Predicting The Energy Level Structure Of The Organic Luminescent Materials And Their Luminescence Properties In The Perovskite Host

With the development of the application of organic light-emitting devices(OLEDs),the requirement of its performance is higher and higher.In order to improve the performance of the device,the choice of the luminescent layer and the improvement of the luminescent material itself are very important.But at present,the prediction method of material energy level is still single and the main material does not have the ability of bipolar transmission.In this thesis,we use machine learning algorithm to predict the energy levels of materials and prepare OLEDs based on perovskite with bipolar transport properties.The energy level structure is one of the essential characteristics of organic luminescent materials.Therefore,we firstly predict the energy level structures of the highly efficient thermally activated delayed fluorescence(TADF)materials using machine learning.We collected more than 300 TADF materials and established a dataset containing more than 200 TADF molecules after data minimizing and feature extraction.We investigated the effect of different feature sets on the learning performance and selected the optimized feature sets.Moreover,we reduced the feature size by correlation analysis.Based on these,the random forest(RF)algorithm was used to map the correlation between the molecule features and the energy levels(HOMO/LUMO)of the TADF materials.The features that induced the prediction error of the RF model were analyzed.Based on the hyperparameter optimization of the algorithm,the accurate prediction of HOMO/LUMO energy levels of the TADF materials was realized.The root mean square error and the correlation coefficient of the RF model on predicting the test set are 0.237 and 0.80 for HOMO and 0.342 and 0.61 for LUMO,respectively.Based on correlation analysis and feature importance analysis,we screened the most critical features governing the energy level structure of the TADF materials and proposed directions for designing TADF materials with desired colors.In this thesis,we adopted perovskite as the host material and prepared organic/perovskite hybrid light-emitting films.After comparing the precr doping with the anti-solvent doping,the complete subject luminescence at low doping concentration was achieved by using the anti-solvent doping.The scanning electron microscopy images show that the organic materials are homogeneously doped under the one-step anti-solvent method.In addition,we also tried to use different dimensional perovskite materials to prepare hybrid light-emitting thin films.Among them,quasi-two-dimensional perovskites show better performance owing to their high tolerance to deformation and better stability.The luminescence properties of organic/perovskite hybrid layers with different energy levels were investigated,and the phenomenon of redshift of emission wavelength was found.These studies greatly facilitate the researchers to TADF material energy level structure prediction and control,reducing unnecessary attempts to cause waste.The feasibility of perovskites as the host materials and provide the optimization strategies,which also highlight the development potential of organic/perovskite hybrid lightemitting devices.