Researchers at Korea's Seoul National University and Sungkyunkwan University have applied AI to reverse-engineer QD-EL device production conditions, and have come up with a solvent that can arrange QDs in an optimal way.
The researchers say that by applying this solvent in a commercial QD-EL device, it can increase the efficiency twofold, and the lifetime by more then 40X.
The research team enabled the AI to learn the relationship between the physical properties of the solvent and the structure of the QD layer. First, thin films of QD were fabricated using five representative solvents, and the uniformity of the surface formation was quantified using an AFM. Then, by applying a range of physical properties (solvent vapor pressure, viscosity, density, etc. ), a machine learning model was trained to inversely predict the solvent characteristics capable of forming the most uniform quantum dot thin films.