A sub-micron NIR imaging device from Hasselt University and imec converts near-infrared patterns into green light at a threshold of just 9 µW/cm², more than 4,000× lower than prior self-powered upconverter films with comparable efficiency. The monolithic stack, thinner than 1 µm, pairs five organic photodetectors with a green-emitting OLED and produces visible images a conventional camera can record.

How does it run without a supply?

The architecture borrows the series-stacking rule from tandem solar cells. Five NIR subcells sit atop a green-emitting OLED; under illumination, each cell adds a fraction of the photovoltage, and the stack sums these voltages to drive the emitter. No external supply is needed, and the readout chain collapses to a standard camera.

What separates this from earlier self-powered upconverters?

Previous thin-film upconverters relied on multiphoton processes inside a single active layer. Those devices needed laser-class intensities, which barred them from routine imaging. The Hasselt team decouples detection from emission, optimizes each subcell independently, and couples them electrically. The result is a 38-layer organic stack with an external upconversion efficiency of roughly 2%, holding a linear response across three orders of magnitude in input power.

Bernhard Siegmund, research project leader at Hasselt/imec and a Senior Postdoctoral Fellow at the Research Foundation–Flanders, led the work:

"Earlier approaches required intense illumination, often from a laser, which limited their use for practical imaging. But we separate NIR detection and visible emission into optimized semiconductor devices and couple them electrically—rather than relying on a nanoscale upconversion process within the active material."

How does the stack preserve spatial fidelity?

Organic semiconductors have low lateral charge mobility. Photocarriers stay confined as they cross toward the OLED, so each NIR pixel maps cleanly to a green emission pixel. The team demonstrated sharper features than a commercial laser detection card up to 880 nm and expects the advantage to extend deeper into the NIR.

Siegmund pegs the architecture's cost target two orders of magnitude below indium gallium arsenide (InGaAs) focal-plane arrays: "Based on the cost of current components, we estimate our device could ultimately make NIR imaging two orders of magnitude less expensive than specialized InGaAs-based camera systems."

What did the optimization actually require?

Five subcells in series must each generate voltage, but the weakest photocurrent caps the entire stack. Recombination layers between subcells had to minimize optical and voltage losses at once. The OLED needed a low drive voltage and high external quantum efficiency, and the positions of absorbers and the emitter inside the optical microcavity had to balance detector currents with light outcoupling.

Siegmund called the build "a multi-year effort with both scientific and technical challenges." It combined specialized material synthesis with multilayer device deposition and characterization, plus millions of optical simulations that shaped the final 38-layer stack. "We had to optimize the entire stack as a coupled system because the requirements of its components are interdependent," he added.

Where could the device land first?

Three application directions emerged from the tests. In optical labs, the green output reads cleanly under an ordinary camera—useful for NIR alignment work and laser visualization. In agriculture and environmental monitoring, future systems could assess crop or grape ripeness, water stress, forest health, or pipeline leaks. The device also resolved subsurface structures in biological samples, opening a path to label-free biomedical imaging.

Siegmund sees a longer arc under the skin: "Farther out on the horizon, upconverters may become part of optical interfaces for smart implants beneath the skin, for example, to monitor healing after an organ transplant."

What does the team still need to demonstrate?

The next configuration is a transmission-mode device with a transparent top electrode—the geometry a standard camera stack demands. "An important next step now is to integrate the upconverter directly with a conventional camera," Siegmund said. Application-specific prototypes and new materials targeting longer NIR wavelengths round out the roadmap.

The work appears in Communications Materials (2026), DOI 10.1038/s43246-026-01286-8. The compliance question is whether the architecture scales to megapixel formats and sensitivity beyond 1 µm at the cost curve Siegmund sketches—and whether imaging integrators will trade an InGaAs focal plane for an organic stack.