AI-designed solvent conditions boost quantum dot display lifespan 40-fold

South Korean researchers have identified optimal design conditions that can dramatically improve the efficiency and lifespan of quantum dot light-emitting diode (QLED) devices.

The National Research Foundation of Korea announced Wednesday that a joint research team led by Professor Kwak Jeong-hun of Seoul National University and Professor Lim Jae-hun of Sungkyunkwan University has developed an AI-powered platform that reverse-engineers the ideal solvent properties for arranging quantum dots uniformly and densely during the QLED fabrication process.

QLED devices use quantum dots — nanometer-scale semiconductor particles — as their light-emitting layer and are considered a promising technology for next-generation displays.

The solution-based coating process, which applies liquid-state quantum dots onto a substrate to form a thin film, also lends itself to low-cost, large-area production.

To achieve high-performance QLED, the quantum dot particles must be arranged uniformly and densely within the thin film, much like bricks in a wall.

The challenge is that brightness and lifespan vary significantly depending on which solvent is used in the solution process to form the thin film.

Because the effect of any given solvent condition on device performance is difficult to predict, researchers have long relied on experience and repeated trial-and-error experiments to find optimal conditions — a process that consumes considerable time and cost.

To untangle this complex relationship, the research team trained an AI model to learn the connections between the physical properties of solvents and the structural characteristics of quantum dot thin films.

The team first fabricated quantum dot thin films using five representative solvents and used atomic force microscopy (AFM) to quantify how uniformly each film's surface had formed.

They then fed solvent property data — including vapor pressure, viscosity, density and dielectric constant — along with thin-film morphology data into a machine learning model, training it to predict in reverse which solvent characteristics would produce the most uniform quantum dot thin film.

No single solvent possessed all the optimal properties the AI identified, but the team combined multiple solvents to replicate the conditions the model proposed.

These composite conditions would have been difficult to discover through repeated experiments alone. When applied to an actual QLED fabrication process, the approach yielded roughly double the efficiency and more than 40 times the operational lifespan compared with a conventional single-solvent process.

"This research demonstrates that AI can be used to design display materials and processes on a data-driven basis," Kwak said. "We expect the platform to be applicable not only to QLED but also to the development of various next-generation electronic devices, including OLED and perovskite solar cells."

He added that the team's goal is to advance the platform into an AI system that recommends optimal processes by considering not only thin-film uniformity but also device efficiency and lifespan.

The research was supported by the Future Display Strategic Research Laboratory Support Project and the Nano and Materials Technology Development Project, both run by the Ministry of Science and ICT and the National Research Foundation of Korea. The findings were published online Wednesday in Reports on Progress in Physics, an international journal of the Institute of Physics in the United Kingdom.