Synthesis guidance of PbS colloidal quantum dots with neural network model for short wave infrared photodetector

PbS colloidal quantum dots (CQDs) have important applications in short-wave infrared (SWIR) detection due to its wide tunable bandgap, low thermoelectric noise, and solution processing capability. Due to the exciton peak of QDs determines the response band of the detector, while QDs with good monodispersed often exhibit better optical performance in photodetectors. The detection performance of PbS CQD-based SWIR photodetectors is closely related to the synthetic properties of QDs in the active layer. In addition, the emergence of machine learning in recent years has accelerated the exploration of QDs synthesis processes. Here, a framework is developed by neural network model which can learn from existing experimental data, through proposed experimental parameters for try, and ultimately point to regions of synthetic parameter space, thereby rapidly and accurately predicting the exciton peak and peak/valley ratio of synthesized CQDs. In terms of model performance, the NN model achieved a correlation coefficient of 0.93 for exciton peak prediction, which is very close to 1. For peak/valley ratio prediction, the correlation coefficient reached 0.75. In prediction of the latest synthesized CQD, the prediction error of exciton peak is only 3.89 %, and the prediction error of peak/valley ratio is 7.24 %. Furthermore, this batch of well synthesized monodisperse CQDs with a peak/valley ratio of 3.105 were used to prepare SWIR photoconductive devices, which demonstrates an excellent device performance, with the responsivity achieving 2.53 A/W, the detectivity reaching up to 2.08 × 10¹² Jones and the noise current of only 7.81 × 10⁻¹³ A/Hz^1/2. This work provides an effective method for preparing PbS CQD of various waveband with uniform particle size, which is expected to reduce costs for high-performance SWIR photodetectors.