AI unlocks QLED recipe that doubles efficiency and boosts lifetime 40-fold

AI unlocks QLED recipe that doubles efficiency and boosts lifetime 40-fold Gaby Clark Scientific Editor Andrew Zinin Chief Editor A technology has been developed that allows artificial intelligence to inversely determine the process conditions for quantum-dot light-emitting diode (QLED) devices—conditions that previously required extensive trial and error to identify. When applied to actual devices, the technology roughly doubled efficiency and extended operational lifetime more than 40-fold, raising expectations that it could accelerate the development of next-generation displays. Seoul National University's College of Engineering announced that a joint research team led by Professor Jeonghun Kwak of the Department of Electrical and Computer Engineering and Professor Jaehoon Lim of Sungkyunkwan University's Department of Energy Science has developed an AI-based platform that inversely designs the optimal solvent properties for arranging quantum dots uniformly and densely during the fabrication of QLEDs. The research was supported by the Ministry of Science and ICT and the National Research Foundation of Korea through the Future Display Leading Technology Program and the Nano & Material Technology Development Program. The findings were published online July 15 in Reports on Progress in Physics, an internationally renowned physics journal published by the United Kingdom's Institute of Physics (IOP). Why solvent choice is critical Quantum-dot LEDs (QLEDs), which use nanometer-scale semiconductor particles called quantum dots as their light-emitting layer, are regarded as a promising technology for next-generation displays. This is because they can be fabricated using a solution process—coating a substrate with quantum dots in liquid form to create a thin film—making them advantageous for low-cost, large-area production. To achieve high-performance QLEDs, quantum-dot particles must be arranged uniformly and densely within the thin film, much like bricks. The challenge is that the choice of solvent used to form the film in this solution process significantly affects brightness and life span. Because it has been difficult to predict how specific solvent conditions influence performance, researchers have largely relied on experience and repeated experimentation to find optimal conditions—a process that consumes considerable time and cost. Teaching AI the film structure To untangle this complex relationship, the research team trained an AI model to learn the connection between the physical properties of solvents and the resulting structure of quantum-dot thin films. They first fabricated quantum-dot films using five representative solvents and quantified the uniformity of the surface using atomic force microscopy (AFM). The team then trained a machine learning model on solvent property data—vapor pressure, viscosity, density, dielectric constant and more—alongside the corresponding film morphology data, enabling it to inversely predict the solvent characteristics that would produce the most uniform quantum-dot film. Atomic force microscopy (AFM) uses a fine probe to scan a sample's surface and measure its height variations and roughness. Blending solvents to match predictions While no single solvent possessed all of the optimal properties suggested by the AI, the research team combined multiple solvents to realize the conditions the AI had proposed. This complex combination—one that would have been difficult to discover through repeated experimentation alone—was applied to an actual QLED fabrication process, resulting in roughly double the efficiency and more than a 40-fold increase in operating lifetime compared with devices made with a single conventional solvent. "This research demonstrates that AI can be used to design display materials and processes on a data-driven basis," said Kwak. "We expect it can also be applied to the development of various next-generation electronic devices, including OLEDs and solar cells." Publication details Beomsoo Chun et al, Machine-learning-enabled solvent engineering for uniform quantum dot packing in efficient and stable quantum-dot light-emitting diodes, Reports on Progress in Physics (2026). DOI: 10.1088/1361-6633/ae8470 Journal information: Reports on Progress in Physics Provided by Seoul National University