Nagoya and Kyushu researchers built a single pipeline from computer-generated molecules to working OLED devices — and produced two boron-free blue emitters, one reaching 35.2 percent external quantum efficiency.

The OLED screen in a modern phone or television is a small triumph of chemistry. Each pixel is built from organic molecules that glow when electricity passes through them, delivering deep blacks, vivid contrast, and thin, flexible designs. Once a luxury, the technology has steadily become more affordable. But one color has always lagged behind the others: blue.

Researchers at the Institute of Transformative Bio-Molecules (WPI-ITbM) at Nagoya University and the Institute for Advanced Study at Kyushu University now report a new way to hunt for better blue materials, combining quantum chemistry with machine learning to screen thousands of virtual molecules before synthesizing the most promising ones. Their results, published July 21 in Angewandte Chemie International Edition, include two new blue emitters with exceptional color purity and efficiency.^1,2

The blue problem

Many of the most efficient red and green OLED pixels rely on heavy metals such as iridium. These phosphorescent emitters can convert nearly 100 percent of the electrical energy that excites them into light internally. Conventional fluorescent blue pixels, by contrast, are capped at 25 percent.²

The reason lies in quantum mechanics. When an electric field is applied across an OLED, it creates excited states in the molecules, but only about a quarter of them are the kind that can directly emit light. The rest are dark. Iridium-based emitters get around this, but blue light requires much higher excited-state energy than red or green, and that extra energy tends to accelerate molecular degradation in metal-based emitters, shortening a pixel’s lifetime.² So blue pixels have generally relied on fully organic molecules — and paid the efficiency penalty.

One promising solution is a class of molecules that exhibit thermally activated delayed fluorescence, or TADF. These molecules use ambient heat to convert non-emissive excited states into ones that can emit light, potentially recovering much of the lost efficiency. But to be useful in commercial displays, TADF emitters need to be readily synthesized and must produce light that is both bright and narrowly focused in color.²

Leaving boron behind

Many of the leading TADF emitters are built around boron-containing frameworks, which can be demanding to make. The Nagoya and Kyushu team deliberately went the other way. They focused on boron-free frameworks built from 13 rings, systematically enumerated from combinations of five- and six-membered rings under defined chemical rules. Every molecule was restricted to just carbon, hydrogen, and nitrogen atoms. The result was a virtual library of more than 19,000 candidate molecules.²

Screening that many molecules with high-accuracy quantum chemistry alone would have been prohibitively expensive in computing time. Machine learning can evaluate huge numbers of candidates quickly, but it needs reliable training data. So the researchers combined the two approaches: they randomly selected 1,000 molecules from the library, calculated their TADF-relevant energy parameters using quantum chemistry, and used those results to train a machine learning model.²

The trained model was then applied to more than 17,000 molecules whose three-dimensional structures could be prepared. From that pool, the team picked 50 promising candidates for more rigorous quantum chemical calculations, and then, weighing both predicted performance and how feasible they would be to make in the lab, narrowed the list to two molecules for synthesis.²

From screen to screen

Both molecules delivered. They emitted vivid blue light in narrow bands, a sign of high color purity, and their photoluminescence quantum yields in thin films ranged from 93 to 99 percent, meaning nearly every absorbed photon was re-emitted as light.²

The researchers then built working OLED devices from the two compounds. A device based on the molecule designated Cz-PAH-1 came close to the blue primary color defined by Rec. 2020, the demanding color standard for next-generation ultra-high-definition displays. A device using Cz-PAH-2 achieved an exceptionally high maximum external quantum efficiency of 35.2 percent.²

A pipeline, not just a molecule

The researchers describe the workflow itself as a key innovation. The study links large-scale virtual molecule generation, quantum chemical calculation, machine learning-based screening, synthesis, photophysical characterization, and device testing into one continuous discovery pipeline — and shows that candidates chosen by an algorithm can survive the trip all the way to a working device.²

Because the method is not specific to blue emitters, the team says it could be extended to other families of organic functional materials with enormous molecular design spaces.² For an industry that has historically discovered new materials through a mix of chemical intuition and painstaking trial and error, that kind of systematic, data-driven search could shorten the road from idea to product.

For consumers, the payoff would be screens that are more efficient, more color-accurate, and potentially longer-lasting — with the stubborn blue pixel finally pulling its weight.

---

Notes

1. Machine-Learning-Guided Discovery of Boron-Free Narrowband Blue Thermally Activated Delayed Fluorescence Emitters. (2026). Angewandte Chemie International Edition. https://doi.org/10.1002/ange.2731055
2. Nagoya University. (2026, July 22). AI and quantum chemistry identify efficient blue OLED materials [Press release]. EurekAlert! https://www.eurekalert.org/news-releases/1137086