Fusion Deep Learning for Predicting Conductivity in Electron-Doped Organic Polymers.

Area of Science:

• Materials Science
• Organic Electronics
• Computational Chemistry

Background:

• Efficient and air-stable n-type organic semiconductors are vital for high-performance organic electronic devices.
• Molecular n-doping is key to achieving desired semiconductor properties.
• Machine learning (ML) accelerates the discovery and design of novel materials by identifying structure-property relationships.

Purpose of the Study:

• To develop and validate a data-driven approach for predicting the conductivity of n-type organic semiconductors.
• To construct a comprehensive database of n-type conductive polymers with experimental and computational data.
• To design and implement a fusion deep learning model for accurate property prediction.

Main Methods:

• A curated database of 84 n-type conductive polymers was created, including experimental n-doping data with 4-(1,3-dimethyl-2,3-dihydro-1H-benzoimidazol-2-yl)phenyl dimethylamine (N-DMBI-H) and density functional theory (DFT) calculations.
• A fusion deep learning model, integrating convolutional neural networks (CNNs) and fully connected artificial neural networks (ANNs), was developed to analyze polymer structures and properties.
• The model was trained and validated using leave-one-out cross-validation and applied to predict the conductivity of polymers with oligoethylene glycol (OEG) side chains.

Main Results:

• The fusion deep learning model successfully identified key physical factors influencing conductivity in n-type polymers doped with N-DMBI-H.
• A double-blind experiment validated the model's predictive capability, with conductivity predictions for N2200-type and BDPPV-type polymers agreeing with experimental values within the same order of magnitude.
• The model demonstrated reliability in predicting the conductivity of doped n-type organic semiconductors.

Conclusions:

• The developed fusion deep learning model is a reliable tool for predicting the conductivity of n-type organic semiconductors.
• This data-driven approach provides a strong foundation for the rational design of high-conductivity n-type polymers.
• The findings facilitate the advancement of durable and high-performance organic electronic devices through accelerated materials discovery.