Optimized Multimetal Sensitized Phosphor for Enhanced Red Up-Conversion Luminescence by Machine Learning

In this research, machine learning including the genetic algorithm (GA) and support vector machine (SVM) algorithm is used to solve the "low up-conversion luminescence (UCL) intensity" problem in order to find the optimal phosphor with enhanced red UCL emission using multielement K/Li/Mn metal modulation. Compared with the first generation of phosphors, the best phosphors' fluorescence intensity occurs in the third generation optimized by the GA, with a stronger brightness (4.91-fold), a higher relative quantum yield (6.40-fold), and an enhanced tissue penetration depth (by 5 mm). The single and multiple dopants effect on the upconversion intensity of K+Li+Mn sensitizers is also studied: the intensity increases first and then decreases with the increase of Yb/Er/K+Li+Mn content, and the optimized K+Li+Mn concentration is 6.03%. In order to confirm the stability of the brightness optimization by the GA, a batch of phosphors was synthesized with the same element proportion, and the similarity of fluorescence intensity of two batches of phosphors was evaluated by the SVM algorithm with the classification accuracy index. Finally, the optimized phosphor was used for bioimaging and phosphor-LED.