Quantum dot LEDs (QLEDs) are of interest because they emit bright light with highly pure colours, making them attractive for display and lighting applications. They also have the potential to be manufactured at lower cost using solution-processing techniques. However, producing high-performance QLEDs remains challenging because the quantum dots must be uniformly distributed within a densely packed film. Poor film uniformity can lead to lower efficiency and faster device degradation.

In this study, machine learning was used to identify solvent conditions that produce the most uniform quantum dot films. Several solvents were characterised using five key solvent parameters that influence how quantum dots arrange themselves as the solvent evaporates. Film uniformity was assessed using atomic force microscopy, which provides detailed information about surface morphology and roughness.

Three machine learning models were trained to predict film uniformity from the solvent properties. Support Vector Regression was found to provide the most accurate predictions and was subsequently used to identify an optimal mixed-solvent formulation. Grazing-Incidence Small-Angle X-ray Scattering confirmed that the resulting films exhibited more homogeneous quantum dot packing.

The optimised solvent formulation produced QLEDs with higher efficiency and longer operational lifetimes than devices fabricated using single solvents. Overall, the study demonstrates that film homogeneity is a critical factor in QLED performance and highlights the potential of machine learning as a powerful tool for optimising solution-processed optoelectronic devices.

Do you want to learn more about this topic?

Materials, photophysics and device engineering of perovskite light-emitting diodes by Ziming Chen et al. (2021)