How a 0.82-volt barrier gave night vision its color

Night vision goggles flatten the world into green. The reason is not that infrared light lacks color information. It is that conventional devices throw that information away, converting everything into shades of brightness. A team at the Beijing Institute of Technology, led by Xin Tang and Ge Mu, has now built a device that keeps the information and turns it into full color, not with a camera or image processor but with a physical energy barrier inside the detector itself. The result, described in Science Advances on 29 July 2026, is an eyeglass that renders infrared scenes in hue, and whose upconverted light can even drive living visual cells.

Vision begins when a photon strikes a light-sensitive pigment in the retina and forces it to change shape, a step that requires a minimum photon energy of roughly 1.6 electron volts. Infrared photons, with wavelengths longer than 700 nanometers, carry less than that, which is why they are invisible to us. That leaves more than half of the Sun’s radiant energy outside human perception, along with everything that emits or reflects heat. Existing night vision devices capture some of that invisible light, but they render it through a photodetector wired to a visible display, which encodes only how much light arrived. A warmer object glows a little more, a cooler one a little less, all in the same greenish tint. The human eye, however, is far better at distinguishing subtle differences in hue than in brightness, so the monochrome approach leaves most of the eye’s sensitivity unused.

The Beijing team’s detector solves the problem by making wavelength itself selectable. Their sensing layer is made of mercury telluride colloidal quantum dots, crystals about 4 nanometers across, so small that quantum confinement breaks their energy levels into discrete steps. In bulk semiconductors, energy bands are continuous and every infrared photon produces roughly the same response. In a quantum dot, photons of different wavelengths excite different electronic transitions: the longest wavelengths the team tested, around 2 micrometers, carry just enough energy to lift an electron across the fundamental gap, while shorter infrared wavelengths open additional higher-energy transitions, and intense light can even kick loose more than one electron-hole pair per photon. The result is that the number of charge carriers leaving the quantum dot layer encodes both the wavelength and the intensity of the arriving light.

Converting that carrier count into color is where the physics does the coding. The device stacks two emissive layers on the path the holes travel: a red-emitting phosphor layer close to the detector and a cyan-emitting layer beyond it, separated by an energy barrier of 0.82 electron volts. When few holes arrive, from dim or long-wavelength light, they are all captured by the red layer, and the device glows red. As the carrier flux rises, the red emitter’s trapping sites saturate and holes begin crossing the barrier into the cyan layer, so the two emissions mix and the perceived color shifts. Because the shift is tied to carrier number rather than a lookup table, the output color encodes wavelength and intensity simultaneously. The paper reports a photon-to-photon upconversion efficiency of 3.85 percent, luminance above 700 candelas per square meter, and the ability to distinguish infrared power differences more than two orders of magnitude finer than a single-color design.

We're building an independent news platform focused on facts over headlines. Join us by supporting our work.

Support 1ban.news

The team demonstrated the approach in a wearable form: a semi-transparent eyeglass weighing 23 grams with an active viewing area of 3.57 square centimeters. Illuminated by shortwave infrared, it projected sharp, color-coded images of test patterns and moving objects that a standard camera could capture, while ordinary visible light still passed through the semi-transparent stack. In principle the same device could switch between an augmented-reality mode, overlaying infrared information on normal sight, and an immersive infrared-only mode. The more ambitious claim is biological. The researchers bound the upconverters to cells producing channelrhodopsin-2, a light-sensitive protein used to make neurons responsive to blue light, and infrared illumination triggered photocurrents in those cells. Infrared pulses delivered through the device produced clear electroencephalogram responses in mice and clear electroretinogram responses in human volunteers; the same pulses without the upconverter produced no measurable response at all. Upconverted infrared light, in other words, can enter the biological visual pathway.

The paper follows a wave of attempts to extend human vision into the infrared. Last year, researchers at the University of Science and Technology of China demonstrated upconversion contact lenses that let volunteers distinguish near-infrared patterns, including in color. The Beijing device differs in two ways: it is a glasses-mounted display rather than an in-eye lens, and its color mechanism is entirely physical, emerging from the barrier rather than from spectral engineering of the phosphors. The authors describe the work as transcending the monochrome paradigm of infrared vision.

Caveats separate the demonstration from a product. Every test ran under controlled conditions with a calibrated blackbody source and high-contrast patterns, not the mixed infrared clutter of a real street at night. The OLED side needs an external power supply, making the eyeglass closer to a small powered display than a passive lens. And mercury telluride is a heavy-metal compound: the paper says nothing about long-term skin contact, and the implantable retinal prosthetic that the authors gesture toward would face biocompatibility questions that isolated neurons in a dish and external EEG readings cannot answer. The demonstration that upconverted light can drive real vision is genuine; the road from that to full-color infrared glasses runs through power, real scenes, and toxicity, in that order.

References

Chengchang Fu, Jintao Zou, Xiaoxue Yang, Qun Hao, Xin Tang, and Ge Mu, Multispectral infrared-to-full-color upconversion expanding human vision. Science Advances 12, eaed0245 (2026). DOI: 10.1126/sciadv.aed0245.

Jacek Krywko, Researchers devise a full-color night vision goggle. Ars Technica, 31 July 2026.

Near-infrared spatiotemporal color vision in humans enabled by upconversion contact lenses. Cell 188(13), 3375-3388 (2025). DOI: 10.1016/j.cell.2025.04.019.

Scroll to Top