details

Published Date

Jun 25, 2026

Featured Departments