In Situ Raman Spectra and Machine Learning Assistant Thermal Annealing Optimization for Effective Phototransistors

The relationship between the structure and function of condensed matter is complex and changeable, which is especially suitable for combination with machine learning to quickly obtain optimized experimental conditions. However, little research has been done on the effect of temperature on condensed matter and how it affects device performance because the difference between the in situ physical property parameters (which are lowered by the surface tension and mixing entropy) and the basic parameters of the bulk makes accurate AI predictions difficult. In this work, P3HT/ITIC was chosen as the donor/acceptor material for the active layer of organic phototransistors (OPTs). The thermal annealing process has been detected by DSC, UV, and Raman, where Raman can catch the lowest critical phase transition temperatures and give the best raw data for exact machine learning. An accurate and reliable model was developed to predict and screen the optimal annealing temperature at 110 °C for OPTs to reach maximum D_shot* values of 3.51 × 10¹² Jones with low power consumption of 54 pJ. This study provides a new idea for the in-depth exploration of the mechanism of the effect of temperature on condensed matter to achieve the precise regulation and optimization of the performance of organic optoelectronic devices.