Predicting Photoluminescence Quantum Yields of Quantum Dots from Time-Resolved Photoluminescence with Artificial Neural Networks

Photoluminescence (PL) dynamics of quantum dots (QDs) are indicative of the quality of the emission channel, band-edge emission from a delocalized excitonic state, or emission from a localized state. PL dynamics determine the PL quantum yield (QY), although the quantitative relationship between the two is not always easy to establish, especially for heterogeneous ensembles of QDs. In this work, we develop a generalized artificial neural network (ANN) model that makes use of a combination of physical characteristics and measured transient PL to predict the QY of QD samples of various compositions. We train the model using literature and our own data for QDs of a number of compositions (CdSe, CdTe, PbS). We deploy the model on new validation data points as well as compositions not used in training (Cd3P2). We observe a solid performance of the model on the validation set, with a root-mean-square error of 13%. Finally, we use the trained model to make predictions of radiative lifetimes of QDs of several compositions.